{ "schemaVersion": 1, "evaluation": { "rubricVersion": "ramen-v3", "verifierSha256": "19066b98b56ae8dd5ad68feba341e8adaa04a15ecfc7b52a8d2a5285b458762b", "judge": { "models": [ "openrouter/openai/gpt-6.1-sol", "openrouter/anthropic/claude-sonnet-5.5", "openrouter/z-ai/glm-5.3-flash" ], "samples": 3, "efforts": { "openrouter/openai/gpt-6.1-sol": "low", "openrouter/anthropic/claude-sonnet-5.5": "low", "openrouter/z-ai/glm-5.3-flash": "default" } }, "criteria": [ "noodles", "toppings", "bowl", "broth", "steam", "composition" ], "weights": { "noodles": 0.2, "toppings": 0.25, "bowl": 0.2, "broth": 0.1, "steam": 0.15, "composition": 0.1 }, "aggregation": "compliance gates \u00d7 weighted mean of visual criteria; equal judge weights" }, "results": [ { "artifactSha256": "37dfcef19740fc51451fb27d36a7e2aac0b1051ccb4088d7d29118b1de753d80", "gradedAt": "2026-09-30T22:29:07.670862+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5944444444444444, "toppings": 0.6877777777777777, "bowl": 0.8444444444444444, "broth": 0.6822222222222223, "steam": 0.768888888888889, "composition": 0.8466666666666667 }, "reward": 0.7279444444444445, "runId": "anthropic/opus-5.5/max", "rewardStdDev": 0.060529224652534894, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.7233333333333333, "bowl": 0.93, "broth": 0.6533333333333333, "steam": 0.78, "composition": 0.88 }, "reward": 0.7371666666666667, "sampleStdDev": 0.002020725942163724, "judgments": [ { "noodles": { "score": 0.5, "reason": "Readable yellow strands have consistent thickness and some overlap, but dominate as a flat bundle of repeated tight zigzag waves rather than supple, varied noodles." }, "toppings": { "score": 0.72, "reason": "Veined greens, scored mushroom caps, textured tofu, corn and scallion rings provide clear variety. Nori is overly rectangular and rigid, and several toppings have simplified, glossy surfaces." }, "bowl": { "score": 0.93, "reason": "Coherent elliptical rim, visible inner wall, rounded body and grounded base establish convincing depth. Speckled ceramic, irregular glaze and restrained rim highlights are polished; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm orange liquid has scattered oil-like flecks and reflections, but the surface is rather flat, with strong topping shadows and limited immersion cues making ingredients appear perched above it." }, "steam": { "score": 0.78, "reason": "Translucent wisps originate around the soup and change shape across normal frames, with broader sideways curling in pointer frames. The tall, similarly oriented streaks remain somewhat repetitive and haze the rear toppings." }, "composition": { "score": 0.88, "reason": "The single dish remains the clear focus on desktop and mobile, with warm wood, dark negative space and coherent directional lighting. No distracting UI is visible; the bright left-side reflection is slightly overpowering." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, nori and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.5, "reason": "Individual yellow strands are readable with consistent thickness, but the mound is dominated by repeated parallel zigzag waves rather than supple, varied tangles." }, "toppings": { "score": 0.73, "reason": "Leafy greens have pale stems and veins, mushrooms have scored caps, and tofu and corn are recognizable. Nori is overly rectangular and rigid, while several ingredients have simplified glossy surfaces." }, "bowl": { "score": 0.93, "reason": "A coherent elliptical rim, deep curved walls, grounded base, speckled ceramic finish and irregular glaze boundary convincingly establish the bowl. No material construction defect is observed." }, "broth": { "score": 0.66, "reason": "Warm orange liquid has small oil-like flecks and highlights, but the large blown-out reflection and strong hard shadows beneath toppings make immersion feel shallow and somewhat staged." }, "steam": { "score": 0.78, "reason": "Translucent wisps visibly change shape across normal desktop frames and rise from the bowl, with broader bending in pointer captures. Some plumes remain overly tall and similarly aligned, creating a smoky veil over the rear toppings." }, "composition": { "score": 0.88, "reason": "The single dish remains clearly framed on desktop and mobile, with warm directional lighting, a restrained dark background and no distracting interface. The intense left-side highlight slightly unbalances food visibility." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Distinct yellow strands have consistent thickness and some overlap, but predominantly form parallel, repetitive zigzag waves in a flat raft rather than supple tangled volume." }, "toppings": { "score": 0.72, "reason": "Leaf veins, white vegetable stems, scored mushroom caps, scallion rings and textured tofu provide readable variety. Nori is overly rectangular and rigid, and several toppings have simplified glossy shapes." }, "bowl": { "score": 0.93, "reason": "Coherent elliptical rim, visible inner walls, rounded body and supported foot establish convincing depth. Speckled ceramic, irregular glaze boundary and fine rim highlights are polished; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm orange liquid has scattered oil-like droplets and surface highlights, but the broad surface is quite uniform and strong shadows make several ingredients appear perched above it rather than immersed." }, "steam": { "score": 0.78, "reason": "Translucent wisps emerge around the soup and visibly change shape across normal frames, with larger sideways curls in pointer frames. The tall, somewhat parallel smoky bands are heavier and more repetitive than delicate steam." }, "composition": { "score": 0.88, "reason": "A single readable dish anchors a restrained dark scene with warm wood, coherent directional lighting and ample negative space; mobile framing retains the bowl. The intense left-side highlight washes out some food detail." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, scallions, seaweed and noodles plausibly constitute vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6066666666666667, "bowl": 0.7533333333333333, "broth": 0.66, "steam": 0.6933333333333334, "composition": 0.7766666666666667 }, "reward": 0.6633333333333333, "sampleStdDev": 0.020132891827388685, "judgments": [ { "noodles": { "score": 0.6, "reason": "The noodles form a wavy, curly mass with believable thickness and overlap, and they sit in the broth. The clump is small and a bit dense, and the strands read as repetitive curls rather than long supple lengths." }, "toppings": { "score": 0.62, "reason": "The toppings are varied and recognizable: bok choy leaves, shiitake with scored caps, tofu cubes, enoki, corn and scallion rings. The nori sheets are flat, rigid rectangles standing upright, and the tofu cubes are fairly blocky." }, "bowl": { "score": 0.78, "reason": "The bowl has a coherent rim, speckled ceramic glaze, a drip-glaze base, a foot and a good contact shadow. The perspective is consistent. Small flaws are the slightly jagged glaze edge and a rim that looks a little thin." }, "broth": { "score": 0.68, "reason": "The orange broth has a specular highlight, oil droplets and chili flecks, and the toppings sit in it with shadows. It looks somewhat flat and opaque, and the glare on the left is strong." }, "steam": { "score": 0.7, "reason": "The steam is translucent and wispy, and its shape differs from frame to frame, so it reads as moving. It rises and spreads in soft plumes without hiding the food. It is slightly faint and diffuse in places." }, "composition": { "score": 0.78, "reason": "The bowl is a clear focal point in a dark, moody scene. A warm spotlight falls on the wood table and the colors are cohesive. There is no clutter or UI. The dark background is a bit empty." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, bok choy, corn, enoki, nori and scallions, with no visible meat, fish or egg." } }, { "noodles": { "score": 0.55, "reason": "The noodles form a small, dense wavy mass with a believable yellow color and some overlap. The strands are fine and somewhat indistinct, and the clump is small and flat for the bowl size. It reads as noodles, but it lacks volume and supple individual strands." }, "toppings": { "score": 0.6, "reason": "The toppings are varied and readable: bok choy leaves, shiitake with scoring marks, tofu cubes, corn, enoki and scallion rings. The nori sheets are flat, rigid rectangles standing upright, which looks artificial. The tofu cubes are a bit blocky and plain, though they are shaded well." }, "bowl": { "score": 0.78, "reason": "The bowl has a coherent elliptical rim, a speckled ceramic glaze, a foot ring and a credible contact shadow on the wood. The wall thickness and highlights read well. The glaze drip edge is slightly odd, but no major distortion is visible." }, "broth": { "score": 0.68, "reason": "The orange broth has a glossy specular highlight, floating oil and chili flecks, and toppings that sit in it with some contact shadows. It looks a little flat and uniformly opaque, and the nori seems to cast odd shadows." }, "steam": { "score": 0.68, "reason": "The steam is translucent, wispy and irregular, and it rises and drifts differently across the frames. It is a little broad and hazy, and it partly covers the nori, but it does not hide the food." }, "composition": { "score": 0.8, "reason": "The bowl is a clear focal point, centered under a dramatic warm spotlight on a dark wood table. The colors are cohesive and the atmosphere is dark and restrained. The food is somewhat small in the frame, and the lower half is empty." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, bok choy, corn, enoki, nori and scallions. No animal products are visible." } }, { "noodles": { "score": 0.55, "reason": "The noodles form a small, dense wavy mass with a recognizable ramen look and some curl and overlap. The clump is small, reads as a tangle, and the individual strands are hard to make out. It is not convincingly voluminous or immersed." }, "toppings": { "score": 0.6, "reason": "The toppings are varied: bok choy leaves, shiitake with scoring, tofu cubes, corn, enoki and scallion rings. The shiitake and bok choy read well. The nori sheets are flat, rigid rectangles standing upright, and the tofu cubes are plain blocks." }, "bowl": { "score": 0.7, "reason": "The speckled ceramic bowl has a coherent rim, walls and foot ring, a believable perspective and a contact shadow on the wood. The drip-glaze base band looks slightly jagged and the rim is fairly plain, but there is no major distortion." }, "broth": { "score": 0.62, "reason": "The orange broth has a sheen, specular highlights and floating chili and oil specks. Toppings sit in it with shadows and contact. The surface looks a bit flat and opaque, and the glare on the left is strong." }, "steam": { "score": 0.7, "reason": "The translucent wisps rise and spread, and they change shape between frames. They do not obscure the food. The plumes are somewhat diffuse and broad, and they fade into the backdrop." }, "composition": { "score": 0.75, "reason": "A single focal bowl is centered under a dramatic spotlight on a wood table, in a dark, moody scene with warm colors and no clutter. The lower half is empty and the nori partly collides with the steam, but the food remains the focus." }, "vegan": { "score": 0.92, "reason": "The toppings are tofu, mushrooms, bok choy, corn, enoki, nori and scallions, and the broth looks plant-based. No animal products are visible." } } ], "diagnostics": { "vegan": 0.9266666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.7333333333333333, "bowl": 0.85, "broth": 0.7333333333333334, "steam": 0.8333333333333334, "composition": 0.8833333333333333 }, "reward": 0.7833333333333333, "sampleStdDev": 0.029297326385411604, "judgments": [ { "vegan": { "score": 1, "reason": "All visible ingredients are plant-typical: noodles, nori sheets, shiitake caps, tofu cubes, corn, bok choy, sprouts, scallions and chili in a miso-colored broth; no meat, egg, fish or dairy forms appear." }, "toppings": { "score": 0.75, "reason": "Good variety and readable identity: cross-cut shiitake caps, bok choy with pale stalks, scallion rings, chili flecks, corn cluster, sprouts and standing nori sheets with subtle texture. Slight defects: tofu cubes read as glossy uniform blocks and the corn is a pile of near-identical spheres." }, "steam": { "score": 0.85, "reason": "Across desktop-1/2/3 and pointer frames the wisps clearly change shape, rising, spreading and thinning into the dark, staying translucent and never obscuring the food. Reduced frames freeze it identically, consistent with prefers-reduced-motion." }, "noodles": { "score": 0.7, "reason": "Golden curly strands with plausible thickness and coherent overlaps, garnished with scallion rings. Shortcomings: the mass reads as a fairly flat radiating mat lying on the surface with a repetitive wave rhythm rather than a supple piled nest, and left strands drape stiffly over the greens." }, "composition": { "score": 0.9, "reason": "Strong dark atmospheric staging: warm spotlight pools on the bowl, soft bokeh, clean wooden table, clear focal dish, no text or clutter, and steam adds vertical balance." }, "broth": { "score": 0.8, "reason": "Appetizing orange miso surface with oil droplets, chili flecks and ring scum near the rim; nori casts shadows onto the liquid and ingredients sit with contact rather than floating. Minor flaw: a large soft specular bloom in the center-left slightly washes out the surface." }, "bowl": { "score": 0.85, "reason": "Coherent speckled ceramic bowl with dripped glaze, consistent elliptical rim and wall thickness, visible footed base resting with a believable contact shadow; material highlights are convincing. No incompatible perspective or broken joins observed." } }, { "noodles": { "score": 0.7, "reason": "Center mass of curly yellow strands with plausible waviness and scattered scallion rings reads as ramen, but the coil pattern is repetitive and the mass lies fairly flat on the broth with limited volume; a lone strand trails unnaturally to the left like a rope." }, "toppings": { "score": 0.7, "reason": "Good variety and most items read well: scored shiitake caps, bok choy with pale stems, enoki bundle, chili flakes and sesame. Defects: scallions are literal green torus rings, corn is a pile of glossy yellow beads, tofu cubes are sharp-edged polished blocks, and the two nori sheets are thick rigid slabs standing like cards." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: clean elliptical rim, speckled interior glaze, tapered body with drip-glaze band, and a small footed base sitting on the table with a grounded cast shadow. Perspective is consistent; only minor stiffness in the drip pattern." }, "broth": { "score": 0.6, "reason": "Orange miso-toned surface with sheen, chili specks and light hotspot, and ingredients show contact/occlusion at their bases. However the liquid reads fairly solid and matte like clay, with a flat single highlight rather than believable liquid depth." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise from the soup and the plume configuration clearly evolves between desktop-1/2/3 and the pointer frames, rising and spreading without hiding the food; slightly broad and diffuse high above the bowl." }, "composition": { "score": 0.85, "reason": "Strong dark atmospheric scene: warm spotlight isolates the bowl on a wood table, dish is the clear focal point, balanced framing on both desktop and mobile, no clutter or competing elements. The long cast shadow slightly dominates the lower right." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are nori, tofu cubes, corn, shiitake, enoki, bok choy, scallions, chili and sesame in a plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.75, "reason": "Yellow curly strands form a coherent central mass with believable volume and scallion rings resting on top; the wave pattern is somewhat repetitive and instant-noodle-like, but strands connect and read clearly rather than as wires or scribbles." }, "toppings": { "score": 0.75, "reason": "Good variety: two upright nori sheets with texture, corn cluster, bok choy with white stems, shiitake-style discs with star scoring, scallion rings, chili flecks and sesame. The tofu cubes look slightly blocky/plasticky and the mushrooms are flat stylized discs, but all ingredients are recognizable and appropriately scaled." }, "bowl": { "score": 0.85, "reason": "Well-formed ceramic bowl with coherent speckled rim, consistent elliptical perspective, drip-glaze body, proper footed base and grounded contact shadow on the wood table; no distortion or detached parts, minor stylization in the glaze drips." }, "broth": { "score": 0.8, "reason": "Appetizing orange miso-style surface with restrained sheen, bubbles, chili flecks and oil spots; toppings sit with visible contact shadows and reflection near the nori. The surface is slightly flat in places but immersion reads believably." }, "steam": { "score": 0.85, "reason": "Steam clearly evolves across time-ordered frames: wisps shift shape and drift (vertical in desktop-1, spreading rightward in desktop-2/3, swirling differently in pointer frames). Translucent, irregular, and does not obscure the food; reduced frames show it appropriately static." }, "composition": { "score": 0.9, "reason": "Balanced centered dish in a dark atmospheric scene with a warm spotlight, wood table and soft bokeh; food is the clear focal point with no text, UI or competing elements, and framing works on both desktop and mobile viewports." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are entirely plant-based: noodles, nori, corn, tofu cubes, mushrooms, bok choy, scallions, chili and sesame; no egg, meat, fish or dairy visible." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9755555555555556, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 1466.6000000000058, "frameIntervalP95Ms": 1466.6000000000058, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "2a4735b83fa30c0c3b8bc929e7e9586d4352773d8145a1c10542f8438b20432a", "gradedAt": "2026-09-30T22:27:29.587655+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.65, "toppings": 0.6577777777777778, "bowl": 0.8055555555555556, "broth": 0.7188888888888889, "steam": 0.6900000000000001, "composition": 0.8166666666666667 }, "reward": 0.7126111111111111, "runId": "anthropic/sonnet-5.5/max", "rewardStdDev": 0.0804883244351029, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6333333333333333, "toppings": 0.6466666666666667, "bowl": 0.88, "broth": 0.72, "steam": 0.67, "composition": 0.86 }, "reward": 0.7228333333333334, "sampleStdDev": 0.005530220007678969, "judgments": [ { "noodles": { "score": 0.64, "reason": "Distinct rounded strands have warm shading and coherent overlap, but the tightly parallel, repeated bends form an overly orderly corrugated mound rather than loose, supple noodles." }, "toppings": { "score": 0.65, "reason": "Porous browned tofu, veined greens, scallion rings and scored mushroom caps provide readable variety. Leaves are stiff and repetitive, mushrooms resemble short cylinders, and the upright nori looks like rigid corrugated cards." }, "bowl": { "score": 0.88, "reason": "The elliptical rim, visible inner wall, curved body and small supported base form coherent depth and perspective. Warm rim highlights and side reflections convincingly suggest glazed ceramic; the dark lower wall loses some material definition." }, "broth": { "score": 0.72, "reason": "Amber shading, scattered oil flecks and submerged details suggest appetizing liquid depth. Ingredient contacts are somewhat dry and sharply outlined, with limited meniscus or surface reflection." }, "steam": { "score": 0.66, "reason": "Translucent vapor emerges over the soup and changes distribution across the normal frames without hiding most food. It reads more as a broad soft haze than distinct rising, spreading wisps; continuous flow is not established by the sparse captures." }, "composition": { "score": 0.86, "reason": "Warm food colors stand clearly against the dark setting, with a balanced focal bowl, restrained table detail and no distracting UI or text. The mobile dish remains fully readable, though its low placement leaves considerable empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, greens, mushrooms, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.62, "reason": "Distinct golden strands have rounded thickness, curved bends and layered volume, but their closely parallel, repeated waves look overly arranged rather than naturally tangled." }, "toppings": { "score": 0.64, "reason": "Tofu has browned porous surfaces, greens have readable veins, and scallions and corn add variety. Mushroom caps resemble thick cylinders, leaves are repetitive, and upright nori looks stiff and card-like." }, "bowl": { "score": 0.88, "reason": "The continuous elliptical rim, shaded inner wall, rounded exterior and grounded base form a coherent vessel. Warm glaze, dark decoration and curved highlights convincingly suggest ceramic; no material construction defect is observed." }, "broth": { "score": 0.72, "reason": "Warm amber liquid has subtle speckles, oil droplets and shaded depth around the inner wall. Ingredient contact is readable, though the surface is rather uniform and several toppings seem perched above it." }, "steam": { "score": 0.66, "reason": "Translucent vapor emerges above the soup and changes its distribution across normal frames without obscuring the food. It reads more as a broad diffuse cloud than distinct rising, spreading wisps; continuous flow cannot be established from these captures." }, "composition": { "score": 0.86, "reason": "Warm food colors stand clearly against the dark wooden setting, with a readable focal bowl, coordinated lighting and no text or UI clutter. Mobile retains the whole dish, though its large upper empty area weakens the balance slightly." }, "vegan": { "score": 1, "reason": "Visible tofu, greens, mushrooms, seaweed, corn, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.64, "reason": "Distinct rounded strands have warm shading and coherent overlap, but the bundles are overly parallel and uniformly bent, with many blunt exposed ends rather than supple tangled continuity." }, "toppings": { "score": 0.65, "reason": "Porous browned tofu, veined greens, scallion rings and scored mushroom caps provide readable variety. Leaves are stiff and repetitive, mushrooms resemble thick cylinders, and the upright nori looks like rigid corrugated cards." }, "bowl": { "score": 0.88, "reason": "A coherent elliptical rim, visible inner wall, rounded body and grounded base establish convincing depth. Warm rim highlights and a curved side reflection support the lacquer finish; the dark lower body loses some material definition." }, "broth": { "score": 0.72, "reason": "Amber shading, scattered oil-like flecks and partial ingredient occlusion suggest appetizing soup. The surface is somewhat matte and uniform, with limited meniscus or wet contact around the larger toppings." }, "steam": { "score": 0.69, "reason": "Normal frames show changing translucent vapor patches above the soup without hiding the food. The vapor is broad and cloud-like rather than clearly articulated rising, spreading wisps, so continuous fluid motion is not established." }, "composition": { "score": 0.86, "reason": "The single dish remains the clear focus against a restrained dark wooden setting, with warm highlights and green-gold food contrast and no distracting UI or text. Mobile framing preserves the whole bowl but leaves considerable empty space above it." }, "vegan": { "score": 1, "reason": "Visible tofu, greens, mushrooms, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.5933333333333333, "bowl": 0.6866666666666666, "broth": 0.62, "steam": 0.5833333333333334, "composition": 0.7233333333333333 }, "reward": 0.6275, "sampleStdDev": 0.004769696007084703, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles form a mound of layered curved strands with some depth and plausible overlap, but they read as repeated parallel arcs and the mass is small and somewhat flat and stylized." }, "toppings": { "score": 0.6, "reason": "Variety is good: bok choy leaves, tofu blocks, shiitake caps with a scored pattern, corn and scallion rings. The nori is two stiff rectangular cards standing upright, and the tofu and mushrooms look like blocky, slightly plastic pieces." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a thick lacquer wall with a drip glaze, a foot and a highlight, and it sits on the wood table. The lower body is very dark and the side reflection looks slightly odd, but no major distortion." }, "broth": { "score": 0.62, "reason": "The broth is a warm brown with some specks and oil sheen, and the toppings sit in it with soft contact shadows. The surface is fairly flat and murky, with little ripple depth." }, "steam": { "score": 0.55, "reason": "Translucent wisps rise from the bowl and shift between frames, with a soft fade. They are hazy and somewhat diffuse, and they overlap the nori, but they do not hide the food." }, "composition": { "score": 0.72, "reason": "The bowl is a centered focal point in a dark, moody scene with a wood table and restrained bokeh. The color palette is cohesive, and there is no text or clutter." }, "vegan": { "score": 0.93, "reason": "Only plant ingredients are visible: tofu, mushrooms, bok choy, corn, scallions, nori and noodles. There are no animal products." } }, { "noodles": { "score": 0.6, "reason": "Noodles form a readable mound with many parallel curved strands and some overlap, but they look like repeated flat swirl lines and the mass is small and somewhat stiff." }, "toppings": { "score": 0.58, "reason": "Toppings are varied and recognizable: bok choy leaves, tofu blocks, shiitake caps with a star cut, corn and scallion rings. The nori is a tall rigid card-like sheet with a jagged edge, and the mushrooms read as chocolate-like discs." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, lacquer drip glaze, a foot and a reflective highlight, and it sits convincingly on the wood table. The lower body is very dark with little detail, and the rim perspective is slightly wide." }, "broth": { "score": 0.62, "reason": "The brown broth has a soft sheen, small floating bits and some ingredient contact, so it is appetizing. It is fairly flat and opaque, and the toppings sit on it with limited immersion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise from the bowl and drift and vary between frames without hiding the food. They are somewhat diffuse and fog-like, with limited distinct wisp detail." }, "composition": { "score": 0.75, "reason": "The bowl is centered and is a clear focal point in a moody dark scene with warm wood and soft bokeh. The palette is cohesive, and the tall nori slightly distracts." }, "vegan": { "score": 0.95, "reason": "Visible ingredients are tofu, mushrooms, bok choy, corn, scallions, nori and noodles. No animal products are apparent." } }, { "noodles": { "score": 0.6, "reason": "Noodle mound has many fine parallel strands with believable volume and color, but the strands read as repetitive flat swirl lines and are slightly rigid, with little clear overlap." }, "toppings": { "score": 0.6, "reason": "Bok choy leaves, tofu blocks, shiitake caps, corn and scallion rings are all recognizable and varied. The nori sheets are tall and rigid like cards, and the mushrooms look like chocolate discs, which limits the score." }, "bowl": { "score": 0.68, "reason": "The lacquer bowl has a coherent elliptical rim, a curved body, a foot and a drip glaze pattern. Highlights are good and it sits on the table convincingly. The dark lower body is somewhat murky." }, "broth": { "score": 0.62, "reason": "The broth is an appetizing warm brown with speckles and some depth, and the toppings sit in it with contact shadows. The surface is fairly flat and matte, with little sheen." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise from the bowl and drift between frames without hiding the food. The plume is somewhat diffuse and blobby, and its motion is modest." }, "composition": { "score": 0.7, "reason": "The bowl is a centered focal point in a dark, moody scene with a warm wood table and restrained bokeh. The nori sticks out awkwardly, but there is no clutter or UI." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, corn, scallions, nori and noodles all read as plant-based. No animal ingredients are visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.7333333333333334, "bowl": 0.85, "broth": 0.8166666666666667, "steam": 0.8166666666666667, "composition": 0.8666666666666667 }, "reward": 0.7875000000000001, "sampleStdDev": 0.027041634565979973, "judgments": [ { "noodles": { "score": 0.75, "reason": "Cream wavy strands fan across the mound with consistent thickness, darker strand lines and believable bends; slightly repetitive flat wavy-line pattern and loose mass at the edges keep it from polished." }, "toppings": { "score": 0.8, "reason": "Good variety and readable ingredients: veined pak choi with white stalks, speckled tofu blocks, textured nori sheets with torn edges, scored dark shiitake-like discs, corn, menma-like rolls and scallion rings; corn reads as simple glossy beads and tofu blocks are a bit simplified." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with a red lacquer band and drip glaze onto the dark body, consistent three-quarter perspective, visible foot and contact shadow on the wood; minor simplicity in surface detail." }, "broth": { "score": 0.85, "reason": "Dark amber liquid surface with oil glints, red chili flecks and floating scallion rings; toppings sit with subtle contact shadows and partial immersion rather than floating above." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps rise from the soup and drift; plume contours clearly change across the time-ordered desktop frames (1048ms, 3818ms, 6891ms, pointer frames), spreading and thinning without opaque fog, though it partially veils the nori." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point on a warm wooden table with dark atmospheric bokeh background; no clutter or UI. Mobile crop leaves considerable empty dark space above and below the dish, slightly weakening balance." }, "vegan": { "score": 0.9, "reason": "Visible toppings read as plant-based: pak choi, tofu, menma-like rolls, scored shiitake-style discs, corn, scallion, nori and dark broth; no egg, meat or seafood forms evident, with only mild ambiguity in the beige rolls." } }, { "noodles": { "score": 0.75, "reason": "Noodles form a coherent fanned mound of fine wavy strands with subtle sheen and believable volume in the broth; strands are continuous, though the uniform comb-like wave pattern reads slightly repetitive and stylized." }, "toppings": { "score": 0.65, "reason": "Good variety: bok choy leaves read as real leaves, speckled tofu blocks and scallion rings are convincing, nori sheets have torn irregular edges. Defects: the two dark glossy disks with snowflake stamps read as chocolate cookies rather than ramen toppings, and the corn is rendered as small glossy gold beads; the beige tubes read ambiguously like chikuwa fish cakes." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with visible inner wall, red lacquer rim band, dark glazed body with an intentional drip motif, visible foot and grounded contact shadow; perspective is consistent and highlights are tasteful. No material distortion observed." }, "broth": { "score": 0.8, "reason": "Warm brown broth with restrained sheen, oil droplets, reddish chili-oil patches and gentle surface variation; toppings sit with soft occlusion shadows and the noodle mound meets the liquid plausibly. Slight flatness in the far-left surface." }, "steam": { "score": 0.85, "reason": "Across desktop-1/2/3 and the pointer frames the translucent plume clearly changes shape: narrow column at ~1s, broader forked spread at ~3.8s, diffuse haze by ~6.9s, denser drift in pointer frames. Wisps are irregular and never obscure the food; no opaque fog or static blobs." }, "composition": { "score": 0.9, "reason": "Balanced centered dish on a warm wood table against a dark atmospheric bokeh background; coherent lighting and depth, no text, UI or clutter. The bowl is clearly the focal point; mobile framing leaves generous dark space but remains readable." }, "vegan": { "score": 0.5, "reason": "Mostly plant-based appearance (greens, tofu, scallions, corn, nori), but the two beige ribbed cylinders strongly resemble chikuwa/narutomaki fish-paste cakes, and the dark embossed disks read as chocolate (possibly dairy), making the ingredient set ambiguous rather than clearly vegan." } }, { "noodles": { "score": 0.65, "reason": "Readable strand bundle fanning across the broth with scallion rings on top, but the strands are uniformly parallel and comb-like, laid flat on the surface with little believable immersion; competent yet somewhat mechanical." }, "toppings": { "score": 0.75, "reason": "Good variety: bok choy with leaf veins, sesame-topped tofu blocks, scored shiitake-like caps, corn kernels, scallion rings and torn-edge nori sheets with ribbed texture. Corn reads as glossy spheres and the two striped beige cylinders are stylized and a bit ambiguous, keeping it below excellent." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim, visible interior wall, dark lacquer body with a red drip motif, footed base resting on the planks with contact shadow and rim highlights. Perspective between rim and body is consistent; the drip pattern is a stylistic flourish rather than a defect." }, "broth": { "score": 0.8, "reason": "Warm brown surface with oil droplets, chili flakes, floating scallions and tonal depth; ingredients sit in the liquid with soft occlusion. Slight flatness in the far-left sheen, but overall appetizing and integrated." }, "steam": { "score": 0.75, "reason": "Translucent wisps rise from the soup and their shape visibly shifts across desktop-1/2/3 and the pointer frames, so the steam is animated, not static. The plume is somewhat concentrated in one blobby mass behind the nori and partially veils it." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a dark wood table with restrained bokeh, warm rim lighting and balanced proportions; mobile view crops cleanly with no overflow. No text, UI or competing elements distract from the dish." }, "vegan": { "score": 0.5, "reason": "Greens, tofu, mushrooms, corn, scallion and nori are plant-appropriate, but the two tan striped cylinders resemble naruto fish cake; appearance is ambiguous, so an intermediate score applies." } } ], "diagnostics": { "vegan": 0.6333333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8588888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 133.40000000000146, "frameIntervalP95Ms": 166.59999999999854, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "42dce0ef7f7add6c20803bd5ad516a08c74ebb49384ff8526f837e7a562f6a57", "gradedAt": "2026-09-30T22:27:30.395399+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.43777777777777777, "toppings": 0.5822222222222222, "bowl": 0.6544444444444445, "broth": 0.5866666666666667, "steam": 0.6033333333333333, "composition": 0.7533333333333334 }, "reward": 0.5885, "runId": "google/gemini-3.8-flash/low", "rewardStdDev": 0.04186983799029241, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.45666666666666667, "toppings": 0.5533333333333333, "bowl": 0.64, "broth": 0.5266666666666667, "steam": 0.49333333333333335, "composition": 0.79 }, "reward": 0.5633333333333334, "sampleStdDev": 0.005507570547286107, "judgments": [ { "noodles": { "score": 0.46, "reason": "Readable pale strands overlap and pass beneath toppings, but angular, wire-like loops dominate, with little supple curvature or noodle volume." }, "toppings": { "score": 0.58, "reason": "Mushroom cap cuts, leaf veins, scallion rings and toasted tofu marks provide variety and recognition. Shapes remain simplified: mushrooms resemble disks, nori is a rigid rectangle, and tofu lacks food texture." }, "bowl": { "score": 0.64, "reason": "Continuous elliptical rim, layered thickness and restrained highlights give a coherent vessel. The nearly invisible exterior and broad dark shadow leave its wall depth and support weakly established." }, "broth": { "score": 0.51, "reason": "Warm amber shading and dim submerged strands suggest depth, but concentric bands and a matte, uniform surface read more like a shaded disc than liquid; topping immersion is limited." }, "steam": { "score": 0.48, "reason": "Translucent haze changes position and density across normal frames without obscuring the food. It forms diffuse detached patches above the bowl rather than clearly connected, tapering rising wisps." }, "composition": { "score": 0.78, "reason": "The single centered dish stays readable against a restrained dark brown background, with harmonious warm food colors and no clutter. The oversized shadow and sparse upper space slightly weaken the balance." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, scallions, seaweed, corn-like kernels and tofu-like grilled pieces plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Pale strands are readable and overlap toppings, but angular, sparse looping paths look more like wire outlines than supple noodles with volume." }, "toppings": { "score": 0.58, "reason": "Veined greens, scored mushroom caps, scallion rings and grilled tofu provide recognizable variety. Shapes remain simplified: mushrooms are smooth disks, tofu is rounded blocks, and nori is a rigid rectangular card." }, "bowl": { "score": 0.64, "reason": "Continuous rim bands and restrained highlights form a coherent near-overhead vessel. The shallow visible exterior and oversized dark oval shadow give limited wall depth and an uncertain sense of contact." }, "broth": { "score": 0.53, "reason": "Warm amber shading and faint submerged details suggest soup, but broad concentric bands resemble a solid shaded surface; ingredient immersion and liquid sheen are weak." }, "steam": { "score": 0.48, "reason": "Normal frames show shifting translucent haze above the bowl without obscuring food. It is mostly diffuse detached patches rather than distinct wisps visibly emerging from the soup and rising." }, "composition": { "score": 0.77, "reason": "A single centered dish remains readable against a restrained dark brown background, with warm colors and no distracting text or UI. The large black shadow and generous empty space make the presentation slightly flat." }, "vegan": { "score": 1, "reason": "Visible mushrooms, greens, seaweed, scallions, corn-like kernels and tofu-like grilled pieces plausibly form vegan ramen; no animal-derived ingredient is evident." } }, { "noodles": { "score": 0.45, "reason": "Pale strands are readable with some overlap and submerged fading, but angular, wire-like loops dominate and convey little supple noodle volume." }, "toppings": { "score": 0.5, "reason": "Mushroom scoring, leaf veins, scallion rings and tofu grill marks identify varied ingredients, but smooth disks, rounded blocks and a rigid rectangular nori sheet remain mostly symbolic." }, "bowl": { "score": 0.64, "reason": "The continuous shaded rim and inner cavity have coherent near-overhead geometry. Minimal exterior wall depth and an oversized offset shadow make support and ceramic volume less convincing." }, "broth": { "score": 0.54, "reason": "Warm amber shading and faint submerged strands suggest soup depth, but broad concentric bands feel graphic and toppings lack convincing liquid contact or menisci." }, "steam": { "score": 0.52, "reason": "Translucent haze changes position and density across normal frames without obscuring the food, but appears as diffuse detached patches rather than clearly soup-rooted rising wisps." }, "composition": { "score": 0.82, "reason": "The centered dish, restrained amber-green palette and uncluttered dark background provide a clear focal point on desktop and mobile. The large heavy shadow slightly weakens the otherwise balanced presentation." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, scallions, corn and nori; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4166666666666667, "toppings": 0.5933333333333333, "bowl": 0.5833333333333334, "broth": 0.6, "steam": 0.6, "composition": 0.67 }, "reward": 0.5653333333333334, "sampleStdDev": 0.008736894948054066, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin cream outline loops that read as wire scribbles on the broth surface. They have no volume or overlap, so they look flat and maze-like." }, "toppings": { "score": 0.58, "reason": "Shiitake with star scoring, a bok choy leaf, nori, tofu with grill marks, corn and scallions are all recognizable and varied. They are small and somewhat flat, the tofu looks like rounded cards, and the nori is a rigid rectangle." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark inner wall, a light outer lip and a cast shadow. The shadow sits low, so the bowl looks unsupported and there is no visible base or body, which limits depth." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a soft radial gradient, faint concentric rings and red oil specks. It reads a bit like a flat disc, and the ingredients sit on the surface more than in it." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps change between frames and drift with the pointer. They are blurry and blobby, sometimes detached from the bowl, and they fade out almost entirely in some frames." }, "composition": { "score": 0.68, "reason": "A single centered bowl on a dark, moody background keeps the focus on the food, with no clutter or text. The palette is harmonious, though the bowl is small and the bottom shadow is heavy." }, "vegan": { "score": 0.93, "reason": "Mushrooms, bok choy, nori, tofu, corn and scallions all look plant-based. There is no visible meat or egg." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin cream outlines that look like wire scribbles or a maze on the broth surface. They have no volume or overlap and only partly read as noodles." }, "toppings": { "score": 0.6, "reason": "Shiitake with star scoring, a bok choy leaf, nori, fried tofu with grill marks, corn and scallion bits are all recognizable and varied. They are small and somewhat flat, and the nori is a rigid rectangle." }, "bowl": { "score": 0.55, "reason": "The bowl is an ellipse with a rim, a dark wall and a ground shadow, but it has no visible base or depth. It reads as a flat disc viewed from above, and the contact with the shadow is weak." }, "broth": { "score": 0.6, "reason": "The broth is warm amber with a soft gradient and faint concentric rings and red oil spots. It looks somewhat flat, and the ingredients sit on the surface instead of being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps change position between frames and drift sideways, and they react to the pointer. They are blotchy and blobby at times, and some are disconnected from the bowl." }, "composition": { "score": 0.68, "reason": "The bowl is a centered focal point on a dark, moody background with a warm palette and no clutter. The bowl is small in the frame and the lower shadow area is large and empty." }, "vegan": { "score": 0.93, "reason": "The toppings are mushrooms, bok choy, tofu, nori, corn and scallion, and no animal products are visible." } }, { "noodles": { "score": 0.45, "reason": "Noodles are thin cream outlines that read as wiggly line scribbles. They lack volume and overlap, and they sit flat on the surface, so they read as a maze of loops." }, "toppings": { "score": 0.6, "reason": "The toppings are recognizable: shiitake with star scoring, a bok choy leaf with veins, nori, grilled tofu with grill marks, corn and scallion. Forms are small and somewhat flat. The tofu is a little plasticky, and the nori is a plain rectangle." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark wall and a cast shadow. It is fairly flat and lacks visible ceramic highlights. The base and contact are vague, and the shadow sits oddly low." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a soft gradient and subtle ripple rings. It is appetizing but fairly flat, and the ingredients look pasted on top without much immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Soft translucent steam wisps change shape and position across frames and drift with the pointer. They are blurry blobs, somewhat mechanical at times, and they fade in and out. They do not obscure the food." }, "composition": { "score": 0.65, "reason": "The bowl is centered in a dark, moody scene with a clear focal dish and warm colors. The bowl is a bit small and the shadow is heavy, but there is no clutter." }, "vegan": { "score": 0.95, "reason": "Toppings are mushrooms, bok choy, nori, tofu, corn and scallion, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.44, "toppings": 0.6, "bowl": 0.74, "broth": 0.6333333333333333, "steam": 0.7166666666666667, "composition": 0.8 }, "reward": 0.6368333333333334, "sampleStdDev": 0.03600462933198085, "judgments": [ { "noodles": { "score": 0.42, "reason": "Noodles read as thin uniform pale lines with sharp angular zigzag bends forming a wire/maze-like network over the broth; continuous and contained within the bowl but lacking supple curves, thickness variation, or layered volume, matching the rigid-scribble defect capped at .50." }, "toppings": { "score": 0.6, "reason": "Recognizable stylized set: two brown mushrooms with star score marks, a leafy green with a pale bulb (scallion/leek), scattered scallion slices, chili flecks, yellow garnish beads, and a dark nori rectangle. Defects: the nori is a featureless flat rectangle and the two grill-marked pieces are card-like flat slabs; several items read as simple geometric tokens rather than textured food." }, "bowl": { "score": 0.72, "reason": "Coherent top-down bowl: metallic rim ring with left highlight, dark inner wall ring, clean circular geometry, and a soft contact shadow grounding it. Slight shortcomings: construction is essentially flat concentric rings with minimal wall depth, and the metallic sheen reads a bit like a saucepan rather than ceramic/lacquer." }, "broth": { "score": 0.65, "reason": "Warm tan broth with radial gradient, restrained sheen, oil droplets and chili specks, and subtle shadows under toppings suggesting immersion. Shortcomings: surface is fairly flat and uniform with faint concentric ring banding, and ingredients sit atop rather than visibly displacing liquid." }, "steam": { "score": 0.55, "reason": "Steam changes across frames (desktop-1/2/3 and pointer frames show differing wisp configurations), so it is genuinely animated, translucent, and does not hide the food. Defects: wisps are amorphous gray blobs that frequently float detached from the bowl (large cloud off to the left in pointer-left, plume disconnected at upper right in pointer-right) rather than rising continuously from the soup surface." }, "composition": { "score": 0.75, "reason": "Centered focal dish on a dark atmospheric ground with warm rim lighting, no clutter, text, or competing elements, and modest overflow-free layout. Shortcomings: large empty dark regions (top half of the mobile frame) and the glow-lit void does most of the atmosphere while the dish itself is fairly small." }, "vegan": { "score": 0.9, "reason": "Visible toppings are mushrooms, leafy green, scallion, nori, corn-like garnish and chili in plant-toned broth. The grill-marked brown rectangles are ambiguous (could be grilled tofu/aburaage rather than meat), so full credit is withheld, but no clearly animal-derived ingredient is visible." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, uniform pale-yellow lines forming wire-like loops and repeated flat rings with polygonal bends; they lack thickness variation, volume, or believable submersion, reading more as connected wire scribbles than supple strands." }, "toppings": { "score": 0.6, "reason": "Recognizable variety: two shiitake-like mushrooms with cross-cut asterisks, a bok-choy-style leaf, a dark nori sheet, scallion rings, corn kernels and chili flecks. However, the tofu pieces are flat glossy rounded rectangles and the greens are simple balloon-like shapes, giving a geometric, sticker-like feel." }, "bowl": { "score": 0.75, "reason": "Coherent top-down construction: metallic gradient rim, dark inner edge, circular cavity and an offset contact shadow grounding the bowl. Rim highlight and consistent circular perspective are clean, though the rim reads as brushed steel rather than ceramic and the bowl is otherwise quite plain." }, "broth": { "score": 0.6, "reason": "Warm tan liquid with a soft radial sheen and subtle depth, but a faint concentric ring is visible inside and toppings/noodles sit as outlines pasted on the surface with little occlusion or immersion, weakening the sense of real soup." }, "steam": { "score": 0.75, "reason": "Across desktop-1/2/3 and pointer frames the translucent grey wisps visibly shift shape and position around and above the bowl (right wisp in pointer-right, left mass in pointer-left), showing continuous rise and fade without obscuring the food; wisps are slightly blobby in places." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark warm backdrop, balanced framing, attractive rim glow and shadow, no clutter or text; the steam drift adds atmosphere without competing with the food." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, greens, nori, tofu, corn, scallions and chili in a tan plant-style broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.5, "reason": "Noodles are continuous thin pale-yellow squiggles spanning the bowl, but they read as a flat tangled wire/maze scribble with uniform hairline width, no volume or layered overlap, floating on the surface rather than sitting in broth." }, "toppings": { "score": 0.6, "reason": "Recognizable stylized set: two shiitake discs with star cuts, a bok choy leaf with pale stem, a speckled dark nori sheet, scallion slices, corn, chili oil, and grill-marked tofu pieces. Details like nori speckles and grill marks help, but the tofu reads as flat rigid rectangles and the mushrooms as flat disks with little texture." }, "bowl": { "score": 0.75, "reason": "Coherent straight top-down bowl: metallic-toned rim band, dark inner rim ring defining the cavity, specular highlight on the lower-right rim, and an elliptical ground shadow supporting it. Slightly plain and shallow, but no broken geometry or perspective clash for the chosen angle." }, "broth": { "score": 0.65, "reason": "Warm tonkotsu-like tan surface with soft central glow, chili-oil specks and a faint inner ring; restrained sheen. However the liquid reads somewhat flat, and toppings/noodles sit pasted on top with little immersion, shadowing or occlusion." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps that change shape frame-to-frame (desktop 1-3), drift right in pointer-right and left in pointer-left, and disappear entirely in reduced-motion frames. Wisps stay clear of the food; only minor blobbiness in places." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric scene with the bowl as clear focal point, warm rim light, soft ground shadow, and no distracting text or UI; mobile framing also keeps the bowl intact. Presentational restraint is good, though the background is minimal." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, bok choy, nori, tofu/grill-marked plant protein, corn, scallions and chili oil; no meat, fish, egg or dairy visible." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9677777777777777, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "c0c5269964a47054e0224bf1984ab61d6e06cd196f05069981c3848e50629b63", "gradedAt": "2026-09-30T22:27:31.227460+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4211111111111111, "toppings": 0.508888888888889, "bowl": 0.6666666666666666, "broth": 0.5833333333333334, "steam": 0.6488888888888888, "composition": 0.7555555555555555 }, "reward": 0.5760000000000001, "runId": "meta/muse-spark-1.3/xhigh", "rewardStdDev": 0.09939371766420205, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.46, "bowl": 0.7166666666666667, "broth": 0.5666666666666667, "steam": 0.58, "composition": 0.82 }, "reward": 0.5800000000000001, "sampleStdDev": 0.0025000000000000022, "judgments": [ { "noodles": { "score": 0.48, "reason": "Thin pale curved strands are visible beneath the toppings, but form a sparse, flat tangle with little coherent overlap, volume or convincing immersion." }, "toppings": { "score": 0.46, "reason": "Mushroom caps have readable cuts and scallion slices offer variety, but glossy yellow spheres, oval leaf symbols and a thick rounded nori card dominate the artificial food depiction." }, "bowl": { "score": 0.73, "reason": "The elliptical rim and curved walls have coherent perspective, thickness and restrained highlights. The flattened foot meets the body awkwardly, weakening the otherwise plausible ceramic construction." }, "broth": { "score": 0.57, "reason": "Warm amber color and a localized sheen suggest soup, but the surface is largely flat and opaque, with ingredients appearing layered above it rather than convincingly immersed." }, "steam": { "score": 0.56, "reason": "Translucent haze changes position and distribution across normal frames without obscuring the food. It resembles broad disconnected blurred clouds more than delicate wisps emerging from the soup." }, "composition": { "score": 0.82, "reason": "A single readable dish anchors the restrained dark scene, with warm food colors, subtle tabletop shadows and ample negative space. No distracting text or UI is present; the dish is somewhat small on mobile." }, "vegan": { "score": 0.92, "reason": "Visible mushrooms, scallions, greens, seaweed and pale tofu-like pieces plausibly indicate vegan ramen. The yellow spherical topping is unclear, but no unmistakable animal ingredient is visible." } }, { "noodles": { "score": 0.48, "reason": "Thin pale strands have curved bends and some overlap, but form a sparse, flat scribble with little coherent noodle volume or convincing immersion." }, "toppings": { "score": 0.46, "reason": "Mushroom caps, scallion rings and pale tofu pieces are recognizable, but glossy yellow spheres, disk-like greens and a thick rounded rectangular nori card make much of the food look generic and plastic." }, "bowl": { "score": 0.7, "reason": "The elliptical rim and rounded walls have coherent perspective, readable thickness and restrained highlights. The flattened foot has an abrupt, weakly articulated connection to the body." }, "broth": { "score": 0.56, "reason": "Warm amber color and localized sheen suggest soup, but the surface is mostly a smooth opaque plane; toppings have limited liquid contact and appear laid on top." }, "steam": { "score": 0.59, "reason": "Translucent clouds change position and shape across normal frames without obscuring the food. They read more as separated soft fog patches than connected wisps emerging from the soup." }, "composition": { "score": 0.82, "reason": "A single clearly readable dish, warm food against a dark blue background, restrained table lighting and ample negative space create a balanced atmospheric presentation without distracting text or UI. The tall steam area slightly outweighs the relatively small dish, especially on mobile." }, "vegan": { "score": 0.9, "reason": "Visible mushrooms, green vegetables, seaweed and tofu-like pieces plausibly form vegan ramen. The yellow round toppings are ambiguous, but no unmistakable animal ingredient is visible." } }, { "noodles": { "score": 0.48, "reason": "Thin pale curves suggest noodles beneath the toppings, but mostly read as flat, loosely tangled linework with little coherent overlap or strand volume." }, "toppings": { "score": 0.46, "reason": "Mushroom caps, tofu and green slices are recognizable, but glossy yellow spheres, oval leaf symbols and thick rounded rectangular nori make much of the food look geometric and plastic." }, "bowl": { "score": 0.72, "reason": "The elliptical rim and curved dark body establish a coherent cavity with restrained highlights. The flattened foot joins the body awkwardly and gives the support a slightly separate appearance." }, "broth": { "score": 0.57, "reason": "Warm amber shading and a soft highlight suggest liquid, but the surface lacks convincing depth and meniscus contact; several toppings appear laid on top rather than immersed." }, "steam": { "score": 0.59, "reason": "Translucent clouds change position and shape across normal frames without obscuring the food. They are broad, disconnected blurry patches rather than fine wisps visibly emerging from the soup." }, "composition": { "score": 0.82, "reason": "A single readable dish anchors the dark scene, with warm food colors contrasting the blue-black bowl and ample uncluttered space. Mobile framing preserves the focal dish; the large upper void slightly weakens the balance." }, "vegan": { "score": 0.9, "reason": "Visible mushrooms, tofu-like pieces, greens, seaweed and noodles plausibly form vegan ramen. The yellow spherical topping is ambiguous, but no clearly animal-derived ingredient is visible." } } ], "diagnostics": { "vegan": 0.9066666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.45, "bowl": 0.5, "broth": 0.5166666666666667, "steam": 0.5833333333333334, "composition": 0.63 }, "reward": 0.4746666666666667, "sampleStdDev": 0.007751343969488997, "judgments": [ { "noodles": { "score": 0.3, "reason": "Noodles are only a few thin pale threads (looks like bean sprouts) in the center; there is no real noodle volume and no supple strands, so they are barely readable as ramen noodles." }, "toppings": { "score": 0.45, "reason": "Shiitake caps, scallion slices, a nori card, tofu cubes and bok choy are recognizable. Yet the nori is a rigid rounded card, the yellow glossy spheres look like plastic, and the green leaves are flat discs." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim, a glossy dark body and a copper band, but the foot is a detached floating ellipse below the bowl with no contact to the surface, which reads as a construction flaw." }, "broth": { "score": 0.5, "reason": "The orange broth has a warm gradient and reads as soup, but it is flat and the toppings sit on it without much immersion or sheen." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise and change shape between frames, without hiding the food. They are blurry blobs, somewhat detached from the bowl." }, "composition": { "score": 0.62, "reason": "The bowl is centered and lit warmly against a dark atmospheric backdrop, with no clutter or text. The floating base and small scale of the food limit it." }, "vegan": { "score": 0.85, "reason": "Mushrooms, vegetables, tofu and nori are all plant-based. The yellow spheres are ambiguous, possibly corn or yuzu rather than egg." } }, { "noodles": { "score": 0.3, "reason": "Noodles are only tiny thin pale threads in the center, reading more like bean sprouts or wisps. There is no noodle volume or supple strands, so the noodle mass is indistinct." }, "toppings": { "score": 0.45, "reason": "The toppings are recognizable: shiitake caps with slits, a nori card, cucumber or scallion slices, tofu cubes and bok choy. However, the nori is a rigid rounded card, the golden balls look like glossy plastic spheres, and the green leaves are plain discs." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent rim, a dark glossy body with a highlight, and a gold inner line. The foot is detached and floats below the body as a separate ellipse with a gap, so it looks unsupported." }, "broth": { "score": 0.5, "reason": "The orange broth is warm and reads as a liquid with a slight gradient. It is fairly flat, and the toppings look pasted on the surface with little immersion or ripple." }, "steam": { "score": 0.55, "reason": "Soft translucent wisps rise and drift differently between frames, and they fade out without hiding the food. They are blurry blobs, somewhat mechanical and faint." }, "composition": { "score": 0.62, "reason": "The bowl is centered as the clear focal point in a dark moody scene with warm lighting and no clutter. The bowl is small in the frame, and the floating base and reflection feel odd." }, "vegan": { "score": 0.85, "reason": "Mushrooms, tofu, greens, nori and sprout-like noodles all look plant-based. The golden balls are ambiguous but are not clearly egg or other animal product." } }, { "noodles": { "score": 0.3, "reason": "Noodles are only faint thin pale threads in the center of the broth, barely visible and reading as bean sprouts or scribbles. There is no noodle volume or supple strands." }, "toppings": { "score": 0.45, "reason": "Mushrooms with gill marks, scallion rings, tofu cubes, bok choy and a dark nori card are recognizable. The nori is a rigid upright rounded rectangle, and the glossy yellow spheres look like plastic balls. The greens are flat disks, which caps the score." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim, a highlight and a copper band, but the foot is a detached dark ellipse floating below the body with a gap. This looks unsupported and breaks the bowl's contact with the table." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm glow and a subtle gradient, and it reads as appetizing. It is fairly flat, and toppings sit on top with little immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise and shift between frames, and they fade without hiding the food. They are blurry blobs, somewhat mechanical and offset from the bowl." }, "composition": { "score": 0.65, "reason": "The bowl is centered and lit warmly against a dark atmospheric background, with no clutter or text. The floating foot and small scale limit the result." }, "vegan": { "score": 0.85, "reason": "Mushrooms, tofu, greens, nori and sprouts all look plant-based. The yellow balls are ambiguous and could be read as egg yolks or corn." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.48333333333333334, "toppings": 0.6166666666666667, "bowl": 0.7833333333333333, "broth": 0.6666666666666666, "steam": 0.7833333333333333, "composition": 0.8166666666666667 }, "reward": 0.6733333333333333, "sampleStdDev": 0.02268442931469363, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles appear only as thin, sparse pale strokes woven among the center toppings; they curve smoothly but lack strand volume, overlap and a convincing noodle bed, reading more as delicate garnish wisps than a full portion." }, "toppings": { "score": 0.65, "reason": "Good variety with readable shiitake caps (gill marks), scallion rings, tofu cubes, chili threads and standing nori sheets; however the nori reads slightly as stiff rounded cards and the gold cluster is a group of glossy plastic-looking spheres rather than food." }, "bowl": { "score": 0.8, "reason": "Coherent dark lacquer bowl with a clean elliptical rim, matching broth ellipse, tapering body, small foot, side highlight and grounded shadow on the table; construction is consistent with only minor stiffness." }, "broth": { "score": 0.6, "reason": "Amber broth surface has a pleasant sheen with lighter swirl highlights and no target-disc artifact, but immersion is weak: the corn cluster, leek pieces and mushrooms sit largely pasted on the surface plane with little occlusion or contact shading." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps clearly change shape and position across the time-ordered desktop frames, rising, spreading and fading above the bowl, and freeze in the reduced-motion frames as expected; they are slightly cloud-blobby but never hide the food." }, "composition": { "score": 0.85, "reason": "Balanced dark scene with the bowl as a clear focal point, warm rim light on the broth, a soft overhead shaft, subtle dust particles and no text, UI or clutter; presentation is atmospheric and restrained." }, "vegan": { "score": 1, "reason": "All visible ingredients are plant-derived: nori, shiitake mushrooms, scallions, tofu, corn-like kernels, chili threads, noodles and plant broth; no meat, fish, egg or dairy appears." } }, { "noodles": { "score": 0.45, "reason": "Noodles appear only as thin pale wavy squiggle lines half-buried under toppings in the broth center. They are flat, wire-like strands without volume or coherent mass, and much of the bowl surface shows bare broth rather than a believable noodle bed." }, "toppings": { "score": 0.6, "reason": "Good variety and readability: a standing nori sheet with highlight, mushroom caps with gill marks, green scallion rings, tofu cubes, and cream menma sticks. However, the cluster of glossy yellow spheres reads as polished plastic balls rather than food, and the flat ring shapes are somewhat generic." }, "bowl": { "score": 0.75, "reason": "Coherent dark lacquer-style bowl with a wide elliptical rim, tapering body, foot ring, body highlight, and grounded contact shadow. Construction is consistent for the flat-illustration style; only minor simplification of the rim band and foot joint." }, "broth": { "score": 0.7, "reason": "Warm amber broth with glossy sheen, chili-oil specks and a lighter swirl; toppings sit at the surface with plausible contact. Slightly flat and the highlight blob is a bit stylized, but it reads as liquid." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps above the bowl visibly change shape, spread and drift across desktop-1 through desktop-3 and pointer frames, indicating continuous motion. Wisps remain irregular and do not hide the food, though some puffs read slightly blobby and detached." }, "composition": { "score": 0.8, "reason": "Centered focal dish against a dark atmospheric background with subtle floating specks, warm under-glow and vignette; no text, UI or clutter. Presentation is clean and balanced, with steam leading the eye into the bowl." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are nori, mushrooms, scallions, tofu, menma, chili and corn-like yellow spheres; no meat, fish, egg or dairy-like toppings are present." } }, { "noodles": { "score": 0.45, "reason": "Noodles are only faint pale squiggled strokes peeking under the scallions and near the corn cluster; they lack volume, thickness variation and coherent strand overlap, reading as thin scratchy lines rather than supple strands." }, "toppings": { "score": 0.6, "reason": "Good variety: readable shiitake caps with gill marks, scallion rings, nori sheet, tofu slabs and chili threads. But the corn is a cluster of glossy golden spheres and a shiny green ball sits by the nori \u2014 plastic-bead look; the nori reads as a thick rounded card." }, "bowl": { "score": 0.8, "reason": "Coherent dark hemisphere: consistent elliptical rim, curving walls with a left highlight, a plausible foot ellipse and soft ground shadow; geometry holds across frames. Minor: rim band is somewhat heavy and the inner rim edge line is thin." }, "broth": { "score": 0.7, "reason": "Warm orange surface with radial glow, darker edge vignette and contact shadows under mushrooms and corn give some depth and integration; but the liquid reads fairly flat and toppings appear laid on the surface rather than partly submerged." }, "steam": { "score": 0.8, "reason": "Across the time-ordered desktop frames the wisps change shape, position and dispersion (shifted plumes in each frame), rising translucently above the bowl without hiding the food; reduced frames show the expected static state." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a restrained dark scene with subtle particle specks, warm floor light pool and steam occupying the upper space; no clutter or text. The bowl is a bit small relative to the canvas and the surroundings are sparse." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, scallions, corn, nori, tofu and chili threads in plant broth; no animal-derived ingredients apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9188888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 83.30000000000109, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "769289633bf544fab1b88247a5f2c3fc73c5fb437616e27a0ae61dc4007db5b9", "gradedAt": "2026-09-30T22:27:32.168131+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6388888888888888, "toppings": 0.8022222222222223, "bowl": 0.8488888888888889, "broth": 0.7100000000000001, "steam": 0.7411111111111112, "composition": 0.8588888888888889 }, "reward": 0.7661666666666668, "runId": "openai/gpt-6-astra/high", "rewardStdDev": 0.03318676108196025, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.8266666666666667, "bowl": 0.92, "broth": 0.68, "steam": 0.79, "composition": 0.9033333333333333 }, "reward": 0.7675000000000001, "sampleStdDev": 0.0032787192621510286, "judgments": [ { "noodles": { "score": 0.5, "reason": "Clearly defined golden strands have consistent thickness and some convincing crossings, but repeated nested oval loops dominate, making the mound look flat and mechanically coiled rather than supple." }, "toppings": { "score": 0.83, "reason": "Veined, scalloped leaves, textured mushroom caps with cut markings, porous browned tofu, radish slices and scallion rings provide readable variety. The nori is rather rectangular and the leaves somewhat uniformly arranged." }, "bowl": { "score": 0.92, "reason": "A coherent elliptical rim, rounded tapering walls and attached foot establish convincing depth and support. Fine ceramic speckles, restrained highlights and a contact shadow enrich the material; no material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading and scattered oil droplets suggest appetizing broth, but the surface is quite flat and opaque, with limited immersion cues around the tofu and noodle mound." }, "steam": { "score": 0.8, "reason": "Translucent curling wisps emerge behind the food and visibly change shape across the normal frames without obscuring ingredients. They remain somewhat evenly spaced and similar in height; continuous fading cannot be fully established from these captures." }, "composition": { "score": 0.91, "reason": "The single dish is cleanly framed against a restrained dark background, with warm food colors balanced by green foliage and dark ceramic. Lighting and contact shadows are coherent, and the mobile view retains the whole bowl; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushrooms, leafy vegetables, radish, scallions and seaweed, with no discernible animal-derived ingredient." } }, { "noodles": { "score": 0.5, "reason": "Individual golden strands have clear outlines and overlaps, but repeated flattened oval loops dominate, making the mound look tightly coiled rather than supple and loosely piled." }, "toppings": { "score": 0.83, "reason": "Veined, scalloped greens, scored mushroom caps with visible undersides, porous browned tofu, and radish slices provide detailed variety. The nori is somewhat rectangular and the leaves slightly uniform." }, "bowl": { "score": 0.92, "reason": "The continuous elliptical rim, shaded curved walls, subtle ceramic speckling, attached foot, and grounding shadow form a convincing vessel. No material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading, scattered oil droplets, and edge highlights suggest appetizing soup, but ingredient boundaries show limited immersion and the surface feels somewhat flat." }, "steam": { "score": 0.79, "reason": "Soft translucent wisps change curvature and position across normal frames, rising above the soup without obscuring it. Their broad blurred shapes provide less delicate branching and fading detail." }, "composition": { "score": 0.9, "reason": "The single centered dish is clearly readable against a restrained dark background, with warm food colors balanced by green toppings and a dark bowl. There is no distracting text or clutter; mobile framing leaves generous empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, radish, scallions, and seaweed; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Thin continuous strands have shaded edges and readable overlaps, but repeated elongated loops dominate, making the mound look like a flat coil rather than supple, loosely tangled noodles." }, "toppings": { "score": 0.82, "reason": "Veined scalloped leaves, speckled mushroom caps with scored tops, porous browned tofu, radish slices and scallion rings provide clear variety and detail. The leaves are somewhat uniformly upright and the nori has an overly straight rectangular edge." }, "bowl": { "score": 0.92, "reason": "A coherent elliptical rim, curved deep walls, attached foot and grounding shadow establish convincing construction. Subtle glaze speckles, horizontal texture and restrained highlights support the ceramic appearance; no material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading, scattered oil droplets and curved highlights suggest an appetizing liquid. Ingredient contact is less convincing: tofu and the noodle mound appear largely perched on the surface with little visible immersion." }, "steam": { "score": 0.78, "reason": "Normal frames show changing translucent wisps extending upward from the soup without obscuring it. Their soft fading edges work well, though the plumes remain fairly parallel and evenly spaced, and the sampled frames cannot establish fully continuous motion." }, "composition": { "score": 0.9, "reason": "The single dish is clearly framed against a restrained dark background, with warm broth contrasting green toppings and dark ceramic. Lighting and contact shadow give depth without clutter or text; mobile framing leaves rather generous vertical empty space." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushrooms, leafy vegetables, radish, scallions and seaweed alongside noodles; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.72, "toppings": 0.78, "bowl": 0.7766666666666667, "broth": 0.6733333333333333, "steam": 0.6, "composition": 0.8066666666666666 }, "reward": 0.7323333333333334, "sampleStdDev": 0.0070237691685684465, "judgments": [ { "noodles": { "score": 0.72, "reason": "Fine, continuous strands form a nested nest with coherent overlap and bends, plus a slight sheen. The nest sits as a flat raft on top of the broth, and the looping pattern looks somewhat repetitive." }, "toppings": { "score": 0.78, "reason": "Leaves with veins, shiitake with star cuts, seared tofu cubes with grain and pepper, radish slices, scallion rings and a nori piece give good variety and readable detail. The tofu cubes look slightly rigid, and the nori is tiny." }, "bowl": { "score": 0.75, "reason": "The dark ceramic bowl has a coherent rim, tapered walls, a foot ring, speckle texture and a cast shadow. Perspective is consistent. The tofu overlaps the rim, which is a minor contact oddity." }, "broth": { "score": 0.66, "reason": "The warm orange-brown broth has oil droplets, chili flecks and a soft sheen. It reads a bit flat and opaque, and the noodles and toppings have little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise above the bowl and shift shape between frames. They are very subtle and somewhat uniform, but they don't hide the food." }, "composition": { "score": 0.8, "reason": "The bowl is centered as a clear focal point against a dark vignette. The palette is warm and harmonious, and there is no clutter or text." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, greens, radish, scallions and noodles are all plant-based, with no visible animal products." } }, { "noodles": { "score": 0.72, "reason": "Fine, overlapping noodle strands form a nest with believable swirls and thickness variation, and scattered scallion and chili adds detail. The nest looks like a tidy flat mat resting on the broth, with repeated, fairly uniform loops and little sense of immersion." }, "toppings": { "score": 0.78, "reason": "Leafy greens with veins, shiitake with star cracks, toasted tofu cubes with speckles, radish slices, scallion rings and a small nori piece are all readable and varied. The tofu cubes are rigid and slightly blocky, the nori is tiny, and the tofu on the rim looks a bit pasted on." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim, dark speckled glaze, curved walls, foot ring, soft contact shadow and rim highlights. The perspective is consistent. Minor flaws are a slightly flat body shading and a faint vertical highlight band." }, "broth": { "score": 0.68, "reason": "Warm orange-brown broth with oil droplets and chili oil swirls has a believable depth gradient. It reads somewhat flat and opaque, and the toppings sit on it with limited occlusion or ripples." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise and shift shape across frames, without hiding the food. They are very subtle, somewhat uniform in shape, and rise from above the rim area rather than clearly from the soup." }, "composition": { "score": 0.82, "reason": "Centered single bowl on a dark, restrained background with warm food colors and a clear focal point. The layout is balanced, and there is no clutter or text. The bowl fills the frame well." }, "vegan": { "score": 0.95, "reason": "Only noodles, greens, mushrooms, tofu, radish, scallion and nori are visible, with no egg, meat or fish." } }, { "noodles": { "score": 0.72, "reason": "Fine, continuous strands form a nest with believable overlap and slight bends, plus scallion and chili garnish. The nest reads as a flat, tidy oval sitting on top of the broth, and the strands are fairly uniform and regular." }, "toppings": { "score": 0.78, "reason": "Veined lettuce leaves, star-scored shiitake, toasted tofu cubes with speckles, radish slices and a nori square are all recognizable and varied. The tofu cubes look slightly rigid and one sits on the rim. The nori is tiny and hidden." }, "bowl": { "score": 0.78, "reason": "Dark ceramic bowl with a coherent elliptical rim, curved walls, foot ring, subtle speckle and highlight, and a grounding shadow. Perspective is consistent. The tofu overlapping the rim and the flat base area are minor flaws." }, "broth": { "score": 0.68, "reason": "Warm orange-brown broth with oil droplets and chili-oil specks reads as appetizing. The surface is fairly flat and opaque, and the noodles and toppings look placed on top with limited immersion or contact." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise above the bowl and change shape across frames, without hiding the food. They are very subtle and somewhat repetitive, and they barely connect to the soup surface." }, "composition": { "score": 0.8, "reason": "A centered bowl on a dark, restrained background keeps the food as the focus. Colors are harmonious, with orange broth against the green bowl and greens. There is no clutter or text." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, lettuce, radish, scallions, nori and noodles are visible. No egg, meat or fish appears." } } ], "diagnostics": { "vegan": 0.9566666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6966666666666667, "toppings": 0.8, "bowl": 0.85, "broth": 0.7766666666666667, "steam": 0.8333333333333334, "composition": 0.8666666666666667 }, "reward": 0.7986666666666667, "sampleStdDev": 0.020039544239661086, "judgments": [ { "noodles": { "score": 0.72, "reason": "Coherent golden nest of thin continuous strands with varied bends, crossings and draped ends; the left half is dominated by a repetitive concentric loop swirl that reads slightly drawn/mechanical rather than natural." }, "toppings": { "score": 0.82, "reason": "Good variety and readability: veined bok-choy leaves, three shiitake caps, seared tofu cubes, pink-rimmed radish slices, scallion rings, chili rounds, sesame and a nori sheet. Minor flaws: the four leaves are near-identical, tofu cubes are blocky/cartoonish, and the nori is barely visible." }, "bowl": { "score": 0.85, "reason": "Well-constructed speckled dark ceramic bowl: coherent elliptical rim, curving walls, visible foot with grounding shadow and consistent perspective. Minor: the rim highlight ring is thin and flat, and the light rim band reads a bit synthetic." }, "broth": { "score": 0.78, "reason": "Appetizing orange miso-toned surface with restrained sheen, oil droplets, chili-oil streaks and a darker ring near the rim. Most ingredients sit convincingly in the liquid, but the right-hand tofu cubes perch on the rim edge above the broth rather than sinking into it." }, "steam": { "score": 0.8, "reason": "Translucent irregular gray wisps rise from the soup without hiding the food. Comparing desktop-1/2/3 and the pointer frames, the plume shapes clearly change over time, showing continuous animated motion rather than static smoke." }, "composition": { "score": 0.87, "reason": "Centered, balanced presentation with the bowl as a clear focal point, warm broth and steam contrasting against a clean dark vignette; no text, UI or clutter, and the mobile crop stays well composed." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, shiitake mushrooms, leafy greens, radish slices, scallions, chili, sesame and nori in plant broth; no egg, meat, fish or dairy forms appear." } }, { "noodles": { "score": 0.75, "reason": "A coherent central nest of golden strands with consistent thickness, continuous curves and believable overlap, garnished with scallion rings and chili threads. The nest is slightly too uniform and loop-like in places, reading a bit more like drawn spaghetti than supple ramen strands, but no broken or maze-like scribbles." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: leafy greens with veins and ruffled edges, shiitake caps with star slits and stems, sliced radish with radial interior, scallion rings, red chili slices, a nori sheet, and seasoned tofu cubes. Minor deductions: tofu cubes are a bit blocky and glossy like candy, and the mushroom star markings are repetitive." }, "bowl": { "score": 0.85, "reason": "Coherent dark speckled ceramic bowl: elliptical rim with a lighter glazed band, consistent interior/exterior perspective, tapered walls, visible footed base with contact shadow, and a bright left-side highlight. Speckle texture is tasteful; no visible distortion or detached parts." }, "broth": { "score": 0.8, "reason": "Rich orange-brown broth with restrained sheen, fine speckle and oil-dot texture, a darker edge ring suggesting depth, and credible occlusion shadows where mushrooms, tofu and greens sit in the liquid. Slightly flat gradient near center but convincingly liquid, not solid." }, "steam": { "score": 0.85, "reason": "Translucent wisps rising from the soup that visibly change shape and position across desktop-1/2/3 and the pointer frames, fading high above the bowl without obscuring food. Wisps are slightly faint and sparse, but clearly animated rather than static, and reduced-motion frames appropriately show fainter, calmer steam." }, "composition": { "score": 0.88, "reason": "Centered, well-proportioned focal bowl on a dark atmospheric background with a subtle warm glow and vignette; steam adds depth without clutter. No text, UI or competing elements, and the mobile viewport crops gracefully with the dish still readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, leafy greens, shiitake mushrooms, tofu cubes, scallions, chili, radish and nori in plant broth; no eggs, meat, fish or dairy appear anywhere in the scene." } }, { "noodles": { "score": 0.62, "reason": "Noodle nest reads as thin, tightly coiled concentric loops \u2014 more spaghetti-nest than ramen, lacking waviness and thickness variation; strands are continuous but flat and repetitive, and the nest sits atop the broth with limited immersion." }, "toppings": { "score": 0.78, "reason": "Good variety: bok-choy leaves with vein detail, shiitake caps with cracked-star tops, radish slices, scallion rings, chili and sesame; tofu cubes are a bit hard-edged and the right cubes perch oddly at the rim, and the nori is a flat dark slab." }, "bowl": { "score": 0.85, "reason": "Coherent speckled ceramic bowl with consistent rim ellipse, deep body, visible foot, contact shadow and left-rim highlight; perspective holds together with no broken joins, only slightly uniform rim treatment." }, "broth": { "score": 0.75, "reason": "Appetizing miso-toned surface with darker depth patches, oil specks and chili-oil hints; toppings mostly sit in it, though the surface is a touch flat in sheen and the noodle nest reads as pasted on top rather than immersed." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rise and fade above the bowl and clearly change shape across desktop frames (1\u21923, pointer frames), indicating continuous motion without opaque fog or obscuring the food." }, "composition": { "score": 0.85, "reason": "Centered focal dish against a restrained dark vignette, warm broth contrasting the teal bowl, soft grounding shadow, no clutter or distracting UI; tofu near the right rim is the only slight awkwardness." }, "vegan": { "score": 1.0, "reason": "Visible ingredients \u2014 noodles, greens, shiitake, tofu, scallions, radish, chili, nori, sesame \u2014 are all plant-based; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9855555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 116.70000000000073, "frameIntervalP95Ms": 149.89999999999782, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "db238963da76b47f29cb2b68f2a94e9cc80d70a615fd99689dc8cb2b5c19a84e", "gradedAt": "2026-09-30T22:27:32.938003+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6777777777777778, "toppings": 0.8022222222222223, "bowl": 0.8411111111111111, "broth": 0.7077777777777778, "steam": 0.5277777777777778, "composition": 0.8511111111111112 }, "reward": 0.739388888888889, "runId": "openai/gpt-6-astra/max", "rewardStdDev": 0.08306295024680803, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6266666666666667, "toppings": 0.78, "bowl": 0.92, "broth": 0.6633333333333333, "steam": 0.25, "composition": 0.9 }, "reward": 0.6981666666666667, "sampleStdDev": 0.0030550504633038962, "judgments": [ { "noodles": { "score": 0.62, "reason": "Distinct golden strands have consistent thickness and readable overlaps, but the densely packed, mostly flat looping pattern lacks supple three-dimensional volume." }, "toppings": { "score": 0.78, "reason": "Scored textured mushroom caps, porous grilled tofu, veined greens, and individual corn kernels are clearly readable. Leaves and nori remain rather stiff and planar, and corn is unusually glossy." }, "bowl": { "score": 0.92, "reason": "Coherent elliptical rim, deep curved walls, attached foot, and contact shadow establish convincing construction. Fine ceramic grain and restrained edge highlights support the material; no material geometry defect is observed." }, "broth": { "score": 0.65, "reason": "Warm brown liquid has subtle oil marks and highlights, but its surface is fairly uniform and opaque, with limited immersion cues around the prominently raised ingredients." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps rise from behind the food without obscuring it, but their contours appear unchanged across the three normal desktop frames, reading as static smoke." }, "composition": { "score": 0.9, "reason": "A single uncluttered dish remains the clear focus against a restrained dark background. Warm food colors, green accents, and directional bowl shading balance well, including on mobile; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, seaweed, scallions, and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.64, "reason": "Thin continuous golden strands have readable overlap and shading, but the densely flattened pile has many angular bends and limited supple volume." }, "toppings": { "score": 0.78, "reason": "Scored mushroom caps, porous grilled tofu, veined greens, textured nori and individual corn kernels are clearly differentiated. Leaves and seaweed remain rather rigid and planar, and corn is unusually glossy." }, "bowl": { "score": 0.92, "reason": "Coherent elliptical rim, substantial curved walls, integrated foot and contact shadow establish convincing depth. Fine ceramic texture and restrained edge highlights reinforce the material; no material construction defect is observed." }, "broth": { "score": 0.67, "reason": "Warm brown liquid has small oil marks and restrained highlights, but its surface is quite uniform and opaque, with limited immersion cues around the prominent ingredients." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps emerge behind the food without obscuring it, but their shapes appear unchanged across the normal time-ordered desktop frames." }, "composition": { "score": 0.9, "reason": "A single well-framed dish dominates the uncluttered dark scene, with balanced green and gold accents and coherent warm lighting. Desktop and mobile preserve the complete silhouette; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, seaweed, scallions and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.62, "reason": "Thin continuous golden strands have readable overlaps and some shading, but the dense, largely flat tangle has angular bends and limited supple volume." }, "toppings": { "score": 0.78, "reason": "Textured scored mushroom caps, veined greens, speckled grilled tofu and individual corn kernels are clearly readable. Nori is rather rigid and card-like, and the greens have stiff stems and repetitive scalloped edges." }, "bowl": { "score": 0.92, "reason": "A coherent elliptical rim, shaded inner wall, rounded ceramic body and attached foot establish convincing depth and support. Fine surface texture, restrained highlights and a contact shadow reinforce the material; no material construction defect is observed." }, "broth": { "score": 0.67, "reason": "Warm brown liquid has subtle oil marks, highlights and edge shading, but its broad surface is rather uniform and opaque, with ingredients appearing mostly perched above it rather than immersed." }, "steam": { "score": 0.25, "reason": "Faint translucent curling wisps rise above the soup without obscuring it, but their contours remain visibly unchanged across the three normal desktop frames, reading as static smoke." }, "composition": { "score": 0.9, "reason": "The single dish is clearly framed against a restrained dark background, with balanced green and gold accents and coherent warm lighting. No distracting text or UI is present; mobile retains the full dish, though with generous empty space." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushrooms, leafy vegetables, corn, scallions and seaweed alongside noodles; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6066666666666667, "toppings": 0.76, "bowl": 0.7366666666666667, "broth": 0.6266666666666667, "steam": 0.5666666666666667, "composition": 0.7866666666666667 }, "reward": 0.685, "sampleStdDev": 0.0243310501211929, "judgments": [ { "noodles": { "score": 0.6, "reason": "The noodles form a fine, dense nest with a convincing golden color and many overlapping strands. The mass is small and looks somewhat flat and textured rather than supple. Individual strands and bends are hard to read." }, "toppings": { "score": 0.72, "reason": "Shiitake with cross-cut scoring, grilled tofu with char marks, crinkled nori, bok choy with veined leaves, corn kernels and scallion rings are all recognizable and varied. The nori is textured and irregular. The tofu blocks are slightly rigid and the corn is a bit clustered and uniform. No plastic-looking forms dominate." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a thick lip with a highlight, a speckled dark-green glaze, a foot ring and a soft contact shadow. The perspective is consistent. The body is fairly flat and the foot looks slightly small and low in contrast." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange-brown with subtle ripple lines and some sheen, and toppings sit in it with reasonable contact. The surface is fairly flat and opaque, with limited depth. Some toppings look slightly pasted on." }, "steam": { "score": 0.5, "reason": "Faint, translucent, wispy plumes rise above the bowl and shift shape slightly between frames, so they do not hide the food. The motion is subtle and the wisps are thin and low in contrast, looking somewhat mechanical and only slightly varying." }, "composition": { "score": 0.78, "reason": "The bowl is centered as the clear focal point against a dark, restrained background with warm lighting and a pleasing palette of orange, green and gold. There is no clutter or text. The bowl is a little small in the frame, with empty space below it." }, "vegan": { "score": 0.96, "reason": "The toppings are tofu, mushrooms, nori, bok choy, corn, scallions and noodles in a plant-looking broth. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.62, "reason": "Fine golden strands form a nest with overlap and a few bends, but they read as a dense, slightly uniform texture with little individual strand definition, and the mass is small relative to the bowl." }, "toppings": { "score": 0.78, "reason": "Shiitake with cross-cuts, grilled tofu with char marks, textured nori, veined bok choy, corn kernels and scallion rings are all readable and varied. Some pieces look slightly pasted on, and the nori stands upright like cards." }, "bowl": { "score": 0.74, "reason": "Coherent elliptical rim, speckled ceramic wall, foot ring and contact shadow give good depth. The rim looks a bit thin and translucent, and the interior wall is mostly hidden behind the broth." }, "broth": { "score": 0.62, "reason": "Warm orange-brown broth with subtle sheen and ripple lines looks appetizing. It is fairly flat and opaque, and some toppings sit on top rather than sinking into the liquid." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise above the bowl and shift shape across frames, without hiding the food. They are thin and low in contrast, and the motion is subtle." }, "composition": { "score": 0.78, "reason": "The bowl is centered on a dark vignette, with a warm palette and good contrast. The food is clearly the focus, with no clutter or text, though the lower half of the frame is empty." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, vegetables, seaweed, corn and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.6, "reason": "The noodle nest has fine, wavy, overlapping strands with a golden color and plausible volume. It is small and dense, somewhat texture-like, and the strands are hard to read individually." }, "toppings": { "score": 0.78, "reason": "The toppings are varied and recognizable: nori sheets with a crinkled edge, shiitake with cross-cut scoring, grilled tofu with char marks, bok choy with veins, corn kernels and scallion rings. Some pieces look a bit clip-art-like and the corn cluster is slightly flat, but there are no plastic blobs." }, "bowl": { "score": 0.75, "reason": "The bowl has a coherent elliptical rim, a textured dark ceramic body, a foot ring and a ground shadow. The perspective is consistent and the rim highlight is convincing. The wall is slightly thin at the rim and the contact shadow is soft." }, "broth": { "score": 0.66, "reason": "The broth is a warm orange-brown with subtle sheen lines and droplets. It reads as liquid, but it is fairly flat and uniform, and the toppings sit on it with limited immersion or ripples." }, "steam": { "score": 0.6, "reason": "The steam is faint, translucent and wispy, and it shifts shape between frames. It is subtle and low in contrast, and its motion is gentle and limited." }, "composition": { "score": 0.8, "reason": "The bowl is centered on a dark vignette with warm, appetizing colors and a clear focal point. There is no clutter and the ingredient layout is balanced. The empty space around the bowl is slightly large." }, "vegan": { "score": 0.97, "reason": "The bowl shows only tofu, mushrooms, nori, greens, corn and noodles, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.8, "toppings": 0.8666666666666667, "bowl": 0.8666666666666667, "broth": 0.8333333333333334, "steam": 0.7666666666666667, "composition": 0.8666666666666667 }, "reward": 0.8350000000000001, "sampleStdDev": 0.017500000000000047, "judgments": [ { "noodles": { "score": 0.8, "reason": "Center nest of fine, wavy golden strands with coherent overlaps and believable immersion in the broth; slightly uniform/scribbly density in places but clearly reads as supple noodles, not ribbons or wire." }, "toppings": { "score": 0.85, "reason": "Well-rendered shiitake with cross-scored caps, grilled tofu slabs, bok choy leaves with veins, textured nori sheets, scallion rings, chili threads and a corn cluster; corn kernels are slightly glossy-spherical, otherwise varied and readable." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with clean elliptical rim, inner band, smooth tapering walls and a visible foot; subtle sheen and grounded contact shadow. Minor odd light streak on the left rim." }, "broth": { "score": 0.8, "reason": "Rich appetizing brown surface with soft edge highlights and gentle ripples; ingredients sit in and contact the liquid with plausible occlusion, though the center surface reads slightly flat." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise above the bowl and visibly shift shape across the time-ordered desktop and pointer frames, thinning as they climb without obscuring the food; motion is gentle and supported by the frames." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a dark, warmly glowing background with restrained palette and no clutter or UI; mobile frame keeps the dish readable despite generous empty space." }, "vegan": { "score": 1, "reason": "Visible toppings are mushrooms, tofu, corn, leafy greens, nori, scallion and chili; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.85, "reason": "Warm-toned tangled strands with fine line detail and believable curvature in the broth; slightly massed/flat in the center but clearly supple, not wire-like." }, "toppings": { "score": 0.9, "reason": "Cross-cut shiitake caps with gill texture, grilled tofu slabs, corn kernels, bok-choy-like greens, textured nori sheets, chili threads and scallion rings \u2014 all organic, well-scaled and readable." }, "bowl": { "score": 0.9, "reason": "Coherent dark ceramic bowl with thick rim, footed base, subtle speckled material, consistent perspective and grounding shadow; only a faint highlight seam on the rim as a minor nit." }, "broth": { "score": 0.85, "reason": "Rich tonkotsu-brown surface with gentle sheen and darker edge occlusion; ingredients sit in the liquid, though broth reads slightly flat/uniform in tone." }, "steam": { "score": 0.7, "reason": "Translucent wisps rise and fade without hiding food; shapes shift subtly between frames indicating motion, though the plume is somewhat sparse and only mildly dynamic." }, "composition": { "score": 0.9, "reason": "Balanced single-bowl focal composition on a dark atmospheric ground, coherent warm lighting, no clutter or UI; mobile crop keeps the dish centered and readable." }, "vegan": { "score": 1, "reason": "Tofu, mushrooms, corn, greens, nori, scallions \u2014 all visibly plant-based with no animal ingredients." } }, { "vegan": { "score": 1, "reason": "Toppings are grilled tofu cubes, shiitake mushrooms, corn, bok choy-style greens, nori, scallion rings and chili threads in a miso-toned broth; no animal-derived ingredients are visible." }, "toppings": { "score": 0.85, "reason": "Strong variety: scored shiitake caps with cross cuts, textured nori sheets with jagged edges, leafy greens with pale stalks and veins, grilled tofu cubes with char lines, corn kernel cluster, scallion rings and thin chili threads. Corn reads slightly beady/glossy and tofu cubes are a bit geometric, but all ingredients are recognizable and organically detailed." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the soup, curl and fade without obscuring the food. Comparing desktop-1/2/3 and pointer frames, the tendril shapes shift subtly between captures, indicating live animation rather than a static plume; motion is gentle and could be more pronounced." }, "noodles": { "score": 0.75, "reason": "Golden strands form a central nest with wavy S-curves, layered overlap and believable broth placement. Strand thickness is plausible and strands read as continuous, though the fine-wave texture is somewhat uniform and slightly repetitive in places." }, "composition": { "score": 0.85, "reason": "Balanced centered bowl on a dark warm-vignette background with a soft contact shadow; steam adds atmosphere and nothing competes with the dish. Colors are harmonious; mobile framing also crops cleanly around the bowl." }, "broth": { "score": 0.85, "reason": "Rich miso-orange surface with tonal gradients, subtle oil swirls near the rim and tiny specks; toppings sit in the liquid with contact shadows and occlusion rather than floating above it. Restrained sheen keeps it appetizing." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with a light glazed rim band, matching elliptical rim and body perspective, visible footed base resting in a shadow, and a soft left-side highlight. Construction is solid; the rim highlight is slightly graphic but not broken." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9888888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 416.70000000000437, "frameIntervalP95Ms": 416.70000000000437, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "568508364829f813850959d613f7a1cefa308ef6f9f46f3c8884de5636fd38f7", "gradedAt": "2026-09-30T22:27:33.741612+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6966666666666667, "toppings": 0.8166666666666667, "bowl": 0.8622222222222222, "broth": 0.66, "steam": 0.5444444444444444, "composition": 0.8655555555555555 }, "reward": 0.7501666666666666, "runId": "openai/gpt-6-astra/ultra", "rewardStdDev": 0.08871129829082904, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.8, "bowl": 0.9133333333333333, "broth": 0.58, "steam": 0.25, "composition": 0.9 }, "reward": 0.6681666666666667, "sampleStdDev": 0.005484827557301418, "judgments": [ { "noodles": { "score": 0.5, "reason": "Numerous continuous golden strands have clear thickness and shading, but densely parallel waves and flat looping bundles dominate, reading more as an ordered line pattern than supple piled noodles." }, "toppings": { "score": 0.79, "reason": "Veined leafy greens, scored mushroom caps with visible gills, speckled grilled tofu, corn and scallion rings provide readable variety. Nori is unusually rigid and card-like, and the mushrooms have somewhat repetitive shapes." }, "bowl": { "score": 0.91, "reason": "A coherent elliptical rim, shaded inner wall, rounded ceramic body and attached foot establish convincing depth and support. Fine horizontal surface texture and restrained rim highlights strengthen the material; no material construction defect is observed." }, "broth": { "score": 0.58, "reason": "Warm brown liquid is readable around the perimeter with small surface flecks, but has little liquid sheen or depth. Most ingredients appear layered on top rather than visibly immersed." }, "steam": { "score": 0.25, "reason": "Soft translucent plumes emerge above the soup without obscuring it, but their shapes appear unchanged across the three normal desktop frames, providing no visible rise, spread or fade." }, "composition": { "score": 0.9, "reason": "The single dish remains a clear focal point against a restrained dark background, with balanced green and golden accents and coherent soft lighting. Desktop and mobile framing preserve the complete bowl without clutter; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, corn, scallions and seaweed, with no apparent animal-derived ingredient." } }, { "noodles": { "score": 0.5, "reason": "Distinct golden strands have continuous curves and clear shading, but tightly repeated parallel waves dominate, making the surface look like a flat maze rather than a supple, overlapping noodle heap." }, "toppings": { "score": 0.82, "reason": "Veined leafy greens, scored mushroom caps with visible gills, porous browned tofu, corn, and scallion rings provide readable variety and detail. The nori is rather rigid and card-like, and the mushrooms are somewhat uniformly shaped." }, "bowl": { "score": 0.92, "reason": "The elliptical rim, thick ceramic lip, shaded curved walls, subtle surface striations, and attached foot form a convincing vessel with coherent depth and support. No material construction defect is observed." }, "broth": { "score": 0.58, "reason": "Warm brown liquid is visible around the ingredients with small surface flecks, but has limited sheen and depth. Most noodles and toppings appear arranged above it with little visible immersion or liquid contact." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps emerge behind the soup without obscuring it, but their shapes appear essentially unchanged across the normal time-ordered desktop frames; rising and fading motion is not demonstrated." }, "composition": { "score": 0.9, "reason": "A single clearly framed dish remains the focus against a restrained dark background. Warm food colors contrast well with the green ceramic, soft lighting establishes depth, and both desktop and mobile views are uncluttered. No material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands are clearly readable, but densely parallel, repeated bends form a flat maze-like mat rather than supple overlapping noodle volume." }, "toppings": { "score": 0.79, "reason": "Veined leafy greens, scored mushroom caps with visible gills, and speckled grilled tofu provide strong ingredient detail and variety. Nori looks stiff and card-like, and the corn is overly bead-like." }, "bowl": { "score": 0.91, "reason": "Coherent elliptical rim, shaded inner wall, rounded ceramic body, subtle horizontal texture, and attached foot convincingly establish depth and support. No material construction defect is observed." }, "broth": { "score": 0.58, "reason": "Warm brown liquid surrounds the ingredients with small surface flecks, but has little liquid sheen or depth; the noodle mat and toppings appear mostly perched above it rather than immersed." }, "steam": { "score": 0.25, "reason": "Soft translucent plumes emerge behind the soup without obscuring food, but their shapes appear unchanged across the three normal desktop frames, providing no visible rising or fading motion." }, "composition": { "score": 0.9, "reason": "The single dish remains prominent against a restrained dark background, with balanced green and golden accents, coherent shading, and no distracting text or UI. Mobile framing also preserves the whole bowl; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, seaweed, and noodles plausibly constitute vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.7733333333333333, "toppings": 0.8, "bowl": 0.8066666666666666, "broth": 0.6066666666666667, "steam": 0.5333333333333333, "composition": 0.8133333333333334 }, "reward": 0.738, "sampleStdDev": 0.013702189606044723, "judgments": [ { "noodles": { "score": 0.8, "reason": "Many fine, supple wavy strands with coherent overlap and a golden color. The central wavy mass looks convincing and the bundles at the edges read as noodles. The pattern is somewhat repetitive and the strands are thin, so they read a bit like vector lines." }, "toppings": { "score": 0.82, "reason": "Bok choy with veined leaves, shiitake with cross cuts and gills, grilled tofu with char marks, textured nori with ragged edges, corn, scallion rings and chili threads. Variety and readable detail are strong. The nori looks like stiff, flat cards and the tofu is slightly blocky." }, "bowl": { "score": 0.82, "reason": "The elliptical rim is consistent with the deep green glazed body, has a light rim highlight, subtle horizontal throwing lines and a foot ring, and sits on a soft shadow. Perspective is coherent. The body is slightly flat in shading, but no material defect is observed." }, "broth": { "score": 0.6, "reason": "Brown broth is visible only at the edges and sides, with little sheen or depth. Noodles cover most of the surface, so the liquid is hardly readable. Immersion of the toppings is plausible but not strongly shown." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and change shape slightly across frames. They are very subtle and low in contrast, somewhat uniform, and only modest motion is visible." }, "composition": { "score": 0.82, "reason": "A centered bowl on a dark, restrained background with a clear focal dish. The warm food against the green bowl is attractive and the lighting is coherent. There is no clutter or text. The bowl sits slightly high, leaving some empty space below." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, bok choy, nori, corn, scallion and noodles are visible. No animal products are apparent." } }, { "noodles": { "score": 0.78, "reason": "Fine golden strands with wavy bends, overlapping layers and a believable mound in the center. Some areas look patterned and repetitive, and the strands sit like a flat texture in places, but no rigid maze or broken lines." }, "toppings": { "score": 0.8, "reason": "Bok choy with veined leaves and white stems, shiitake with cross-cut caps and gills, grilled tofu with char marks, nori with jagged edges and flecks, corn kernels and scallion rings. All are varied and readable. Nori is a bit card-like and the tofu is slightly blocky." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with thickness, a green glazed body with subtle throwing lines, a foot ring and a soft contact shadow. Perspective is consistent. The glaze highlights are modest and the rim is a little flat." }, "broth": { "score": 0.62, "reason": "Amber broth is visible only at the edges and in the right-hand pocket. It has decent depth and the toppings sit in it, but most of the surface is covered by noodles, so the sheen and texture are limited." }, "steam": { "score": 0.5, "reason": "Faint translucent wisps rise above the bowl and change shape slightly across frames without hiding the food. They are very faint and low in contrast, with limited spread." }, "composition": { "score": 0.8, "reason": "Centered bowl, dark atmospheric background, harmonious earthy palette, and the food is the clear focus. No UI or text clutter. The bowl sits slightly low and the top space is empty." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn, scallion and noodles. No visible animal products. The red threads are likely chili." } }, { "noodles": { "score": 0.74, "reason": "Fine, supple wavy strands with coherent overlap and a convincing flowing mass in the center. They are slightly uniform and tightly packed, and edges near the rim look a bit pattern-like." }, "toppings": { "score": 0.78, "reason": "Bok choy with veined leaves, shiitake with cross-cut scoring and gills, grilled tofu with char marks, corn, scallions and jagged textured nori are all recognizable and varied. The nori looks a bit like flat cards, and the tofu is somewhat stylized." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a thick lip, curved walls with subtle glaze lines, a foot base, and a soft shadow. The perspective is consistent. The bowl is slightly dark and plain, with minimal highlights." }, "broth": { "score": 0.6, "reason": "Warm amber broth is visible at the edges and around the noodles, and gives depth. Little surface sheen is shown and most of it is covered, so immersion cues are limited." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and change subtly between frames, without hiding the food. They are very faint and low in contrast, with limited spread." }, "composition": { "score": 0.82, "reason": "The bowl is centered as a clear focal point on a dark, restrained background, with harmonious green and amber colors. No distracting UI or text. The toppings are balanced, though somewhat crowded to the sides." }, "vegan": { "score": 0.96, "reason": "Tofu, mushrooms, bok choy, corn, scallion, nori and noodles make up the visible ingredients. No egg, meat or fish is visible." } } ], "diagnostics": { "vegan": 0.96 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.8166666666666667, "toppings": 0.85, "bowl": 0.8666666666666667, "broth": 0.7933333333333333, "steam": 0.85, "composition": 0.8833333333333333 }, "reward": 0.8443333333333334, "sampleStdDev": 0.006027713773341713, "judgments": [ { "noodles": { "score": 0.8, "reason": "Golden continuous strands with plausible bends and coherent braided mass; slight repetitive wave patterning in places makes parts read a touch flat, but no rigid ribbons or broken strands." }, "toppings": { "score": 0.85, "reason": "Well-detailed bok choy leaves with white stems, scored grilled tofu cubes, shiitake caps with visible gills, nori sheets with speckle, scallion rings and chili threads; corn kernels are slightly bead-like but otherwise convincing variety and scale." }, "bowl": { "score": 0.9, "reason": "Coherent dark ceramic bowl with consistent elliptical rim, cream inner lip, tapering walls, speckled glaze, rim highlight, visible foot and grounded shadow; no distortion or detached parts." }, "broth": { "score": 0.8, "reason": "Amber liquid surface reads as broth with restrained sheen; mushrooms, greens and tofu show believable contact and occlusion, though the surface is somewhat matte and the noodle bed sits fairly flat in it." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps rise above the bowl and change shape across desktop-1 through desktop-3 and pointer frames, indicating continuous animation; wisps are irregular, fade naturally and never obscure the food." }, "composition": { "score": 0.9, "reason": "Centered focal bowl against a dark atmospheric vignette with balanced proportions, warm food tones popping off the background, and no clutter, text or competing elements; scales cleanly on mobile." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are bok choy, tofu, shiitake mushrooms, corn, nori, scallions and chili \u2014 all plant-derived with no animal toppings." } }, { "noodles": { "score": 0.8, "reason": "Continuous golden wavy strands with layered shading and believable immersion in broth; strands read as supple noodles rather than rigid ribbons, though the center swirl pattern is somewhat repetitive and stylized." }, "toppings": { "score": 0.85, "reason": "Strong variety and detail: shiitake with cross-scored caps and visible gills, bok choy with veined leaves, grilled tofu with sesame and grill marks, speckled nori sheets, corn kernels, and scallion rings with chili threads. Minor flaws: scallion rings are uniform circles and the corn cluster is slightly grape-like." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with a glazed light rim, elliptical opening, curved walls, visible foot, and grounding shadow with highlight streaks; perspective is consistent. Slightly shallow/wide proportions but reads as an intentional ramen bowl shape." }, "broth": { "score": 0.78, "reason": "Amber broth is visible around the noodles with darker edges suggesting depth, and toppings sit at the surface rather than floating pasted above. Broth reads slightly flat in the noodle-dominated center, but integration is believable." }, "steam": { "score": 0.85, "reason": "Translucent faint wisps rise from the soup and visibly change shape across the time-ordered desktop frames (thin plume at ~5s, broader plume at ~8.5s, different plumes in pointer frames), indicating continuous animated steam that never obscures the food." }, "composition": { "score": 0.9, "reason": "Centered focal bowl on a dark atmospheric ground, warm food tones against near-black background, clean presentation with no text, UI, or competing elements; mobile crop keeps the dish readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are bok choy, mushrooms, tofu, corn, scallions, chili threads, and nori \u2014 all plant-based with no meat, egg, fish, or dairy present." } }, { "noodles": { "score": 0.85, "reason": "Dense golden strands with continuous flowing S-curves and a coherent central swirl; consistent thickness and believable layering. Minor flatness in the painted swirl pattern and sparse broth gaps, but no rigid ribbons or scribble maze." }, "toppings": { "score": 0.85, "reason": "Recognizable bok choy with veined leaves, grilled tofu cubes with speckling, shiitake caps with cross scores and visible gills, textured nori sheets, scallion rings, chili threads and corn. Corn reads slightly bead-like and nori a bit rigid, but variety, scale and detail are strong." }, "bowl": { "score": 0.85, "reason": "Coherent dark-green ceramic bowl with a thick elliptical rim, consistent cavity perspective, speckled glazed body, rim highlights, visible foot and grounded contact shadow. Very clean; only slightly idealized symmetry." }, "broth": { "score": 0.8, "reason": "Appetizing golden-brown liquid ring with darker edge shading and restrained sheen; toppings sit with contact shadows and look immersed. Broth is largely occluded by the noodle blanket, limiting visible depth." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps that change shape across desktop-1/2/3 and pointer frames, rising and dispersing above the bowl without obscuring food. Curls are somewhat stylized but clearly animated and irregular; static in reduced frames as expected." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark atmospheric backdrop with subtle warm glow; no text, UI or clutter; appetizing color palette. Mobile view leaves generous empty space above and below the dish." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, bok choy, nori, corn, scallions and chili threads in plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9866666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 149.95000000000073, "frameIntervalP95Ms": 200.09999999999854, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "3f1efffc5a7f1c08a1eff33e141740bf5a3bf26e240ff0d57372396305747144", "gradedAt": "2026-09-30T22:27:34.301119+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6, "toppings": 0.5833333333333334, "bowl": 0.7533333333333334, "broth": 0.5766666666666667, "steam": 0.5633333333333334, "composition": 0.7766666666666667 }, "reward": 0.6363333333333334, "runId": "xai/grok-4.7/medium", "rewardStdDev": 0.05199065086895702, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.5666666666666667, "bowl": 0.78, "broth": 0.4633333333333333, "steam": 0.45666666666666667, "composition": 0.81 }, "reward": 0.6135, "sampleStdDev": 0.005894913061275766, "judgments": [ { "noodles": { "score": 0.61, "reason": "Continuous golden strands have varied curves and readable crossings, but several become broad, angular ribbons with abrupt ends and little convincing submerged volume." }, "toppings": { "score": 0.58, "reason": "Tofu, mushroom caps, corn, greens and fine white mushroom stems provide recognizable variety. Texture is sparse, leaves are glossy simplified shapes, and the nori resembles a thick smooth upright strip." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, dark cavity, curved walls and small foot form coherent bowl geometry with restrained highlights. Broad, nearly uniform wall shading and a weak ground contact limit material depth." }, "broth": { "score": 0.47, "reason": "Warm orange color clearly separates the soup from the bowl, but concentric elliptical bands make it look like a flat solid disc. Ingredients show little meniscus, immersion or liquid sheen." }, "steam": { "score": 0.46, "reason": "Normal frames show changing positions and opacity of translucent vapor above the soup without obscuring food. The vapor consists mainly of disconnected blurred oval puffs rather than irregular connected wisps; continuous flow is not established." }, "composition": { "score": 0.81, "reason": "A centered solitary dish, restrained warm highlights and an uncluttered dark background keep the ramen clearly focal. Mobile framing preserves the dish but leaves excessive empty space and makes food details small." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, corn, greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present. No material compliance defect is observed." } }, { "noodles": { "score": 0.61, "reason": "Readable golden strands have curved bends and some overlap, but several are broad, angular ribbons with abrupt ends and little convincing submerged volume." }, "toppings": { "score": 0.56, "reason": "Tofu, mushroom caps, corn, greens and scallion rings provide recognizable variety. Texture is sparse, the leaves are simplified, and the nori reads as a thick smooth upright strip rather than irregular seaweed." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, dark cavity, rounded walls and small foot form a coherent vessel with restrained highlights. The oversized rim and weak ground contact make it slightly heavy and floating-looking." }, "broth": { "score": 0.46, "reason": "Warm orange coloring suggests soup, but concentric elliptical bands create a flat disc-like surface. Minimal reflections and immersion cues leave ingredients looking laid over it." }, "steam": { "score": 0.44, "reason": "Translucent patches change position and density across normal frames without obscuring the dish. They resemble disconnected blurred oval puffs more than irregular, connected rising wisps." }, "composition": { "score": 0.82, "reason": "The single centered dish, warm food colors, restrained dark backdrop and uncluttered presentation establish a clear focal point. Mobile framing leaves the bowl relatively small within extensive empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, corn, greens and seaweed with noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.58, "reason": "Distinct golden strands have varied curves and readable overlaps, but several are broad, angular ribbons with abrupt ends and little convincing submerged volume." }, "toppings": { "score": 0.56, "reason": "Tofu, mushroom caps, corn, scallions and a small leafy sprig provide recognizable variety. Texture is sparse, mushrooms are simplified ovals, and the tall nori resembles a smooth rigid strip rather than irregular seaweed." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, dark inner cavity, curved walls and small foot form a coherent vessel with restrained highlights. The oversized rim and minimally modeled exterior make the ceramic feel somewhat schematic." }, "broth": { "score": 0.46, "reason": "Warm orange color clearly distinguishes the soup, but concentric outlines and a nearly uniform surface make it resemble a flat disc. Ingredients show limited immersion or liquid contact." }, "steam": { "score": 0.47, "reason": "Translucent steam changes position and density across normal frames without obscuring the food. It consists mainly of separated blurry oval puffs rather than connected irregular wisps; continuous flow is not established by these captures." }, "composition": { "score": 0.8, "reason": "A centered, unobstructed dish and warm food colors against a restrained dark background create a clear focal point. The mobile dish is relatively small within extensive empty space, and desktop proportions emphasize the vessel more than the food." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredients are present. No material compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.5666666666666667, "bowl": 0.68, "broth": 0.5833333333333334, "steam": 0.5, "composition": 0.6866666666666666 }, "reward": 0.5996666666666667, "sampleStdDev": 0.00880814017448252, "judgments": [ { "noodles": { "score": 0.6, "reason": "Golden strands curve and overlap with a readable tangle and some enoki. They are thin and small, with a fairly flat, somewhat wiry look. The mass is a bit clustered and the volume is limited." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake mushrooms, enoki, corn, scallion rings and a nori strip are recognizable and varied. They are tiny and simplified, and the nori reads as a dark rigid band standing up at the rim. The bok choy sprout is small and neat." }, "bowl": { "score": 0.68, "reason": "The dark ceramic bowl has a coherent rim, a smooth curved body, highlights and a foot. The rim and liquid ellipses are consistent. The base looks slightly floating over the glow, and the body is fairly plain." }, "broth": { "score": 0.55, "reason": "The red-orange broth has a warm gradient and a concentric ring that reads slightly like a target disc. Ingredients sit on it plausibly, but the surface lacks texture and depth, and the liquid sits quite low." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps rise and drift between frames, but they look like blurry blobs. They are faint and somewhat disconnected from the soup, and they are not very wispy." }, "composition": { "score": 0.68, "reason": "The bowl is centered as the focal point in a dark, atmospheric scene with a warm glow and no clutter. The food is small within the bowl and the lower half of the frame is empty." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, corn, greens, nori and noodles are all plant-based, and no animal ingredients are visible." } }, { "noodles": { "score": 0.6, "reason": "Noodles are supple, overlapping, curved strands in a tangled mass with good color. They are small and somewhat thin-looking, and a few read as flat looping ribbons. Volume is modest." }, "toppings": { "score": 0.6, "reason": "Tofu cubes, shiitake mushrooms, enoki, corn, bok choy sprout, scallion rings and a nori strip are all recognizable. They are very small and stylized, and the nori is a stiff dark band that looks like a rigid card." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, inner cavity, rounded body with gradient shading, rim highlights and a foot. The dark foot/shadow looks slightly detached from the body and the bowl appears to float over the glow. It is otherwise consistent in perspective." }, "broth": { "score": 0.6, "reason": "The broth is an appetizing orange-red with a subtle inner ring and highlight. It reads somewhat like a flat disc with a concentric ring, and the ingredients sit on it with little immersion or occlusion." }, "steam": { "score": 0.5, "reason": "Soft translucent blobs rise and shift between frames, so the motion is visible. They look like separate faint puffs rather than continuous wisps and are very dim." }, "composition": { "score": 0.68, "reason": "The bowl is a centered focal point in a dark atmospheric scene with a warm glow and no clutter. The food is small relative to the bowl and the image is very dark, with large empty space below." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, vegetables, seaweed and noodles are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.6, "reason": "Noodles are supple, curved, overlapping strands with a golden color and some volume. They are thin and a bit uniform, and the mass is small and slightly scribbly, so they read less convincingly as a tangle." }, "toppings": { "score": 0.55, "reason": "There are recognizable shiitake mushrooms, tofu cubes, enoki, a nori strip, bok choy leaves, corn and scallion rings. They are tiny and simplified, and the nori reads as a dark rigid bar. Variety is good but detail is limited." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, inner cavity, rounded body with a gradient and a rim highlight. The base foot is small and looks slightly detached and floating above the glow, and the rim and body proportions feel slightly off." }, "broth": { "score": 0.6, "reason": "The broth is an appetizing orange-red with a ring highlight and depth. It looks somewhat flat and disc-like, and the ingredients sit on it without much immersion or ripple." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps rise and drift between frames. They are blobby, faint and somewhat disconnected, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point in a dark atmospheric scene with a warm glow. The food is small relative to the bowl, and there is a lot of empty space in the frame." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, vegetables, corn, nori and noodles, with no visible animal products." } } ], "diagnostics": { "vegan": 0.93 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.6166666666666667, "bowl": 0.8, "broth": 0.6833333333333333, "steam": 0.7333333333333333, "composition": 0.8333333333333334 }, "reward": 0.6958333333333334, "sampleStdDev": 0.005773502691896263, "judgments": [ { "bowl": { "reason": "Coherent elliptical rim, inner lip, tapering body and small foot with grounded contact shadow; consistent perspective and ceramic highlights, only minor simplification at the foot.", "score": 0.8 }, "broth": { "reason": "Orange-red surface with edge darkening, subtle sheen and scattered specks reads as liquid, but it is fairly flat and noodles appear to lie on top rather than submerge convincingly.", "score": 0.65 }, "composition": { "reason": "Well-centered focal bowl on a restrained dark ground with a soft warm under-glow; balanced margins, coherent lighting, no clutter, text or competing elements.", "score": 0.85 }, "noodles": { "reason": "Yellow strands are continuous with varied bends and some layered shading, but the tangle relies on repeated curly loops that mostly sit atop the broth, reading somewhat flat and scribbly.", "score": 0.6 }, "steam": { "reason": "Frames show translucent gray wisps that shift position, shape and opacity over time, rising above the bowl without obscuring the food; blobs are a bit discrete and puffy rather than continuous.", "score": 0.75 }, "toppings": { "reason": "Good variety: corn cluster, tofu cubes, scallion rings, brown mushroom shapes, enoki-like white strands and a leafy sprout; but forms are simple, the vertical dark strip reads as a rigid card and the mushrooms are near-featureless ellipses.", "score": 0.6 }, "vegan": { "reason": "Visible ingredients are tofu, corn, scallion, mushrooms, sprouts, seaweed and noodles in plant broth; no animal-derived toppings apparent.", "score": 1 } }, { "noodles": { "score": 0.6, "reason": "Yellow strands are continuous with plausible bends and consistent thickness, readable as a noodle tangle draping over tofu. However the mass is largely flat loops resting on the broth surface with uniform thin strokes, little volume, and scribbly overlap; immersion in the broth is not conveyed." }, "toppings": { "score": 0.6, "reason": "Good variety: corn clusters, tofu cubes with shading, brown mushroom slices, green sprout leaves, scallion rings, and a dark nori sheet. Shapes are recognizable but simplified; the nori is a thick rigid card-like slab rather than thin seaweed, and the mushroom slices are plain glossy ovals." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: rim ellipse with highlight arc, visible wall thickness and cavity depth, smooth tapering body, small foot with glaze highlight, and a grounded contact shadow. Minor shortcoming is the slightly detached-looking highlight stroke on the rim and somewhat simplified gradient body." }, "broth": { "score": 0.7, "reason": "Warm reddish broth with subtle sheen, darker edge ring reading as a meniscus, and toppings casting small contact shadows so they sit in the liquid. It reads slightly flat/solid and the surface texture is minimal, but integration of ingredients is believable." }, "steam": { "score": 0.75, "reason": "Translucent blurred wisps rise above the bowl and clearly change shape and position across the time-ordered desktop frames (faint in desktop-1, fuller plumes in desktop-2, sparser in desktop-3), so motion is supported. Wisps are somewhat blobby and occasionally read as three mechanical vertical plumes; food remains visible." }, "composition": { "score": 0.8, "reason": "Balanced dark scene with the bowl as clear focal point, warm under-glow and grounding shadow, pleasant contrast between orange broth, yellow noodles and dark ceramic, and pointer parallax that stays tasteful. No text, UI or clutter. Slightly generous empty dark space above the bowl." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are corn, tofu cubes, mushrooms, sprout greens, scallion rings and nori in plant broth; no eggs, meat, fish or dairy appear." } }, { "noodles": { "score": 0.6, "reason": "Pale yellow strands curve coherently across the broth with varied bends and some volume, but many are thin and wiry with repeated hook/loop curls, and the nest lies quite flat on the surface rather than showing volume in the liquid." }, "toppings": { "score": 0.65, "reason": "Good variety of readable vegan items: brown shiitake-like mushrooms, two tofu cubes, corn kernels, scallion rings, a two-leaf sprout, and a nori sheet. Defects: the nori reads as a thick rigid dark card standing in the soup, and the mushrooms are simple glossy ellipses." }, "bowl": { "score": 0.8, "reason": "Well-constructed stylized bowl: coherent elliptical rim with visible rim thickness, smooth tapered walls with soft gradients, foot/base ellipse, rim highlight, and believable ground contact with a warm under-glow. Minor shortcoming: the foot is a simple flat ellipse and interior shading is basic." }, "broth": { "score": 0.7, "reason": "Orange-red surface with tonal gradient and a lighter meniscus ring near the cavity edge; toppings rest on the surface with subtle shadows suggesting immersion. Slightly flat/solid-looking in the center and noodles sit atop rather than dipping in." }, "steam": { "score": 0.7, "reason": "Translucent irregular gray wisps that clearly change across the normal frames (dense low plume at 1s, dispersing higher at 2.6s, fading by 4.2s), indicating continuous rise-and-fade motion; steam is faint/static in reduced frames as expected. Slightly blobby in places with discrete circular puffs near the top." }, "composition": { "score": 0.85, "reason": "Centered dish with clear focal read, restrained dark atmosphere, warm glow grounding the bowl, steam framing above, and no clutter, text, or competing elements; attractive warm-orange-on-black palette. Mobile crop keeps the bowl well placed." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, corn, scallions, sprouts, nori and noodles in a reddish plant broth; no animal-derived toppings are apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9766666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 49.94999999999982, "frameIntervalP95Ms": 66.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "da6f69112dc4be96f0e4904d58cb6c13987cfa83fd6cac01d1b5c8c9b920e15f", "gradedAt": "2026-09-30T22:51:21.046764+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5822222222222222, "toppings": 0.6644444444444444, "bowl": 0.6933333333333334, "broth": 0.6811111111111111, "steam": 0.73, "composition": 0.8099999999999999 }, "reward": 0.6798333333333333, "runId": "anthropic/fable-5.1/high", "rewardStdDev": 0.03442907363132375, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6233333333333333, "bowl": 0.7633333333333333, "broth": 0.6466666666666667, "steam": 0.7466666666666667, "composition": 0.84 }, "reward": 0.6691666666666667, "sampleStdDev": 0.0035472994422987916, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous pale strands have readable highlights and overlaps, but mostly form flat nested oval loops with little supple variation or submerged volume." }, "toppings": { "score": 0.62, "reason": "Scallion rings, scored mushroom caps, carrot flowers, tofu and veined leaves provide readable variety. Mushrooms are disk-like, greens are stiff and repetitive, and nori resembles a rigid rounded card." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner wall, shaded speckled exterior and attached foot establish a coherent vessel. The broad, nearly flat lower wall and abrupt foot transition make its profile somewhat boxy." }, "broth": { "score": 0.64, "reason": "Warm amber shading, small oil bubbles and restrained surface lines suggest soup. Most ingredients appear layered atop the surface, with limited convincing immersion or liquid contact." }, "steam": { "score": 0.74, "reason": "Normal frames show changing translucent vapor rising above the soup without obscuring the dish. The vapor is mostly broad blurred puffs rather than finely articulated irregular wisps." }, "composition": { "score": 0.84, "reason": "The warm food contrasts clearly against the dark restrained setting, and diagonal chopsticks balance the centered bowl without text or clutter. Generous upper empty space makes the dish slightly low and small, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, scallions, carrots, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable overlaps and highlights, but broad, repeated flat loops dominate and look arranged rather than supple or volumetric." }, "toppings": { "score": 0.62, "reason": "Scallion rings, scored mushroom caps, carrot flowers, tofu and veined greens provide readable variety. Mushrooms are disk-like, greens unusually rigid, and the smooth rectangular nori lacks organic texture." }, "bowl": { "score": 0.77, "reason": "The elliptical rim, inner cavity, shaded speckled walls and attached foot establish a coherent vessel. The nearly flat lower wall silhouette and broad pedestal make its profile somewhat boxy." }, "broth": { "score": 0.65, "reason": "Warm amber shading, small oil bubbles and restrained surface lines suggest soup. Depth and immersion remain shallow, with several ingredients appearing laid on the surface rather than partly submerged." }, "steam": { "score": 0.76, "reason": "Normal frames show translucent steam changing position and spreading above the soup without obscuring it. Wisps are rather broad and softly blob-like instead of finely irregular tendrils." }, "composition": { "score": 0.85, "reason": "Warm food colors contrast effectively with the dark bowl and uncluttered background; diagonal chopsticks balance the upright toppings, and mobile framing preserves the dish. Excess upper empty space slightly weakens the framing." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous pale strands have readable highlights and overlaps, but predominantly flat nested loops and angular bends make the pile look diagrammatic rather than supple and voluminous." }, "toppings": { "score": 0.63, "reason": "Scallion rings, scored mushroom caps, carrot flowers, tofu and veined greens provide readable variety. Mushrooms are disk-like, greens unusually rigid and upright, and nori resembles a regular rounded card." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner wall, shaded speckled exterior and attached foot establish a coherent vessel. The broad, nearly flat lower body and minimally shaded foot make its ceramic volume somewhat artificial." }, "broth": { "score": 0.65, "reason": "Warm amber color, scattered oil bubbles and gentle surface lines suggest soup. Limited immersion shading and uniformly outlined ingredients leave the surface fairly flat and some toppings pasted on." }, "steam": { "score": 0.74, "reason": "Normal frames show translucent vapor changing shape and position above the soup without obscuring the dish. Broad blurred patches dominate over finer irregular rising wisps, limiting the natural steam impression." }, "composition": { "score": 0.83, "reason": "The isolated dish, warm food colors against a dark tabletop, restrained lighting and diagonal chopsticks create a clear, balanced focal presentation with no text or clutter. Excess upper negative space makes the dish slightly low and small, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible tofu cubes, mushrooms, leafy greens, scallions, carrots, seaweed and noodles plausibly form vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.68, "toppings": 0.6866666666666666, "bowl": 0.5166666666666667, "broth": 0.68, "steam": 0.66, "composition": 0.74 }, "reward": 0.652, "sampleStdDev": 0.013527749258468695, "judgments": [ { "noodles": { "score": 0.68, "reason": "Noodles form a readable swirled nest with soft bends, overlap, and highlights. The strands look somewhat flat and looped, and the mass is small relative to the bowl." }, "toppings": { "score": 0.68, "reason": "Nori, shiitake with cross scoring, tofu cubes, bok choy/leek, scallion rings, carrot flowers and corn are all readable and varied. Some look like clip-art, and the tofu is blocky." }, "bowl": { "score": 0.5, "reason": "The rim and dark speckled glaze read well. The lower body has a wobbly, flared outline that clashes with the rim perspective, and it sits detached over a narrower foot, so the geometry looks off." }, "broth": { "score": 0.68, "reason": "Orange broth has bubbles, oil specks and a warm sheen, and the ingredients sit in it. It is fairly flat in places, with a visible lower band of color below the rim." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps rise and change shape between frames without hiding the food. They are diffuse and blobby, not clearly wispy." }, "composition": { "score": 0.75, "reason": "A centered bowl on a dark, moody background with warm glow keeps the food as the focus, and the chopsticks add a diagonal. The bowl's odd shape and the empty space around it weaken the presentation slightly." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, vegetables, nori and noodles are visible, with no egg or meat." } }, { "noodles": { "score": 0.68, "reason": "Noodles form a readable, cream-colored swirl of supple, overlapping strands with highlights and shading. The mass is small and looks somewhat flat and stacked. It sits at the front of the bowl and reads slightly pasted on." }, "toppings": { "score": 0.68, "reason": "Toppings are varied and recognizable: a textured nori sheet, shiitake with cross scoring, bok choy stalks with leaves, tofu cubes, scallion rings, corn and carrot flowers. Shapes are stylized and small. The nori is a fairly rigid rectangle, and the mushrooms are simple." }, "bowl": { "score": 0.5, "reason": "The dark speckled bowl has a glossy rim and a foot. The body widens into a strange blobby lower skirt that does not match the elliptical rim. The base looks detached and narrower, so the perspective is inconsistent." }, "broth": { "score": 0.68, "reason": "Orange broth has a warm sheen, small oil bubbles and floating flecks, and it is appetizing. Contact with the ingredients is mostly believable. It is somewhat flat in places." }, "steam": { "score": 0.62, "reason": "Soft translucent wisps rise from the bowl and shift shape across frames. They stay subtle and do not hide the food. The plumes are blurry and blobby rather than wispy." }, "composition": { "score": 0.72, "reason": "The bowl is a centered focal point on a dark, moody background with a warm glow and no clutter. The chopsticks add a diagonal line. The bowl is slightly small in the frame and the bowl distortion detracts." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, vegetables, seaweed and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.68, "reason": "The noodle nest has soft, thick strands that curve and overlap with highlights and shadows. The loops are somewhat repetitive and concentric, and the mass is small in the bowl." }, "toppings": { "score": 0.7, "reason": "Shiitake with cross scoring, irregular nori, bok choy stalks with leaves, cubed tofu, scallion rings and carrot flowers are all readable. Some pieces look clip-art flat, and the tofu cubes are plain." }, "bowl": { "score": 0.55, "reason": "The rim is elliptical with highlights and the speckled glaze is convincing. The lower body has a wobbly, flared outline that is wider than the rim and sits oddly on a small foot, so the body and base join looks disjointed." }, "broth": { "score": 0.68, "reason": "The orange broth has bubbles, oil droplets and a reflective sheen. Ingredients sit in it reasonably well, though the surface is fairly flat and slightly opaque." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps rise and shift between frames. They are fairly diffuse and blobby, and they drift above the bowl without a clear link to the surface." }, "composition": { "score": 0.75, "reason": "The bowl is centered and clearly the focus against a dark warm vignette with no clutter. The chopsticks add a diagonal, and the warm palette is attractive. The bowl is small in the frame and the lower body shape is awkward." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, tofu, vegetables, nori and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6833333333333333, "bowl": 0.8, "broth": 0.7166666666666667, "steam": 0.7833333333333333, "composition": 0.85 }, "reward": 0.7183333333333334, "sampleStdDev": 0.023228933107943953, "judgments": [ { "noodles": { "score": 0.55, "reason": "Tangled loop bundle reads as noodles but strands flatten into repeated overlapping loops with indistinct mass at center; plausible thickness and color though." }, "toppings": { "score": 0.7, "reason": "Readable bok choy, mushroom slices with cross-hatching, tofu cubes, scallion rings, and a textured nori sheet; shapes are somewhat flat/sticker-like and variety is modest." }, "bowl": { "score": 0.8, "reason": "Coherent dark speckled bowl with rim highlight, speckle texture, footed base, and believable chopstick contact; slight geometric simplification but no broken joins." }, "broth": { "score": 0.75, "reason": "Rich orange broth with bubbles, oil droplets, and sesame specks; ingredients mostly sit in the liquid, though immersion is shallow and edges look slightly pasted." }, "steam": { "score": 0.8, "reason": "Translucent wisps shift shape and position across the temporal frames, rising and spreading without obscuring the dish; slightly blobby but clearly dynamic." }, "composition": { "score": 0.85, "reason": "Focused single dish on warm dark gradient, balanced proportions, restrained lighting, no clutter; mobile crop keeps the bowl readable." }, "vegan": { "score": 0.5, "reason": "Tofu, mushrooms, bok choy, scallions and nori are vegan-appropriate, but the orange star-shaped slice resembles a narutomaki fish cake, making appearance ambiguous." } }, { "bowl": { "reason": "Black ceramic donburi with a coherent elliptical rim, inner-rim highlight, speckled glaze shading and a plausible footed base with contact shadow; chopsticks rest convincingly on the rim. Only minor flatness in the foot.", "score": 0.85 }, "broth": { "reason": "Warm orange miso-toned surface with oil droplets, sesame, chili flecks and swirl arcs that give some depth, but the swirls read slightly as concentric rings and toppings sit mostly on top with shallow immersion cues.", "score": 0.65 }, "composition": { "reason": "Centered focal dish on a dark atmospheric backdrop with a soft warm glow, balanced proportions and chopsticks adding a diagonal; no clutter, text or competing elements, and mobile framing stays clean.", "score": 0.85 }, "noodles": { "reason": "Cream-colored strands form a readable curly nest with plausible thickness, bends and some shading, but the central tangle relies on repetitive overlapping loops and strand continuity/volume is only partly convincing.", "score": 0.65 }, "steam": { "reason": "Translucent irregular wisps rise above the bowl without hiding food, and the plume shape clearly differs across desktop-1/2/3 and the pointer frames, showing animated drift; some upper patches look slightly blobby/fog-like.", "score": 0.75 }, "toppings": { "reason": "Good variety: cross-cut shiitake-style caps, a textured nori sheet leaning on the rim, tofu cubes, scallion rounds, upright scallion stalks, corn and sesame. Shapes are readable but somewhat iconic \u2014 mushroom caps are flat marked discs and tofu cubes are crisp geometric blocks.", "score": 0.7 }, "vegan": { "reason": "Visible ingredients are tofu, mushrooms, nori, scallions, corn, sesame, chili and noodles in plant-toned broth; no meat, fish, egg or dairy forms appear. Small orange star slices read as vegetable garnish rather than naruto.", "score": 1.0 } }, { "noodles": { "score": 0.5, "reason": "Noodles read as a flat pale tangle of outlined strands with repeated circular loops and hard edges; little volume or coherent overlap with the broth, bordering on a scribbled nest rather than supple continuous strands." }, "toppings": { "score": 0.65, "reason": "Good variety: two bok choy with convincing leaves and stems, tofu cubes, scallion rings, corn, sesame, flower-cut carrot. Weaker elements: mushrooms are flat uniform disks with cross marks, the nori is a rigid green card with grid lines, and tofu blocks are somewhat featureless." }, "bowl": { "score": 0.75, "reason": "Coherent dark speckled bowl with elliptical rim, highlight arc, visible inner wall and a plausible pedestal foot; slight ambiguity where the body meets the foot and a somewhat flat base, but no broken geometry." }, "broth": { "score": 0.75, "reason": "Rich orange broth with restrained sheen, bubbles, chili flecks and sesame; toppings mostly sit in the liquid with believable contact and a consistent meniscus against the inner wall; slight ring-like edge highlight but overall appetizing." }, "steam": { "score": 0.8, "reason": "Normal frames show translucent irregular wisps that change shape, drift left and fade across time without hiding food; reduced frames show fainter static haze as a reduced-motion treatment. Wisps are slightly blobby in places." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric scene: warm glow behind a centered focal bowl, chopsticks resting on the rim, subtle wood table, no text or clutter; generous empty space slightly shrinks the dish's dominance." }, "vegan": { "score": 1.0, "reason": "Visible toppings are bok choy, mushrooms, tofu, scallion, corn, sesame, nori, carrot flowers and plant-looking broth; no meat, egg, fish or dairy forms observed." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9233333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100, "frameIntervalP95Ms": 116.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "2e0cbd313cbeb6186b4dc04c1162e43ecb5d20e34e4d699059be468616124bee", "gradedAt": "2026-09-30T22:52:16.333702+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.40222222222222226, "toppings": 0.5077777777777778, "bowl": 0.7477777777777778, "broth": 0.5666666666666667, "steam": 0.5766666666666667, "composition": 0.7566666666666667 }, "reward": 0.5757777777777778, "runId": "anthropic/fable-5.1/low", "rewardStdDev": 0.04337252501210791, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.3666666666666667, "toppings": 0.49, "bowl": 0.77, "broth": 0.55, "steam": 0.4633333333333333, "composition": 0.75 }, "reward": 0.5493333333333333, "sampleStdDev": 0.009865765724632504, "judgments": [ { "noodles": { "score": 0.38, "reason": "Pale strands are individually readable and overlap, but mostly resemble short rigid sticks with angular bends rather than supple, continuous noodles." }, "toppings": { "score": 0.49, "reason": "Scallion rings, corn and scored mushroom caps provide recognizable variety. Mushrooms are simple oval disks and the rear greens are smooth, featureless shapes without organic leaf detail." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, continuous decorative band and shaded rounded walls form a coherent vessel with convincing depth. Material highlights and base contact remain understated and somewhat flat." }, "broth": { "score": 0.55, "reason": "Warm amber shading and scattered highlights suggest soup, but the surface is largely uniform and ingredients lack convincing immersion or liquid contact." }, "steam": { "score": 0.47, "reason": "Normal frames show translucent clouds growing and spreading above the soup without obscuring the food. They read as broad blurred blobs rather than irregular rising wisps, with no clear fading cycle shown." }, "composition": { "score": 0.76, "reason": "A restrained dark palette keeps the single bowl readable, and diagonal chopsticks add direction without UI clutter. The dish sits low amid considerable empty space, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible mushrooms, scallions, corn, greens and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.38, "reason": "Pale strands are distinct and overlap, but mostly resemble short rigid sticks with angular bends rather than supple, continuous noodles." }, "toppings": { "score": 0.5, "reason": "Scallion rings, corn and cross-scored mushroom caps are recognizable, but mushrooms are smooth oval disks and the large green leaves lack veins, folds or organic detail." }, "bowl": { "score": 0.77, "reason": "Elliptical rim, inset cavity and rounded shaded walls form a coherent bowl with readable thickness. Material highlights are restrained, though the base and supporting contact remain poorly defined." }, "broth": { "score": 0.55, "reason": "Warm amber shading and small surface highlights suggest soup, but the surface is quite solid-looking and ingredients have little visible immersion or liquid contact." }, "steam": { "score": 0.46, "reason": "Normal frames show pale vapor rising and expanding above the soup. It remains mostly a collection of broad blurred clouds rather than irregular wisps, with no clear fading sequence visible." }, "composition": { "score": 0.75, "reason": "The uncluttered dark scene and warm palette keep the bowl readable, with diagonal chopsticks adding direction. Chopsticks obscure much of the food, and the mobile dish is rather small within substantial empty space." }, "vegan": { "score": 1, "reason": "Visible mushrooms, greens, scallions, corn and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.34, "reason": "Pale strands have readable thickness and overlaps, but mostly resemble short rigid sticks with angular bends rather than supple, continuous noodles." }, "toppings": { "score": 0.48, "reason": "Scallion rings, corn and cross-scored mushroom caps are recognizable. However, mushrooms are smooth oval disks and the rear greens are featureless rounded shapes, with little organic texture." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inset cavity and shaded curved body form coherent geometry, with a readable decorative band. Material highlights are subdued and contact with a supporting surface is poorly established." }, "broth": { "score": 0.55, "reason": "Warm amber shading and scattered highlights suggest soup, but the surface is unusually smooth and opaque, with weak immersion cues making ingredients appear laid on top." }, "steam": { "score": 0.46, "reason": "Normal frames show pale clouds growing and shifting above the soup, so steam is not static. It forms broad blurry blobs rather than irregular rising wisps, becoming conspicuously dense above the rear rim." }, "composition": { "score": 0.74, "reason": "A restrained dark brown palette and diagonal chopsticks clearly frame the focal bowl without text or clutter. Chopsticks obscure much of the food, and the mobile dish is rather small within extensive empty space." }, "vegan": { "score": 1, "reason": "Visible mushroom caps, greens, scallions and corn plausibly accompany noodles in plant-based broth; no visibly animal-derived topping is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4066666666666667, "toppings": 0.5, "bowl": 0.6733333333333333, "broth": 0.55, "steam": 0.5833333333333334, "composition": 0.6866666666666666 }, "reward": 0.5521666666666667, "sampleStdDev": 0.016096065771900127, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are short, thin, straight pale sticks, more like bean sprouts or matchsticks than supple ramen strands. They are rigid with little bend or overlap and read as a scatter rather than a coherent noodle mass." }, "toppings": { "score": 0.5, "reason": "Shiitake caps with cross scoring read well, and there are bok choy, tofu, corn and scallion rings. The pieces are small and generic, the scallion rings look like flat ovals, and the chopsticks cover much of the arrangement. The variety is decent but the detail is limited." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a dark glazed wall with a terracotta band and a soft highlight. The perspective is consistent. The base and ground contact are vague and sink into the dark background, and the bowl is fairly plain." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a vignette and some soft glowing patches that give depth. It looks fairly flat and uniform, with little sheen or immersion detail. The ingredients sit on top rather than in it." }, "steam": { "score": 0.55, "reason": "Steam grows over time and looks soft and translucent, and it is visibly animated across the frames. It becomes a heavy, opaque-looking white blob with little wisp structure, and it is a bit foggy." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm colors and no distracting text or UI. The diagonal chopsticks add interest, though they obscure the food. A lot of empty space surrounds the bowl." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are mushrooms, tofu, bok choy, corn, scallion, chili and noodles, with no meat, egg or fish." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, short, stiff pale sticks scattered like bean sprouts or matchsticks. They lack supple continuous strands, overlap or volume, and don't read as ramen noodles." }, "toppings": { "score": 0.5, "reason": "Shiitake with cross scoring, scallion rings, corn kernels, chili bits, tofu cubes and a bok choy leaf are recognizable. They are small and flat, and the tofu and leaf are crude and partly hidden by chopsticks." }, "bowl": { "score": 0.68, "reason": "Dark ceramic bowl with a coherent elliptical rim, red band, curved body and highlight gradient. Perspective is consistent. The base is undefined and the bowl is fairly plain, with little contact shadow." }, "broth": { "score": 0.55, "reason": "Warm orange broth with a vignette and some glowing patches gives depth. It is fairly flat and opaque, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Steam grows and drifts across the frames, translucent and soft. It becomes a large bright fog-like blob that looks somewhat opaque and mechanical, not wispy." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point against a dark, moody background with warm colors. The chopsticks add a diagonal. The bowl is somewhat small and there is a lot of empty space." }, "vegan": { "score": 0.92, "reason": "Only mushrooms, vegetables, tofu, corn and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.42, "reason": "Noodles are thin, short, straight pale sticks that look like bean sprouts or matchsticks. They are rigid and lack continuous supple strands, and the mass is small and scattered on top of the broth." }, "toppings": { "score": 0.5, "reason": "Shiitake caps with a cross score read as mushrooms, and there are a bok choy leaf, tofu cubes, scallion rings and corn. The pieces are tiny and symbolic, and the scallion and chili bits look like confetti. Variety is decent but the details are small." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a dark glazed wall, a red accent band and a soft highlight on the left. The perspective is consistent and the base curves naturally. It looks slightly flat and has no visible foot or contact shadow." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange with a vignette and some depth, and small glow patches suggest oil. It reads rather flat and uniform, and the toppings sit on top instead of being immersed." }, "steam": { "score": 0.6, "reason": "Steam builds up over the frames as soft translucent plumes that rise and spread, and it responds to the pointer. It becomes a dense, opaque white fog that partly obscures the background, and it is blobby rather than wispy." }, "composition": { "score": 0.68, "reason": "A single bowl is centered in a dark, moody scene with a warm palette and the food as the focus. The chopsticks add a diagonal, and there is no clutter. The bowl is somewhat small and the lower area is empty." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are mushrooms, tofu, bok choy, scallion, corn and chili, all plant-based. There is no meat or egg." } } ], "diagnostics": { "vegan": 0.9066666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.43333333333333335, "toppings": 0.5333333333333333, "bowl": 0.8, "broth": 0.6, "steam": 0.6833333333333333, "composition": 0.8333333333333334 }, "reward": 0.6258333333333334, "sampleStdDev": 0.010103629710818464, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are pale thin straight sticks scattered across the surface with almost no curvature, no coherent bundle or volume; they read as rigid splinters rather than supple strands, dominating the surface." }, "toppings": { "score": 0.6, "reason": "Readable variety: scallion rings, yellow corn kernels, red chili flecks, sesame seeds, two brown mushroom-caps with cross cuts, and a green/cream leaf. Shapes are flat vector forms and the mushroom discs are somewhat symbolic, but they avoid generic plastic blobs." }, "bowl": { "score": 0.8, "reason": "Coherent dark lacquer-style bowl: elliptical rim with gray lip, consistent cavity and hemisphere body, red accent band, subtle sheen and a faint ground contact. Stylized but well-constructed with no broken joins or perspective clash." }, "broth": { "score": 0.6, "reason": "Warm orange-brown gradient surface with oil-spot ellipses is appetizing and readable, but the liquid looks flat and toppings/noodles appear pasted on the surface rather than immersed, with weak occlusion." }, "steam": { "score": 0.5, "reason": "Steam is clearly animated across normal frames (near-none at 1s, small wisps at 2.7s, larger cloud at 4.3s, big plume in pointer frames), so it is not static. However the mature plume is one amorphous, fairly opaque glowing fog blob rather than translucent irregular wisps." }, "composition": { "score": 0.8, "reason": "Centered focal dish in a restrained dark warm scene, diagonal chopsticks add interest, no text/UI/clutter. Later frames' steam is over-bright relative to the food, and the mobile crop leaves the dish small with much empty space, but presentation is clean and balanced." }, "vegan": { "score": 0.9, "reason": "Visible toppings (scallions, corn, chili, sesame, mushroom-like caps, greens, plant-toned broth) plausibly form vegan ramen; no egg, meat, or seafood shapes are visible." } }, { "noodles": { "score": 0.4, "reason": "Noodles appear as short, rigid pale rods scattered flat across the broth surface, like broken sticks or matchsticks rather than continuous supple strands. No coherent pile, volume, or believable bends; strands are broken and lie stiffly on the surface." }, "toppings": { "score": 0.55, "reason": "Recognizable scallion rings, yellow corn, red chili bits, white sesame seeds and small beige cubes (plausibly tofu) plus a green leaf blob. However, shapes are very simple flat vector symbols: the dark brown circles with plus marks read as generic tokens, and the leaf is a crude two-tone blob with little organic texture." }, "bowl": { "score": 0.8, "reason": "Coherent dark bowl with a consistent elliptical rim, thin highlight arcs, a red-brown lacquer accent band, smooth gradient body and a soft ground shadow. Chopsticks rest convincingly across the rim. Minor quibbles: the inner rim ring is heavy and the accent band is a bit graphic." }, "broth": { "score": 0.6, "reason": "Warm orange-brown liquid surface with a soft radial gradient, darker pools and blurred sheen ellipses that suggest depth. But ingredients read as pasted onto the surface rather than sitting in the broth \u2014 noodles and toppings show little immersion or occlusion." }, "steam": { "score": 0.75, "reason": "Translucent whitish wisps rise above the bowl and clearly evolve across the time-ordered desktop frames: faint at 1s, a fuller spreading cloud by 4-8s, with changing shape between frames. The plume grows fairly large and grazes the chopsticks but stays translucent and does not hide the food." }, "composition": { "score": 0.85, "reason": "Centered, readable focal dish on a warm dark atmospheric background with a subtle glow behind the bowl; chopsticks add a pleasing diagonal and there is no text, UI or clutter. Attractive and balanced, though the scene is very minimal." }, "vegan": { "score": 0.5, "reason": "Visible toppings (scallions, corn, chili, sesame, leafy green, pale cubes) are plant-plausible, but the dark brown circles with cross marks are ambiguous \u2014 they could be braised mushroom slices or stylized narutomaki fish cakes, so appearance is ambiguous rather than clearly animal-derived." } }, { "noodles": { "score": 0.45, "reason": "Noodles are pale straight sticks lying flat in a jumbled pile across the broth; uniform thickness, no supple bends or coherent strand volume, reading as rigid scattered straws rather than noodles." }, "toppings": { "score": 0.45, "reason": "Small scallion rings, corn, chili flakes and sesame are readable, but the two large dark discs with pale plus-marks and the flat cream wedge are featureless generic disks that dominate the topping area; the green shape is a simple blob." }, "bowl": { "score": 0.8, "reason": "Well-constructed flat-style bowl: coherent elliptical rim, matching interior, tapered dark ceramic body, red accent band, subtle rim highlight and consistent perspective between opening and base; only minor weakness is the lack of visible ground contact/shadow." }, "broth": { "score": 0.6, "reason": "Amber broth surface has a warm radial gradient with lighter oil pools, but toppings (especially the discs and noodle mass) sit fully on top with no occlusion or immersion shadow, giving a pasted-on feel; oil droplets are simple flat ellipses." }, "steam": { "score": 0.8, "reason": "Steam clearly evolves across frames: faint wisps at 1s, a larger translucent plume by 2.7s and 4.3s, shifting further in pointer frames, rising, spreading and fading without obscuring the food; slightly amorphous cloud-like shape in later frames." }, "composition": { "score": 0.85, "reason": "Centered focal dish in a dark atmospheric scene with warm vignette, chopsticks resting across the bowl add interest, coherent warm palette, no text, UI or clutter; dish remains the clear focus." }, "vegan": { "score": 0.75, "reason": "Visible ingredients (scallion, corn, chili, sesame, noodles, plant-colored broth) are plausible vegan items, but the dark cross-marked discs and pale cream wedge are ambiguous and could read as fish-cake-like toppings." } } ], "diagnostics": { "vegan": 0.7166666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8744444444444445, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "304ea78eff9f0ce79493b8b4242c953b2093372076d8e79dd29a6a06b7b34e3d", "gradedAt": "2026-09-30T22:49:38.819989+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6055555555555555, "toppings": 0.7166666666666667, "bowl": 0.8077777777777778, "broth": 0.6866666666666666, "steam": 0.7122222222222222, "composition": 0.8111111111111111 }, "reward": 0.7184444444444444, "runId": "anthropic/fable-5.1/max", "rewardStdDev": 0.07481279104340183, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.66, "bowl": 0.88, "broth": 0.6533333333333333, "steam": 0.73, "composition": 0.86 }, "reward": 0.7018333333333334, "sampleStdDev": 0.0005773502691896262, "judgments": [ { "noodles": { "score": 0.5, "reason": "Cream-colored strands have consistent thickness and some layered shading, but the central mass is dominated by tightly packed maze-like bends with little readable continuous overlap." }, "toppings": { "score": 0.66, "reason": "Corn, scored mushroom caps, porous tofu, scallion rings and veined greens are recognizable and varied. However, the mushrooms are disk-like, leaves are uniformly smooth, and nori resembles a stiff rectangular card." }, "bowl": { "score": 0.88, "reason": "A coherent elliptical rim, visible inner wall, rounded body and grounded shadow establish convincing depth. Fine horizontal ceramic texture, speckles and restrained rim highlights add polish; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm orange color, scattered oil droplets and submerged noodle silhouettes suggest appetizing soup. The surface remains rather uniform and opaque, with weak immersion cues around several large toppings." }, "steam": { "score": 0.73, "reason": "Translucent vapor changes position and distribution across the normal frames, extending from the soup and fading upward. The broad blurred clouds lack distinct curling wisps and slightly wash out the central food." }, "composition": { "score": 0.86, "reason": "The single warmly lit bowl is clearly isolated against a restrained dark wooden setting, with harmonious orange and green accents and no distracting UI. Mobile framing preserves the complete dish; the placement is somewhat low with generous empty space above." }, "vegan": { "score": 0.93, "reason": "Visible tofu, mushrooms, corn, greens and seaweed plausibly form vegan ramen. A reddish curved garnish is less identifiable, but no unmistakable animal-derived ingredient is visible." } }, { "noodles": { "score": 0.5, "reason": "Numerous cream-colored strands form a recognizable mound, but tight angular maze-like bends dominate and individual continuous strands are difficult to follow." }, "toppings": { "score": 0.66, "reason": "Corn, scored mushrooms, porous tofu, scallion rings and veined greens provide readable variety. Mushroom caps are disk-like, leaves are overly uniform, and nori resembles a rigid rectangular card." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, visible inner wall, rounded body and fine horizontal ceramic texture establish convincing depth and material. Highlights follow the rim and the bowl sits plausibly on the table; no material construction defect is observed." }, "broth": { "score": 0.66, "reason": "Warm orange liquid with scattered oil droplets and partly submerged noodles is appetizing, but its broad surface is rather flat and several toppings have weak liquid-contact cues." }, "steam": { "score": 0.73, "reason": "Translucent vapor changes position and density across normal frames, spreading above the soup without hiding most ingredients. It reads more as broad blurred haze than distinct irregular rising wisps." }, "composition": { "score": 0.86, "reason": "A single clearly framed dish, warm food colors against a restrained dark wooden setting, and coherent illumination keep the ramen central on desktop and mobile. There is no distracting text or clutter; the generous upper negative space slightly weakens balance." }, "vegan": { "score": 0.94, "reason": "Visible tofu, mushrooms, corn, greens, seaweed and scallions plausibly form vegan ramen. The reddish strips are somewhat ambiguous, but no unmistakable animal-derived topping is visible." } }, { "noodles": { "score": 0.5, "reason": "Cream-colored strands have consistent thickness and some overlap, but the central mound is dominated by tightly packed, maze-like squiggles with little readable strand continuity." }, "toppings": { "score": 0.66, "reason": "Corn, scored mushroom caps, speckled tofu, scallions and veined greens are recognizable and varied. Leaves are overly regular, mushrooms resemble decorated disks, and nori is a stiff rectangular sheet." }, "bowl": { "score": 0.88, "reason": "A coherent elliptical rim, shaded inner wall and rounded exterior establish convincing depth. Fine horizontal ceramic texture, rim highlights and a contact shadow support the material and placement; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm orange broth has scattered oil-like droplets and some submerged noodle detail. The surface remains rather uniform and opaque, with several toppings appearing laid on top rather than convincingly immersed." }, "steam": { "score": 0.73, "reason": "Translucent vapor changes position and density across the normal frames and spreads above the soup. It remains somewhat broad and fog-like rather than forming clearly articulated rising wisps, and lightly washes out central ingredients." }, "composition": { "score": 0.86, "reason": "The single dish remains a clear focal point against restrained dark wood, with warm food colors and coherent soft lighting. Desktop and mobile preserve the whole bowl without clutter; the large upper empty area makes the framing slightly bottom-heavy." }, "vegan": { "score": 0.95, "reason": "Visible tofu, mushrooms, corn, greens, seaweed and noodles plausibly form vegan ramen. The reddish strip is not clearly identifiable, but no unmistakable animal-derived topping is visible." } } ], "diagnostics": { "vegan": 0.94 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.6733333333333333, "bowl": 0.6933333333333334, "broth": 0.6133333333333333, "steam": 0.6066666666666667, "composition": 0.74 }, "reward": 0.6533333333333333, "sampleStdDev": 0.006506407098647717, "judgments": [ { "noodles": { "score": 0.6, "reason": "The noodle mass has a fine wavy texture and reads as instant-style ramen. It is small and compact, and the steam glare washes out the center. The strands are repetitive and not clearly continuous, and the red line across them looks odd." }, "toppings": { "score": 0.68, "reason": "The toppings are varied and recognizable: shiitake with scored tops, tofu cubes, bok choy, corn, nori, bamboo strips and scallions. They are illustrated in a clean vector style with some sheen. The nori is a rigid rectangle, the tofu is blocky, and the scattered dots are generic." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a dark speckled glaze body, a tapered base and a warm rim highlight. The perspective is consistent. The base contact is soft and the wall has limited detail." }, "broth": { "score": 0.62, "reason": "The orange broth reads as a spicy miso-style liquid with floating oil droplets and a subtle sheen. It looks fairly flat and opaque, and the toppings sit on it without much immersion or occlusion." }, "steam": { "score": 0.62, "reason": "Soft translucent wisps drift and change position across frames, so the motion is visible. The plumes are blobby, and a glare haze over the noodles and toppings partly obscures the food." }, "composition": { "score": 0.75, "reason": "The bowl is a centered focal point on a dark wood background with warm lighting and no distracting UI. The bowl sits a little low, and there is a lot of empty dark space above it." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, bok choy, corn, nori, scallions and bamboo, all plant-based. No egg, meat or fish is visible." } }, { "noodles": { "score": 0.6, "reason": "The noodle mass is dense and wavy with a believable ramen-curl texture, and it sits in the broth. It reads as a uniform patterned clump with little individual strand definition. A steam glow washes out the center." }, "toppings": { "score": 0.68, "reason": "The toppings are varied and readable: nori, shiitake with scoring marks, tofu blocks, bok choy, corn, scallions, and menma. The nori is a flat rectangle, the tofu is blocky, and the bok choy is stylized. Overall this is a competent illustration without much texture." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a dark glazed body, a tapered base, and a warm rim highlight. The perspective is consistent. The lower body is somewhat flat and murky, and contact with the table is weak." }, "broth": { "score": 0.62, "reason": "The orange broth has a sheen, floating oil droplets, and chili flecks, so it is appetizing. It looks fairly flat and uniform, and the ingredients sit on top more than they sit immersed." }, "steam": { "score": 0.6, "reason": "The steam is translucent and soft, and it changes position and shape across frames with wisps drifting up. It is fairly diffuse and blobby, and it partly hazes the food in the center." }, "composition": { "score": 0.75, "reason": "The bowl is a centered focal point against a dark wood background with warm colors and a vignette. There is no clutter or text. The bowl sits a little low, with empty space above it." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, vegetables, corn, and nori, with no visible animal products. The red strip on the noodles is ambiguous but probably chili." } }, { "noodles": { "score": 0.6, "reason": "The noodle mass is a dense, curly, wavy cluster that reads as instant ramen and has believable volume. It is small and somewhat texture-patterned, and the flat repeated wave pattern is partly washed out by glow. A red chili line sits on top." }, "toppings": { "score": 0.66, "reason": "The toppings are varied and recognizable: nori, shiitake with scored tops, tofu blocks, bok choy, corn, bamboo shoots and scallions. The shapes are fairly clean and vector-like, and the tofu blocks and mushrooms are a bit generic. The bok choy is decent." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a dark speckled glaze body, a foot and a consistent perspective. The rim has a sheen and thickness. The base contact is soft, and the body is a little flat and dim." }, "broth": { "score": 0.6, "reason": "The orange broth has oil droplets and some depth, and the toppings sit in it reasonably. The surface is fairly flat and uniform with little reflection, and it is slightly washed by steam glow." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise and drift differently between frames, with a glow over the noodles. They are fairly blobby and diffuse, and they haze the food in the center." }, "composition": { "score": 0.72, "reason": "The bowl is a centered focal point against a dark wood background with warm colors and a vignette. The framing is balanced and uncluttered, with no text. There is a lot of empty dark space above." }, "vegan": { "score": 0.9, "reason": "The toppings are tofu, mushrooms, vegetables, corn, nori and noodles, with no visible animal products. The orange broth is ambiguous but plausible as a plant-based broth." } } ], "diagnostics": { "vegan": 0.9266666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.8166666666666667, "bowl": 0.85, "broth": 0.7933333333333333, "steam": 0.8, "composition": 0.8333333333333334 }, "reward": 0.8001666666666667, "sampleStdDev": 0.006525590650150606, "judgments": [ { "noodles": { "score": 0.75, "reason": "Dense mass of wavy yellow strands with highlights and a red chili strand woven through reads as curly cooked ramen with plausible volume; slight uniformity and flatness in the central mass keep it below top marks." }, "toppings": { "score": 0.8, "reason": "Good variety and legibility: shiitake slices with cross-cut pattern, charred tofu cubes, fanned bok choy, nori sheet with fold, bamboo strips, corn, scallion rings and sesame. Mushroom asterisk cuts and blocky corn kernels are stylized but recognizable; no generic plastic shapes dominate." }, "bowl": { "score": 0.85, "reason": "Coherent dark stoneware bowl: elliptical rim with lighter inner band, speckled glazed walls tapering to a footed base, rim highlight and proper contact shadow on the wood. Perspective is consistent between rim ellipse and body; no broken joins observed." }, "broth": { "score": 0.78, "reason": "Appetizing orange miso-toned surface with restrained sheen, oil droplets and chili specks; noodles, mushrooms and tofu sit in the liquid with some occlusion. Some scattered red/green specks look slightly pasted on top rather than fully submerged." }, "steam": { "score": 0.8, "reason": "Across desktop frames steam plumes change shape and height: modest wisps at ~1s, taller developed plumes at ~3.4s, a different configuration at ~5.7s and thinner wisps in pointer frames, indicating continuous animated rising and fading. Translucent, never obscuring the food; frozen in reduced-motion frames as expected." }, "composition": { "score": 0.8, "reason": "Warm-lit bowl is the clear focal point against a dark wood background with vignette, no text or clutter. Desktop framing is balanced; the mobile viewport leaves large empty dark areas above and below the small dish, weakening proportion slightly." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, corn, bok choy, bamboo shoots, scallions, nori, chili and noodles in plant broth; no egg, meat, fish or dairy imagery present." } }, { "noodles": { "score": 0.7, "reason": "Curly ruffled noodle bed with coherent wavy texture, believable thickness and a chili-oil swirl draped over it; reads clearly as noodles in the stylized art, though the mass is somewhat a flat ruffled patty with few strands visibly dipping into the broth and a few stray squiggles pasted on the surface." }, "toppings": { "score": 0.85, "reason": "Strong variety and detail: scored shiitake slices with classic star cuts, a textured nori sheet, tofu slabs with browned spots, bok choy with white stalks and green leaves, bamboo strips, corn, scallion rings, chili slices and sesame. Only minor quibbles: corn kernels slightly blocky and one tofu edge has an odd melted lip." }, "bowl": { "score": 0.85, "reason": "Coherent donburi with consistent elliptical rim, visible inner wall, thick rim highlight, speckled ceramic body, and grounded contact shadow. Minor: exterior streak texture and front drip stains read slightly arbitrary." }, "broth": { "score": 0.8, "reason": "Appetizing orange miso-tonkotsu surface with oil droplets, bubbles and floating chili/scallion specks; ingredients sit partially submerged with reasonable occlusion. Slightly uniform in the center where the glow flattens detail." }, "steam": { "score": 0.8, "reason": "Normal frames at ~1.0s/3.4s/5.7s/9.9s/13.5s show translucent irregular wisps that rise, drift and change shape continuously without hiding the food; reduced frames correctly freeze. Slight blobbiness near the plume tops and a strong flare at the source." }, "composition": { "score": 0.85, "reason": "Well-centered focal bowl on dark wood with warm rim light, no clutter or UI, attractive warm palette. Desktop framing is excellent; mobile leaves somewhat generous empty dark space around the smaller bowl." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, shiitake, nori, bok choy, bamboo, corn, scallion and chili in plant broth; the red swirl is chili oil/thread, not fish cake. No animal-derived ingredients apparent." } }, { "noodles": { "score": 0.7, "reason": "Central mound of curly ramen with wiggle texture and shading reads as noodles, but strands stay confined to a repeating mass without continuous strands dipping into the broth; a glow washes over the center." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: bok choy with leafy detail, toasted tofu cubes, shiitake slices with cut-face gill marks, corn pile, bamboo strips, upright nori sheet, scallion rings. Minor stylization: mushroom gills are mechanical asterisk marks and corn cluster is uniform." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim with thickness and inner highlight, tapered walls, glazed base with drip details, believable contact shadow on the wood. Perspective is consistent; only minor flattening of the foot." }, "broth": { "score": 0.8, "reason": "Appetizing miso-toned surface with restrained sheen, oil droplets, chili specks and floating scallion rings; ingredients sit in the liquid with soft contact rather than pasted on top. Surface reads slightly flat in spots under the steam glow." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the wisps change shape, rise, spread and fade over the bowl \u2014 translucent, irregular, food stays visible. Reduced frames correctly freeze it; wisps are occasionally blobby near the rim." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark atmospheric wood table with warm rim light and balanced negative space; no clutter, text or competing elements. Steam glow adds depth without distracting from the food." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu, corn, bok choy, bamboo, nori, scallions and chili specks; no egg, meat, fish or dairy appears, and the red squiggle reads as chili drizzle rather than naruto." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9555555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100.0, "frameIntervalP95Ms": 116.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8c189c63b3e7c6e25bcc799c4b5b9ca5e76adc8d2f26888c05364c67f3e5cd64", "gradedAt": "2026-09-30T22:52:11.983893+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5677777777777778, "toppings": 0.6644444444444444, "bowl": 0.7577777777777778, "broth": 0.6166666666666667, "steam": 0.5755555555555556, "composition": 0.7688888888888888 }, "reward": 0.6561111111111111, "runId": "anthropic/fable-5.1/medium", "rewardStdDev": 0.07124377815589177, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.45333333333333337, "toppings": 0.64, "bowl": 0.78, "broth": 0.52, "steam": 0.49333333333333335, "composition": 0.7733333333333333 }, "reward": 0.6100000000000001, "sampleStdDev": 0.006244997998398403, "judgments": [ { "noodles": { "score": 0.45, "reason": "Yellow strands are readable with some curved overlaps, but mostly form stiff, angular sticks and a dense central scribble rather than supple noodles." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, veined greens, scallion rings and tofu provide recognizable variety. Mushroom caps are disk-like, tofu lacks texture, and corn resembles glossy beads." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inner cavity and smoothly shaded tapering walls form a coherent vessel. The dark oval foot and weak surface contact make its support less convincing." }, "broth": { "score": 0.52, "reason": "Warm amber color, scattered oil spots and subtle ripples suggest soup, but the surface feels flat and opaque, with ingredients largely layered above it rather than immersed." }, "steam": { "score": 0.48, "reason": "Normal frames show changing translucent clouds above the soup without hiding the toppings. They resemble broad glowing fog patches rather than fine, irregular rising wisps; continuous ascent and fading are not clearly demonstrated." }, "composition": { "score": 0.78, "reason": "A single readable dish, restrained dark background and warm green-gold palette establish a clear focus without clutter. Strong steam bloom competes with the food, and mobile chopsticks are clipped at the right edge." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, tofu-like cubes, scallions, corn and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.46, "reason": "Golden strands have readable overlaps and highlights, but mostly form stiff straight segments and angular bends around a thin scribbled center rather than supple noodle volume." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, leafy greens with pale stems, scallion rings and tofu are recognizable and varied. Mushrooms are smooth disk-like forms, tofu lacks texture, and the yellow kernels resemble glossy beads." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inset cavity and shaded tapered walls form a coherent deep vessel. Rim highlights convey thickness, though the dark oval foot and weak table contact make the base feel somewhat unsupported." }, "broth": { "score": 0.53, "reason": "Warm amber color, scattered oil spots and faint ripples suggest soup, but the surface is largely flat and opaque, with ingredients appearing laid on top rather than partially immersed." }, "steam": { "score": 0.52, "reason": "Normal frames show changing translucent clouds above the soup, leaving the food readable. The steam appears as broad glowing blurred patches rather than distinct irregular rising wisps; continuous rise and fading are not clearly established." }, "composition": { "score": 0.78, "reason": "A single readable dish, warm food colors against a dark setting and diagonal chopsticks create a clean focal composition without text or clutter. The large bowl body outweighs the food, and bright steam glow competes with the toppings." }, "vegan": { "score": 1, "reason": "Visible mushrooms, greens, scallions, tofu-like pieces, yellow kernels and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.45, "reason": "Yellow strands are readable and overlap, but mostly form stiff straight segments and angular bends, with a thin scribbled center rather than supple noodle volume." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, veined greens, scallion rings and shaded tofu provide recognizable variety. Shapes remain overly smooth and geometric, with little organic texture and bead-like corn." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, visible inner wall and tapered shaded body form a coherent deep vessel. Broad highlights suggest dark ceramic, though the black foot and weak ground contact feel simplified." }, "broth": { "score": 0.51, "reason": "Warm amber color and scattered oil spots suggest soup, but the surface looks mostly flat and opaque; ingredients cast graphic shadows and appear laid atop it rather than immersed." }, "steam": { "score": 0.48, "reason": "Normal frames show changing translucent clouds above the soup, leaving food readable. Steam resembles broad glowing fog patches rather than irregular rising wisps; continuous rise and fading are not clearly established." }, "composition": { "score": 0.76, "reason": "A single readable dish, warm accents and uncluttered dark surroundings create a clear focus. The tall bowl dominates the food, steam glare is strong, and chopsticks nearly reach the mobile edge." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, scallions, corn, tofu-like cubes and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.62, "bowl": 0.66, "broth": 0.6066666666666667, "steam": 0.55, "composition": 0.7 }, "reward": 0.6201666666666666, "sampleStdDev": 0.009928914005737631, "judgments": [ { "noodles": { "score": 0.6, "reason": "Thin pale strands overlap in a believable tangle with soft shadows and a nest-like volume. They read as short straight segments, somewhat stiff and repetitive, with few long curved bends." }, "toppings": { "score": 0.62, "reason": "Shiitake with cross scoring, bok choy with leaves and stalks, tofu cubes, scallion rings, corn and chili threads give good variety and readable shapes. The tofu cubes are blocky, the scallions are small and slightly plastic, and there is no nori." }, "bowl": { "score": 0.68, "reason": "The dark bowl has a coherent elliptical rim, a thick lip with highlight, tapered walls and a base shadow. Matte shading is convincing. The rim is slightly off from the body, the base looks a bit flat, and the chopsticks rest oddly." }, "broth": { "score": 0.6, "reason": "The orange broth has a warm gradient, a glow highlight, chili oil specks and faint ripples in the pointer frame. It reads somewhat flat, and the noodles and toppings sit on it more than in it." }, "steam": { "score": 0.5, "reason": "Soft translucent steam changes shape and brightness across frames, but it is a blurry glowing blob rather than wispy strands. It is bright and hazy, sometimes blooming into a hot spot, and it is detached from the surface." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point in a dark, warm, atmospheric scene with good color contrast and no clutter. The chopsticks add some diagonal interest, and the steam glow is slightly overpowering." }, "vegan": { "score": 0.93, "reason": "Mushrooms, tofu, bok choy, scallions, corn and noodles all read as plant-based. There is no egg, meat or fish." } }, { "noodles": { "score": 0.6, "reason": "Thin pale strands form a coherent nest with shadows and overlap, but they read as short straight matchstick-like pieces with little continuous curvature, so they look somewhat rigid and scattered." }, "toppings": { "score": 0.62, "reason": "Shiitake with cross scoring, bok choy leaves, tofu cubes, scallion rings, corn and chili threads are all recognizable and varied. The tofu cubes are plain and boxy, and the small elements look like clip-art." }, "bowl": { "score": 0.6, "reason": "The dark bowl has a coherent elliptical rim, tapered body and a base shadow. The rim perspective and the body width don't fully match, the body is flat and dark with little material detail, and the base looks slightly detached." }, "broth": { "score": 0.6, "reason": "Orange broth has a warm gradient, a glow highlight, chili oil specks and faint ripples, and the ingredients have contact shadows. It still looks fairly flat and opaque, and the ingredients sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "A soft translucent glow rises above the bowl and changes shape and position across frames, and it reacts to the pointer. It reads as a blurry blob or light bloom rather than distinct wisps, and it is bright enough at times to wash out the rim." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point against a dark warm background, and the chopsticks add a diagonal. The palette is cohesive, but the bowl body is large, dark and empty, and the steam glow is a bit distracting." }, "vegan": { "score": 0.93, "reason": "The visible toppings are mushrooms, tofu, bok choy, scallion, corn and noodles, with no meat, egg or fish." } }, { "noodles": { "score": 0.6, "reason": "Thin yellow strands overlap in a believable nest with shading and some bends. They read as short straight bean-sprout-like sticks, somewhat rigid and repetitive, and sit as a pile on the broth." }, "toppings": { "score": 0.62, "reason": "Shiitake with cross scoring, bok choy leaves, tofu cubes, scallion rings, corn and chili threads give good variety. Tofu cubes are blocky, scallions are a bit glossy and symbolic, and there is no nori." }, "bowl": { "score": 0.7, "reason": "Dark ceramic bowl with a coherent elliptical rim, inner wall, tapered body and a grounding base shadow. Rim highlight is nice. The body is rather flat and the base shadow looks slightly detached." }, "broth": { "score": 0.62, "reason": "Warm orange broth with a glow gradient and small chili oil spots, and faint ripples appear in a later frame. Noodles sit on top with little immersion, and the surface is fairly flat." }, "steam": { "score": 0.55, "reason": "Soft glowing steam changes shape across frames and rises and fades. It is a blurry blob or glow, sometimes overly bright, rather than wispy translucent strands. Reduced motion shows it dimmed." }, "composition": { "score": 0.72, "reason": "Centered bowl with a dark moody backdrop, warm palette and a chopstick diagonal keep the food as the focus. There is no clutter, though the lower half is an empty bowl body and the steam glow is somewhat strong." }, "vegan": { "score": 0.93, "reason": "Mushrooms, tofu, bok choy, corn, scallions and noodles are visible. There is no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.93 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7333333333333333, "bowl": 0.8333333333333334, "broth": 0.7233333333333333, "steam": 0.6833333333333333, "composition": 0.8333333333333334 }, "reward": 0.7381666666666666, "sampleStdDev": 0.02191080403210555, "judgments": [ { "noodles": { "score": 0.7, "reason": "Wavy pale-yellow strands with varied bends tangle believably in the broth center; some strands read slightly flat and uniform, with a few abrupt ends and limited volume layering, but they are coherent noodles rather than maze scribbles." }, "toppings": { "score": 0.75, "reason": "Clear variety: scored shiitake slices with cross cuts, bok choy with green leaves and white stems, beige tofu cubes, scallion rings, corn kernels, chili threads and sesame specks. Corn is slightly glossy-spherical and tofu blocks a bit plain, but most items read as real ingredients." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with a highlight arc, smooth tapering walls, dark foot with contact shadow, chopsticks resting convincingly across the rim. Stylistic simplification, but no broken joins or perspective conflicts." }, "broth": { "score": 0.7, "reason": "Warm orange miso-toned surface with chili-oil specks and a soft specular glow; ingredients sit on the surface with subtle shadows and faint swirl lines appear in one frame. Reads slightly flat and the broth sheen is modest." }, "steam": { "score": 0.8, "reason": "Across desktop frames 1-3 the translucent warm plumes clearly shift shape, rise, spread and fade (left plume drifts and changes between frames), staying wispy and not hiding the food. Reduced-motion frames appropriately show only a faint static glow." }, "composition": { "score": 0.85, "reason": "Balanced centered dish against a dark atmospheric backdrop with warm key light, steam as focal accent, and chopsticks as a restrained secondary element; no text, UI or clutter. Food remains the clear focus; mobile framing crops gracefully." }, "vegan": { "score": 1, "reason": "Visible ingredients are mushrooms, bok choy, tofu, corn, scallions, chili and sesame in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.6, "reason": "Yellow strands are continuous and readable as noodles, but most strokes are straight, straw-like sticks crossing in a haystack; few natural bends and little volume coherence." }, "toppings": { "score": 0.75, "reason": "Good vegan variety: shiitake slices with cross gill marks, bok choy leaves with pale stalks, tofu cubes, corn kernels, scallion rings, chili threads and sesame. Shapes are stylized but recognizable; tofu cubes and mushroom discs are slightly plain/flat." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: clean rim ellipse, consistent cavity and tapered walls, subtle glossy rim highlight, foot shadow at base, and chopsticks plausibly resting on the rim. No broken joins or perspective conflicts." }, "broth": { "score": 0.72, "reason": "Warm miso-toned surface with chili-oil specks and a light sheen; toppings sit on and read as immersed in the liquid. Slightly flat/uniform gradient limits depth." }, "steam": { "score": 0.65, "reason": "Steam shape visibly changes across normal frames (left blobs, then wider right plume), showing motion; it is translucent and does not hide the food, but reads as diffuse fog/light glow rather than distinct rising wisps." }, "composition": { "score": 0.85, "reason": "Well-balanced dark scene with the bowl as clear focal point, warm atmospheric palette, diagonal chopsticks adding interest, no clutter or text; the upper steam glow is slightly dominant but food stays central." }, "vegan": { "score": 1, "reason": "Visible toppings are mushrooms, bok choy, tofu, corn, scallion and chili; no animal-derived ingredients apparent." } }, { "noodles": { "score": 0.65, "reason": "Thin pale-yellow strands with wavy bends and occasional highlights read as noodles, but the mass is a fairly uniform squiggle heap concentrated mid-bowl rather than coherent strands emerging from the broth; repetition makes it slightly artificial." }, "toppings": { "score": 0.7, "reason": "Good variety: scored shiitake caps with contact shadows, bok choy with distinct leaves and stems, scallion rings, corn kernels, chili threads and sesame. Tofu cubes are clean geometric blocks with little texture, and shapes are quite stylized/flat." }, "bowl": { "score": 0.85, "reason": "Coherent donburi-style bowl: elliptical rim with a glazed inner band, matching wall shading, dark ceramic/lacquer material, plausible foot with cast shadow, and chopsticks resting convincingly across the rim. Minor softness where the rim highlight arc reads as a simple stroke." }, "broth": { "score": 0.75, "reason": "Warm miso-toned surface with chili-oil flecks and subtle shading; toppings sit in the liquid with small occlusion shadows. The back-center glow washes part of the surface, slightly flattening its texture." }, "steam": { "score": 0.6, "reason": "Steam is translucent and does not hide the food, and the plume's shape and position clearly change across the desktop frames (1037ms/3005ms/4881ms) and pointer frames, so it is animated. However it reads as an amorphous soft glow/fog blob rather than irregular rising wisps that taper and fade." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a dark atmospheric scene with warm backlight, chopsticks adding a diagonal accent, no text/UI/clutter, and the bowl fits cleanly on desktop and mobile viewports with no overflow. The heavy glow slightly competes with the food's focus." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, bok choy, tofu, corn, scallions, chili and sesame in a plant-toned broth; no animal-derived ingredients appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9766666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.29999999999927, "frameIntervalP95Ms": 83.40000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "6840f23694128c4b5ad633b77d7fa8c38cbd35796b3e9640817c3aa07664bb57", "gradedAt": "2026-09-30T22:50:50.443668+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6033333333333334, "toppings": 0.6522222222222223, "bowl": 0.7677777777777778, "broth": 0.6722222222222223, "steam": 0.67, "composition": 0.7877777777777778 }, "reward": 0.6837777777777778, "runId": "anthropic/fable-5.1/xhigh", "rewardStdDev": 0.04082120001139563, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6166666666666667, "bowl": 0.8633333333333333, "broth": 0.68, "steam": 0.6166666666666667, "composition": 0.8133333333333334 }, "reward": 0.6686666666666667, "sampleStdDev": 0.0007637626158259498, "judgments": [ { "noodles": { "score": 0.5, "reason": "Readable golden strands with shaded edges and overlapping volume, but predominantly straight, stiff ribbon-like segments and angular bends lack supple noodle flow." }, "toppings": { "score": 0.61, "reason": "Tofu has porous markings, mushrooms have scored caps, and greens have readable veins. However, leaves are repetitive round symbols, corn looks bead-like, and nori resembles rigid smooth panels." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, visible inner wall, rounded exterior, ceramic speckling and restrained highlights establish convincing depth and thickness. Ground shadow supports the bowl; no material construction defect is observed." }, "broth": { "score": 0.69, "reason": "Warm amber liquid has subtle surface highlights and partially submerged garnish. Broad blurry color patches and limited contact detail make some ingredients appear layered above the surface." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent plume shapes above the soup without hiding the food. The steam is excessively broad and blurred, with glowing cloud-like lobes rather than delicate irregular wisps." }, "composition": { "score": 0.82, "reason": "Warm food colors stand clearly against the dark wooden setting, with a single uncluttered focal bowl and coherent lighting. The diagonal chopsticks provide direction, but extend off-screen on mobile and the tall steam occupies substantial space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Clearly readable golden strands with layered overlap, but many are stiff, flat ribbons with angular bends rather than supple noodles." }, "toppings": { "score": 0.62, "reason": "Varied tofu, scored mushrooms, corn and greens have readable details. However, leaves resemble repeated round badges and nori resembles smooth rigid panels, limiting organic plausibility." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, recessed cavity and curved walls establish convincing depth; speckled dark ceramic, edge highlights and a contact shadow support the material. No material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading and restrained surface highlights suggest soup, with some submerged garnish. The noodle mound and tofu appear perched above a softly blurred surface with limited liquid contact." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent plume shapes above the soup without obscuring the food. The steam is broad and heavily blurred, resembling glowing fog more than fine irregular wisps." }, "composition": { "score": 0.8, "reason": "The isolated bowl is readable against a restrained dark wooden setting with coordinated warm lighting and no UI clutter. Long diagonal chopsticks compete somewhat with the food and extend beyond the mobile frame." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, seaweed, corn, greens and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.5, "reason": "Distinct pale strands have shading and readable overlaps, but the pile is dominated by stiff, flattened ribbons and angular bends rather than supple continuous noodles." }, "toppings": { "score": 0.62, "reason": "Porous tofu, scored mushroom caps, scallion rings and corn provide recognizable variety. However, the leaves are repetitive oval symbols and the nori resembles smooth upright cards with little organic texture." }, "bowl": { "score": 0.87, "reason": "The continuous elliptical rim, consistent inner wall and rounded exterior establish convincing depth and perspective. Speckled dark ceramic, restrained highlights and a contact shadow support the vessel; no material construction defect is observed." }, "broth": { "score": 0.67, "reason": "Warm amber coloring, soft reflections and submerged flecks suggest appetizing soup. Limited meniscus detail and the sharply layered ingredient mound make immersion less convincing." }, "steam": { "score": 0.61, "reason": "Normal frames show changing translucent plumes above the soup without obscuring the food. The steam is unusually broad and blurred, with bright floating patches rather than finely tapering irregular wisps." }, "composition": { "score": 0.82, "reason": "A clear focal bowl, warm restrained palette, dark wooden setting and coherent shadows create an attractive uncluttered scene without text or UI. Mobile framing leaves substantial empty space and clips the chopstick ends." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.66, "toppings": 0.64, "bowl": 0.6733333333333333, "broth": 0.62, "steam": 0.6266666666666667, "composition": 0.7 }, "reward": 0.6526666666666667, "sampleStdDev": 0.01887679351302371, "judgments": [ { "noodles": { "score": 0.62, "reason": "Noodles are cream-colored, overlapping and have some bends and volume, sitting in the broth. They look somewhat short and stubby, like cut pieces, with flat repetitive shapes and a few thin red scribble threads. They are competent but not supple." }, "toppings": { "score": 0.6, "reason": "The toppings are varied and recognizable: bok choy with veined leaves, shiitake with a scored star pattern, tofu cubes, enoki-like golden beads, scallion rings and two nori sheets. The nori stands upright as stiff rectangular cards, the tofu cubes are blocky and the golden beads look like beads, so they are still somewhat artificial." }, "bowl": { "score": 0.68, "reason": "The dark stoneware bowl has a coherent elliptical rim, visible inner wall, speckled texture, highlights and a soft base shadow. The base and foot on the lower left look slightly odd and the bowl merges into the dark background, but the perspective is mostly consistent." }, "broth": { "score": 0.62, "reason": "The orange-brown broth has subtle sheen and red chili oil patches, and the ingredients sit in it with some occlusion. The surface is fairly flat and murky, with little depth or ripple." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes drift and change shape across the frames, rising from the bowl and fading. They look blurry and somewhat large and blobby, and they lack wispy detail, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "The bowl is a clear focal point with warm colors against a dark wood atmosphere, and the chopsticks add a diagonal. The bowl is a bit small and low in the frame, and the large steam haze dominates the top of the image. There is no text or clutter." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, bok choy, nori, scallions, sesame and noodles. There are no visible animal products." } }, { "noodles": { "score": 0.68, "reason": "Noodles are a readable tangle with varied bends, overlap and some shading, but they look short and stubby with uniform thickness, and a few red threads read oddly. They sit on the broth plausibly, though the mass is slightly flat." }, "toppings": { "score": 0.66, "reason": "Variety is good: bok choy with veined leaves, shiitake with scoring, tofu cubes, nori sheets, corn or enoki-like beads and scallions. The tofu cubes are blocky and the nori stands upright like stiff cards. The scallion rings are fairly generic." }, "bowl": { "score": 0.68, "reason": "A dark speckled stone bowl with a coherent elliptical rim, inner wall and base, plus a highlight on the rim. The lower left of the base has an odd seam or lip, and the bottom is dim and fades into the background." }, "broth": { "score": 0.62, "reason": "The warm orange-brown broth has chili oil patches and some immersion shading. The surface is fairly flat and matte, with little depth or sheen, and the toppings sit on it without many ripples or contact cues." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes change shape across frames and rise and fade. They are a bit blurry and blobby, and they become bright and glowy in the later frames. They do not obscure the food." }, "composition": { "score": 0.68, "reason": "The bowl is a clear focal point in a dark, warm atmosphere with balanced colors. The bowl sits slightly low, with a lot of empty space above filled by steam. The chopsticks add a diagonal, and there is no distracting UI." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, bok choy, nori, corn and scallions, with no visible animal products. The red-orange broth is ambiguous but plausible." } }, { "noodles": { "score": 0.68, "reason": "Noodle strands are cream-colored with thick, supple curves and plausible overlap, sitting in the broth. They look a bit short and stubby, and the pile reads somewhat tangled and flat, with little strand definition." }, "toppings": { "score": 0.66, "reason": "Nori sheets, bok choy, shiitake with a scored pattern, tofu cubes, corn and scallions are all readable and varied. The nori stands upright like stiff cards, the tofu is blocky, and the scallion bits look like small beads." }, "bowl": { "score": 0.66, "reason": "The dark stone bowl has a coherent rim, a visible inner wall and a speckled texture. The base is murky, with an odd lighter outline at lower left, and the lower contour is a little unclear." }, "broth": { "score": 0.62, "reason": "The broth is warm orange-brown with red chili patches and ingredients sitting in it. It is fairly flat and opaque, and there is not much sheen or depth." }, "steam": { "score": 0.68, "reason": "A soft, translucent plume rises and changes shape across frames without hiding the food. It is a bit blurry and blobby, and it looks somewhat like a generic haze." }, "composition": { "score": 0.72, "reason": "The bowl is a clear focal point in a dark, moody scene with warm lighting and a vignette. The bowl sits low, there is a lot of empty space above it, and the chopsticks extend off to the right." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, bok choy, nori, corn and scallions all appear plant-based. There is no visible meat or egg." } } ], "diagnostics": { "vegan": 0.92 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7, "bowl": 0.7666666666666667, "broth": 0.7166666666666667, "steam": 0.7666666666666666, "composition": 0.85 }, "reward": 0.73, "sampleStdDev": 0.039686269665968894, "judgments": [ { "noodles": { "score": 0.7, "reason": "Continuous pale strands with varied bends, tube shading, and red chili threads, emerging from the broth; slightly uniform/stiff spaghetti-pile feel with a few flat loops, but readable and coherent." }, "toppings": { "score": 0.75, "reason": "Good variety: wavy nori sheets, bok choy fans, speckled tofu cubes with red garnish, scored shiitake caps, scallion rings and sesame. Leaves and mushrooms read correctly; glossy yellow corn beads and perfectly circular bok choy leaves look a bit stylized/plastic." }, "bowl": { "score": 0.8, "reason": "Coherent wide stone/ceramic bowl: elliptical rim with visible thickness, tapering speckled body, foot and contact shadow; chopsticks rest plausibly on the rim. Minor silhouette waviness on the left wall, otherwise clean construction." }, "broth": { "score": 0.75, "reason": "Amber broth surface with subtle sheen, darker patches and chili-oil spots; nori, bok choy and mushrooms appear seated with contact. Red oil blobs read as flat discrete ellipses rather than liquid droplets." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps that clearly change shape and position across desktop-1/2/3 and pointer frames, rising and fading without obscuring the food; motion is supported by the temporal frames." }, "composition": { "score": 0.85, "reason": "Balanced dark wooden scene with warm rim lighting, the dish as clear focal point, chopsticks balancing the right side, and no text or clutter; mobile crop keeps the bowl readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, bok choy, nori, mushrooms, corn, scallions and noodles in plant-toned broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.6, "reason": "Pale yellow strands form a coherent tangled pile with draping strands toward the chopsticks, but the strands are fairly uniform in thickness and color with a slightly stiff, wire-like quality and repetitive hooked loops at the left edge." }, "toppings": { "score": 0.65, "reason": "Good variety: tofu cubes with speckles, bok choy with veined leaves on pale stalks, shiitake disks with gill pattern, scallion rings, sesame and chili oil. Shortcomings: nori reads as rigid dark cards with a grid texture, and corn appears as glossy bead clusters rather than kernels." }, "bowl": { "score": 0.75, "reason": "Coherent wide stoneware bowl with a consistent elliptical rim, visible inner wall, speckled ceramic texture, tapering exterior to a small foot with contact shadow. Perspective between rim and base reads correctly; the cavity looks slightly shallow and the right inner edge is thin." }, "broth": { "score": 0.7, "reason": "Amber broth with restrained sheen, chili-oil flecks, floating sesame, and darker depth near the rim; tofu, mushrooms and nori sit believably in the liquid. Some red oil spots look pasted on and the surface is flat in places." }, "steam": { "score": 0.75, "reason": "Across desktop-1 through pointer-left the plume clearly changes shape, position and density over time, rising as translucent irregular wisps that fade with height without obscuring the food; it is slightly foggy/dense mid-column but not opaque." }, "composition": { "score": 0.85, "reason": "Warm dark wood scene with a soft spotlight, bowl as clear focal point slightly left of center, chopsticks adding a dynamic diagonal, and animated steam above; no text or clutter, and the mobile crop centers the dish well. Restrained and atmospheric presentation." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, bok choy, corn, scallions, sesame and nori in a plant broth; no egg, meat, fish or dairy appears in any frame." } }, { "noodles": { "score": 0.65, "reason": "Pale yellow strands form a coherent mound with overlapping bends and visible tapered ends, but they read somewhat flat and wire-like, uniform in thickness, and sit as a pile on the surface rather than being plausibly immersed; red chili threads add variety." }, "toppings": { "score": 0.7, "reason": "Good variety: wavy nori sheets, bok choy with drawn veins, speckled tofu cubes, corn kernels, scallion rings, and two brown mushroom caps. Shapes are readable but stylized and flat; the mushroom caps carry an odd white asterisk mark, and leaves border on lollipop-like." }, "bowl": { "score": 0.75, "reason": "Dark speckled bowl with a coherent elliptical rim, cavity, thick walls and a small foot with table shadow; perspective is consistent. Minor issues: the rim highlight is thin and slightly uneven on the left, and the wall/rim join is a bit soft." }, "broth": { "score": 0.7, "reason": "Brown miso-toned surface with red chili-oil pools, light swirls, and scattered sesame/grain specks; toppings mostly make believable contact. The surface is somewhat flat and the broth reads slightly solid around the edges." }, "steam": { "score": 0.7, "reason": "Wisps clearly change shape and position across desktop-1/2/3 and pointer frames, showing continuous rising motion that is properly suppressed in reduced-motion frames. However, the plumes are fairly opaque, cloud-like puffs concentrated above the bowl rather than delicate translucent wisps emerging from the soup." }, "composition": { "score": 0.85, "reason": "Balanced dark wood scene with warm vignette lighting, centered focal bowl, chopstones resting across the rim adding diagonal interest, and no distracting text or clutter; the food remains the clear focus. Slight heaviness of the steam mass at top center is the only nit." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, nori, bok choy, tofu, corn, scallions, mushrooms and chili oil; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9733333333333334, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50.0, "frameIntervalP95Ms": 66.60000000000036, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "4972041d41d899069903d409907dd196065826deb2ec0338f3f52ebb2bb53d36", "gradedAt": "2026-09-30T22:54:39.671878+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5955555555555555, "toppings": 0.6433333333333333, "bowl": 0.7811111111111111, "broth": 0.6311111111111111, "steam": 0.6544444444444444, "composition": 0.8033333333333333 }, "reward": 0.6777777777777778, "runId": "anthropic/opus-5/high", "rewardStdDev": 0.0379218250459391, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6733333333333333, "bowl": 0.8566666666666667, "broth": 0.62, "steam": 0.6633333333333333, "composition": 0.86 }, "reward": 0.7005, "sampleStdDev": 0.019371370627810507, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable overlaps, but broad, similarly curved loops form a sparse wire-like cage rather than a supple, integrated noodle mound." }, "toppings": { "score": 0.65, "reason": "Scallion rings, corn, scored mushroom caps, tofu and veined leaves provide readable variety. However, mushrooms are smooth ovals, tofu lacks texture, and the nori resembles a stiff rectangular card." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, curved walls and rounded base form a coherent vessel with restrained lacquer highlights and a grounding shadow. The thin rim and very dark lower wall slightly weaken material depth." }, "broth": { "score": 0.61, "reason": "Warm brown liquid has subtle oil-like flecks and shading within the cavity. Its surface remains rather flat, with limited wet contact or partial immersion around noodles and toppings." }, "steam": { "score": 0.67, "reason": "Normal frames show changing translucent vapor patches rising above the soup without obscuring food. The vapor is broad and diffuse, more like blurred haze than distinct irregular wisps; continuous fluidity cannot be established from these samples." }, "composition": { "score": 0.86, "reason": "A single clearly readable bowl anchors the dark scene, with warm food colors, restrained illumination and no distracting text or clutter. Mobile framing preserves the whole dish; substantial upper negative space is a minor imbalance." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.61, "reason": "Continuous golden strands have rounded bends and readable overlaps, but the sparse, broad looping arrangement looks somewhat ribbon-like and elevated rather than a supple immersed nest." }, "toppings": { "score": 0.69, "reason": "Scallion rings, corn, scored mushroom caps, veined leaves and shaded tofu provide readable variety. The nori is overly rectangular and stiff, while mushrooms and tofu have limited organic texture." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim joins smoothly to curved walls and a supported rounded base; restrained edge highlights and dark reflections suggest lacquer. No material construction defect is observed, though the exterior is quite dark." }, "broth": { "score": 0.64, "reason": "Warm brown liquid has small oil-like highlights and a readable surface inside the rim. Limited meniscus and immersion cues make several ingredients appear laid atop a relatively flat surface." }, "steam": { "score": 0.68, "reason": "Normal frames show translucent clouds changing position and density above the soup without obscuring the food. The steam reads more as a broad blurred haze than distinct irregular rising wisps; continuous motion cannot be established from these sparse captures." }, "composition": { "score": 0.86, "reason": "The single dish remains a clear focal point against a restrained dark background, with warm food colors, green accents and coherent highlights. Desktop and mobile framing are uncluttered; generous upper negative space is a minor imbalance." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.59, "reason": "Continuous golden strands have rounded bends and readable overlaps, but form a sparse, elevated cage of broad loops rather than a supple mound settling into the soup." }, "toppings": { "score": 0.68, "reason": "Scallion rings, corn, scored mushroom caps, veined leaves and shaded tofu provide readable variety. The nori is unusually straight and card-like, while mushrooms and tofu lack much organic surface texture." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim joins smoothly to curved dark walls, with restrained edge highlights, a rounded base and grounding shadow. The lacquer-like shading conveys depth; no material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm brown liquid has small oil-like highlights and a readable surface inside the rim. Broad flat color bands and limited immersion cues make several ingredients appear laid above the surface." }, "steam": { "score": 0.64, "reason": "Normal frames show translucent vapor changing shape and spreading above the soup without obscuring it. It reads more as broad blurred clouds than fine irregular rising wisps; continuous motion cannot be established from these sparse captures." }, "composition": { "score": 0.86, "reason": "The single dish remains a clear focal point against a restrained dark background, with warm food colors, coordinated lighting and ample breathing room on desktop and mobile. No text, clutter or material framing defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, corn, scallions and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.62, "toppings": 0.6333333333333333, "bowl": 0.68, "broth": 0.6066666666666667, "steam": 0.5666666666666667, "composition": 0.7 }, "reward": 0.634, "sampleStdDev": 0.014730919862656249, "judgments": [ { "noodles": { "score": 0.62, "reason": "Noodles form a tangled mound with overlapping curved strands and some volume. They look slightly flat and uniform in thickness, with loops that read as vector lines, and they sit atop the broth rather than deeply immersed." }, "toppings": { "score": 0.66, "reason": "Nori, scallion leaves, shiitake with cross scoring, tofu cubes, corn, scallion rings and sesame give good variety and recognizable shapes. The nori is a fairly rigid rectangle, the tofu is blocky, and the mushrooms are fairly dark and plain." }, "bowl": { "score": 0.68, "reason": "Dark hemispherical bowl with a translucent rim, consistent perspective and a subtle highlight reads as coherent. The bowl lacks a visible foot or base contact, and the rim looks glassy rather than ceramic." }, "broth": { "score": 0.62, "reason": "Dark brown broth with small oil droplets and a warm glow reads as appetizing soup. It is murky and fairly flat, with limited visible depth and immersion." }, "steam": { "score": 0.6, "reason": "Soft translucent steam rises and varies between frames, drifting and fading without hiding the food. It is a diffuse blob-like glow rather than distinct wisps, and it fades noticeably in the later frames." }, "composition": { "score": 0.7, "reason": "Centered bowl with a dark atmospheric background, warm lighting and the food as the clear focus. The bowl sits somewhat low, leaving a large empty top area, but there is no clutter or text." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, seaweed, corn, scallions and noodles are all plant-based, and no animal ingredients are visible." } }, { "noodles": { "score": 0.62, "reason": "The noodles form a readable tangled nest with overlapping strands of plausible thickness and warm color. The loops look somewhat uniform and flat, and the nest sits on top of the broth without much immersion." }, "toppings": { "score": 0.62, "reason": "Nori, shiitake with cross-cut scoring, tofu cubes, scallion rings, corn and a leaf are all readable and vegan-friendly. The shapes are simple and stylized, the tofu is blocky, and the scallion rings are scattered like confetti." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a deep hemispherical body and a soft highlight, and the perspective is consistent. The dark body is flat, the base contact is weak, and the rim looks a little like glass." }, "broth": { "score": 0.6, "reason": "The broth is a dark brown with droplets and some depth near the rim. It reads rather flat and murky, with little sheen, and the noodles and toppings sit on it without much contact or occlusion." }, "steam": { "score": 0.6, "reason": "A soft translucent plume rises above the bowl and changes shape and intensity across frames, from dense to faint. It reads as a diffuse glow rather than distinct wisps, and it is somewhat blobby." }, "composition": { "score": 0.72, "reason": "The bowl is centered in a dark, moody scene with warm lighting, and the food is the clear focus. There is no clutter. The bowl sits a bit low with empty space above, and the overall palette is dim." }, "vegan": { "score": 0.95, "reason": "The toppings are mushrooms, tofu, nori, scallions, corn and a leaf, with no visible animal products." } }, { "noodles": { "score": 0.62, "reason": "Noodles form a tangled nest with overlapping curved strands and some volume, but they look like uniform flat-shaded loops and sit on top of the broth rather than being immersed in it." }, "toppings": { "score": 0.62, "reason": "Nori, shiitake with cross cuts, tofu cubes, scallion rings, corn and a leaf are all readable and varied. The scallion rings are uniform and glossy, the tofu is blocky, and the nori is a rigid rectangle." }, "bowl": { "score": 0.68, "reason": "Coherent elliptical rim and a rounded dark body with subtle highlights and a grounded shadow. The rim looks slightly glassy and thin, and the base is barely defined." }, "broth": { "score": 0.6, "reason": "Dark brown broth with small oil droplets and a warm glow reads as soup. It is fairly flat and murky, and the toppings only partly feel immersed." }, "steam": { "score": 0.5, "reason": "Soft translucent haze rises and changes shape across frames. It is a blurry blob, not distinct wisps, and it fades out in the later frames." }, "composition": { "score": 0.68, "reason": "Centered single bowl in a dark, moody scene with warm lighting and no clutter. The bowl is small in the frame and the bottom area is empty." }, "vegan": { "score": 0.95, "reason": "Only noodles, mushrooms, tofu, nori, scallions and corn are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.6233333333333333, "bowl": 0.8066666666666666, "broth": 0.6666666666666666, "steam": 0.7333333333333333, "composition": 0.85 }, "reward": 0.6988333333333333, "sampleStdDev": 0.018036999011291594, "judgments": [ { "bowl": { "reason": "Dark lacquer-style bowl with a coherent elliptical rim, readable rim thickness, smooth tapering body, soft left highlight and a grounded contact shadow; perspective is consistent. The bright gray rim edge reads slightly metallic and there is no foot ring, but no broken geometry.", "score": 0.8 }, "broth": { "reason": "Dark brown broth with a restrained glossy sheen, subtle oil bubbles near the right, and scallion rings, corn and sesame sitting on the surface. The broth is quite dark and murky in places, the noodle nest rests on top rather than visibly sinking in, and the right tofu cube sits ambiguously at the inner wall edge.", "score": 0.65 }, "composition": { "reason": "Balanced centered focal dish on a clean dark background with warm backlight glow and steam; no text, UI or clutter across desktop and mobile frames. The scene is attractively presented, though the very dark, near-monochrome palette and large steam plume leave some upper space unoccupied.", "score": 0.85 }, "noodles": { "reason": "A cohesive pale noodle nest with continuous curved strands, plausible uniform thickness, visible cut ends and coherent overlaps. Strands are somewhat tubular and the looping is mildly repetitive, and the nest floats atop the broth, but no rigid ribbons or maze scribbles.", "score": 0.7 }, "steam": { "reason": "Translucent brownish wisps clearly animate across normal frames: narrow column in desktop-1, wider plume in desktop-2, diffuse faint cloud in desktop-3, fading in pointer-right and reforming rightward in pointer-left. Reduced frames are static as expected. The plumes are slightly blobby fog-like rather than delicate wisps but never hide the food.", "score": 0.7 }, "toppings": { "reason": "Good variety: scallion rings, corn, sesame, chili flakes, leafy herb, brown shiitake-like caps, two tofu cubes and a nori sheet. Mushrooms and scallions read well, but the tofu cubes look like clean sugar cubes, the nori is a stiff rounded-corner card shape, and garnish is scattered fairly uniformly.", "score": 0.65 }, "vegan": { "reason": "Visible ingredients are noodles, nori, mushrooms, tofu, scallion, corn, sesame and chili in plant broth; no meat, fish, egg or dairy is depicted.", "score": 1 } }, { "noodles": { "score": 0.5, "reason": "Noodles read as a tangled spaghetti-like mass with stiff, repeated overlapping loops; some continuity but little supple strand definition or depth in the pile." }, "toppings": { "score": 0.62, "reason": "Recognizable scallion rings, corn, tofu cubes, leafy herb, nori sheet and mushroom caps; however mushrooms carry odd X-shaped markings, tofu cubes look slightly flat, and variety is modest." }, "bowl": { "score": 0.82, "reason": "Coherent dark ceramic bowl with convincing rim highlight, smooth wall gradient, base contact shadow and consistent perspective across all frames; only minor glossiness keeps it from higher." }, "broth": { "score": 0.65, "reason": "Warm brown liquid surface with soft sheen and scattered sesame/chili specks; toppings mostly sit in the broth, though the surface is fairly flat and dark with limited depth." }, "steam": { "score": 0.75, "reason": "Translucent wisps that visibly shift shape and drift upward between the time-ordered frames without obscuring the dish; motion is subtle but real, and plumes are somewhat diffuse." }, "composition": { "score": 0.85, "reason": "Well-balanced dark atmospheric scene, clear focal bowl, warm rim-lit glow, no clutter or UI; dish stays the focus at both desktop and mobile sizes." }, "vegan": { "score": 1, "reason": "All visible toppings (noodles, scallions, corn, tofu, mushrooms, nori, herbs) are plausibly plant-based; no animal-derived ingredients appear." } }, { "bowl": { "reason": "Coherent dark lacquer bowl: clean elliptical rim with a metallic highlight, smooth walls tapering to a shadowed base with a soft contact shadow; perspective of rim, cavity and body agree. Minor: base dissolves into shadow and rim thickness is uniform.", "score": 0.8 }, "broth": { "reason": "Dark brown broth with warm central glow, oil droplets, sesame and chili flecks; tofu and mushrooms sit with small contact shadows suggesting immersion. Slightly flat/uniform in places and the speckle dots repeat mechanically.", "score": 0.7 }, "composition": { "reason": "Centered bowl as clear focal point in a restrained dark atmospheric scene with warm vignette glow; no text, UI or clutter; mobile crop keeps the dish readable and balanced.", "score": 0.85 }, "noodles": { "reason": "Pale tan nest with consistent strand thickness, some draping bends and highlight shading, but the tangle is fairly flat with repeated similar loops and little submergence into the broth.", "score": 0.6 }, "steam": { "reason": "Normal frames show the translucent plume evolving: taller and diffuse at 3s, narrower at 4.9s, nearly dissipated at 7.5s, faint wisps at 10.1s \u2014 continuous rise and fade without hiding food, though the plume is somewhat amorphous.", "score": 0.75 }, "toppings": { "reason": "Good variety: nori sheet, two shiitake slices, tofu cubes, scallion rings, corn, sesame, chili threads and a veined herb leaf. But tofu is rigid blocky cubes, mushrooms read as flat brown discs with a drawn X, and the nori looks like a stiff standing card.", "score": 0.6 }, "vegan": { "reason": "All visible ingredients are plant-based: noodles, nori, shiitake mushrooms, tofu, scallion, corn, sesame, chili and greens; no meat, egg or dairy forms appear.", "score": 1 } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50.0, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "66ef59c198ccbad857cd4c09292c72d482106c8c08101b3e9729ff75c183d24a", "gradedAt": "2026-09-30T22:55:52.835238+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6133333333333334, "toppings": 0.5833333333333334, "bowl": 0.77, "broth": 0.6444444444444445, "steam": 0.5622222222222222, "composition": 0.7444444444444445 }, "reward": 0.6457222222222223, "runId": "anthropic/opus-5/low", "rewardStdDev": 0.0844091117602131, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6733333333333333, "toppings": 0.52, "bowl": 0.8133333333333334, "broth": 0.6166666666666667, "steam": 0.38666666666666666, "composition": 0.73 }, "reward": 0.62, "sampleStdDev": 0.005567764362829967, "judgments": [ { "noodles": { "score": 0.68, "reason": "Continuous golden strands have readable bends, shading and layered crossings, but the closely repeated arches form a somewhat stiff, arranged mound rather than supple noodles settling into soup." }, "toppings": { "score": 0.52, "reason": "Scored mushroom caps, tofu cubes, corn and green garnish provide recognizable variety. However, the nori is a rigid rectangular panel, the leaf is a geometric fan, and tofu surfaces lack food texture." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, inset cavity and smoothly tapering body form coherent perspective and depth, with a fine rim highlight and grounding shadow. The very dark walls obscure material character and base definition." }, "broth": { "score": 0.62, "reason": "Warm brown liquid has scattered sheen and visible garnish, without target-like rings. Ingredient contact is weak, with several toppings appearing placed on the surface and little convincing immersion." }, "steam": { "score": 0.38, "reason": "Normal frames show a bright patch rising and expanding above the soup, supporting motion. It resembles a luminous blurred blob and broad haze rather than translucent irregular wisps, with an excessively bright core." }, "composition": { "score": 0.72, "reason": "A single centered dish, restrained dark background and warm food colors create a clean focal composition across desktop and mobile. The bright glow above the bowl competes with the food, while the lower bowl nearly disappears into darkness." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.67, "reason": "Continuous golden strands have rounded bends, highlights and readable overlaps, but the tightly stacked parallel arches feel arranged rather than naturally tangled and immersed." }, "toppings": { "score": 0.54, "reason": "Scored mushroom caps, tofu cubes, corn and scallion rings provide recognizable variety. The uniformly rectangular nori, fan-shaped leaf and smooth geometric tofu lack organic irregularity and food texture." }, "bowl": { "score": 0.83, "reason": "The elliptical rim, consistent inner cavity and smoothly tapered body form a coherent vessel with restrained edge highlights. No material construction defect is observed, though the very dark lower wall and contact area weaken material and support readability." }, "broth": { "score": 0.61, "reason": "Warm brown color, scattered sheen and small floating garnishes suggest soup. The surface has limited liquid depth, and several toppings appear laid on top with weak immersion cues." }, "steam": { "score": 0.38, "reason": "Normal frames show the bright vapor mass rising, expanding and dimming. However, it resembles a glowing blurred blob and broad haze rather than delicate irregular translucent wisps." }, "composition": { "score": 0.72, "reason": "A single readable dish sits against a restrained dark background with no clutter or text, and mobile framing preserves the bowl. The intense white glow above the rim competes with the food, while the bowl exterior is excessively dark." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, seaweed, greens, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.67, "reason": "Continuous golden strands have rounded bends, shading and readable overlap, but the tightly stacked parallel arches feel arranged and somewhat stiff rather than supple." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, tofu, greens and nori are recognizable and varied, but smooth blocks, oval caps, a fan-shaped leaf and a rigid rectangular nori sheet dominate; organic texture is limited." }, "bowl": { "score": 0.79, "reason": "The elliptical rim, inset cavity and curved walls form a coherent vessel with plausible thickness and a grounding shadow. Very dark walls obscure the material and base detail." }, "broth": { "score": 0.62, "reason": "Warm brown liquid has scattered sheen and garnish flecks, with ingredients enclosed by the surface. Limited menisci and immersion cues make several toppings look placed on a flat surface." }, "steam": { "score": 0.4, "reason": "Normal frames show the plume rising, expanding and fading, but it appears as a bright blurred glow or broad fog column rather than delicate irregular translucent wisps." }, "composition": { "score": 0.75, "reason": "A single readable dish is balanced against a restrained dark background, with warm food colors and no clutter. The intense white steam glow competes with the food, and the bowl exterior loses definition." }, "vegan": { "score": 1, "reason": "Visible tofu cubes, mushrooms, greens, nori and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5166666666666667, "toppings": 0.53, "bowl": 0.68, "broth": 0.6, "steam": 0.5333333333333333, "composition": 0.6533333333333333 }, "reward": 0.5771666666666667, "sampleStdDev": 0.009699656351300965, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are a small cluster of tan arcs with decent overlap and some curvature, but they look like repeated flat loops with a rigid, stylized feel and little volume or immersion." }, "toppings": { "score": 0.52, "reason": "Tofu cubes, a green leaf, nori with texture lines, shiitake with cross scoring and sesame/corn are all recognizable and varied. They are small, simple and somewhat flat, and the tofu cubes look blocky." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a hemispherical body, a rim highlight and a dark glaze. Perspective is consistent. It is very dark with little base or contact definition, so it looks slightly floating." }, "broth": { "score": 0.6, "reason": "The broth is a rich brown with warm glints, floating scallion bits and oil-like speckles, and it reads as liquid. It is fairly flat, with little depth or ingredient immersion." }, "steam": { "score": 0.6, "reason": "Steam changes across frames, from a bright glowing blob to a rising translucent column with wisps. Early frames show an overly bright blob that looks like a light flare rather than steam, and later plumes are soft, hazy and somewhat uniform." }, "composition": { "score": 0.66, "reason": "The bowl is centered in a dark, moody scene and is the clear focal point, with warm lighting and no clutter. The food is small, the bowl is very dark, and the early flare distracts." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, nori, leaf, scallions, sesame and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.5, "reason": "Noodles are a readable tan cluster of curved strands with some overlap, but they look like repeated arcs and are small and somewhat rigid, with limited volume and no real immersion in the broth." }, "toppings": { "score": 0.52, "reason": "Nori, shiitake with cross cuts, tofu cubes, a green leaf and scallion bits are all recognizable and vegan. The tofu cubes are blocky, the nori is a flat rectangle, and the items are small and sparse." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a hemispherical body and a subtle rim highlight. Perspective is consistent. It is very dark, has little ceramic detail, and lacks a visible base or strong contact shadow." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown with speckled oil and scallion dots and some depth. It reads slightly flat and muddy, and the toppings sit on the surface with little immersion." }, "steam": { "score": 0.5, "reason": "Steam rises and changes between frames. Early frames show a blown-out glowing orb, which reads as a light blob rather than wisps. Later frames show softer translucent plumes, but they are still somewhat uniform." }, "composition": { "score": 0.65, "reason": "A centered bowl in a dark atmospheric scene gives a clear focal point and warm lighting. The food is small and dim, and the early bright glow distracts." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, nori, leaf, scallions and noodles are visible. No animal ingredients appear." } }, { "noodles": { "score": 0.55, "reason": "Noodles form a readable cluster of curved, overlapping golden strands with some volume, but they look like repeated arcs and the mass is small and stiff, with little sense of sitting in the broth." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, a green leaf, shiitake with cross scoring, nori and scallion bits are recognizable and varied. They are small and fairly generic. The nori is a flat dark rectangle and the tofu cubes look blocky." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, inner wall, deep hemispherical body and a soft highlight on the rim. The dark body is a bit flat, the base is barely visible, and it seems to float without a clear contact shadow." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown with oily speckles and scattered scallion dots, which reads as miso-like soup. The surface is fairly flat, and the ingredients sit on it with little immersion or occlusion." }, "steam": { "score": 0.5, "reason": "Steam changes over time, from a bright blown-out glow in the first frames to a rising translucent column in later frames, so it is animated. The early glow is an opaque-looking blob that reads as a light flare. The later column is soft but fairly uniform." }, "composition": { "score": 0.65, "reason": "The bowl is centered, the dark moody atmosphere keeps the focus on the food, and the warm colors are pleasing. The food is small within the bowl, and the early bright flare is distracting." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, nori, leaf, scallion and noodles are all plant-based. No animal ingredients are visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7, "bowl": 0.8166666666666667, "broth": 0.7166666666666667, "steam": 0.7666666666666666, "composition": 0.85 }, "reward": 0.74, "sampleStdDev": 0.015612494995996018, "judgments": [ { "noodles": { "score": 0.65, "reason": "Strands are supple and curved with believable thickness, but they gather into one centralized clump with several repeated hook/loop shapes, and the nest reads as pasted atop the broth rather than emerging from it." }, "toppings": { "score": 0.7, "reason": "Good variety: striped nori sheet, scored shiitake ellipses, tofu cubes, corn kernels, scallion bits and chili flakes. Shapes are readable but stylized and flat; the green half-disc with radial lines is ambiguous, and mushrooms are simple flat ovals." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with a thin highlight and inner cavity shadow, smooth tapering dark ceramic body, plausible perspective and grounded contact shadow; only minor shortcoming is the near-black body hiding wall detail." }, "broth": { "score": 0.7, "reason": "Dark brown surface with oil droplets, pepper flecks and tonal variation; ingredients sit within the liquid with plausible contact, but the surface is fairly flat and murky with limited depth or sheen." }, "steam": { "score": 0.7, "reason": "Frames show the plume clearly evolving: low glow at rim, then a taller bright column, then a spreading fading wisp that dims over time \u2014 genuine rise and fade. Defect: the plume core becomes a near-opaque white blob (desktop-2) rather than translucent wisp." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a restrained dark atmospheric backdrop with warm glow, no clutter or competing elements; reads coherently on desktop and mobile, though the very dark palette leaves the dish slightly under-lit." }, "vegan": { "score": 0.85, "reason": "Visible toppings are tofu, mushrooms, nori, corn, scallion and chili \u2014 all plant-based. The green radially-striped half-disc is ambiguous (could read as a fish-cake) but is not unmistakably animal-derived." } }, { "noodles": { "score": 0.6, "reason": "Strands are continuous with varied bends and plausible thickness, but the mass reads as a tall, worm-like heap sitting on the broth rather than submerged strands; a few red strands look pasted in and the pile lacks believable immersion." }, "toppings": { "score": 0.7, "reason": "Recognizable variety: striated nori sheet, scored shiitake-style slices, tofu cubes, corn, scallion dots and chili flakes. Shapes are readable but flat-shaded and cutout-like at edges, with the green fan item somewhat ambiguous." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a lighter inner-wall band, tapering dark body, subtle rim highlight and grounded contact shadow; perspective is consistent, though the near-black body blends into the background and hides material detail." }, "broth": { "score": 0.75, "reason": "Dark brown surface with tonal swirls, floating chili flakes, scallion bits and soft shadows under toppings giving believable occlusion; slightly murky in areas and the sheen is mostly around the glowing center." }, "steam": { "score": 0.75, "reason": "Time-ordered frames show the plume rising, widening and dispersing (compact bloom at ~1s, taller column by ~4.8s, spread wisps later), so motion is supported; defect: the hot bright core reads more like a light bloom than translucent wisps." }, "composition": { "score": 0.85, "reason": "Centered dish as clear focal point against an atmospheric dark backdrop, warm amber palette, no text, UI or clutter, and consistent glow-based lighting; the large central bloom slightly competes with the food." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, nori, tofu cubes, mushrooms, greens, corn, scallions and chili in dark broth; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.7, "reason": "A coherent pale noodle mound sits centered in the broth with curved, hooking strands of plausible thickness and readable overlap. Slight repetition of similar curl shapes and some flattened loops keep it below the polished tier, but it reads clearly as noodles in the stylized illustration, not a maze scribble." }, "toppings": { "score": 0.7, "reason": "Good variety and recognizable forms: a textured dark nori sheet, white tofu cubes with simple 3D shading, brown scored mushroom caps, yellow corn, green scallion flecks and red chili bits. Shapes are somewhat simplified (flat mushroom ellipses, very clean cubes) but all read as food, not generic plastic stand-ins." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with visible wall thickness and a subtle lit edge, deep cavity, smoothly tapering walls to a rounded base, and a soft contact shadow beneath. Perspective between rim and body is consistent; the very dark body hides fine material detail but no distortion or detached parts are visible." }, "broth": { "score": 0.7, "reason": "Dark brown broth surface with appetizing scattered chili, scallion and corn specks, darker pockets, and a warm glow near the center. Ingredients sit in the liquid with contact shadows rather than floating above; the surface is a bit flat/gradient-driven and lacks fine ripple texture." }, "steam": { "score": 0.85, "reason": "Across the time-ordered desktop frames the plume clearly evolves: low bright core at 1s, taller rising column by ~5s, then a wider diffuse fading haze in the pointer frames \u2014 rising, spreading and fading continuously. It stays translucent and never hides the food; the reduced-motion frames are a matching static plume as expected." }, "composition": { "score": 0.85, "reason": "Bowl is the clear focal point, centered with balanced proportions against a dark atmospheric gradient, warm rim/steam lighting, and gentle depth. No text, UI, clutter or competing bowls; the scene remains restrained and readable. Only minor shortcoming is the very dark presentation muting the food's color appeal." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, nori, corn, scallions and chili in plant broth; no egg, meat, fish or dairy shapes are present." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9644444444444444, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "bef22bc54935e137eb18097c63bb154ebb1e70812ffdc3cbb020b6b04e31b5af", "gradedAt": "2026-09-30T22:53:17.044329+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5955555555555555, "toppings": 0.6833333333333333, "bowl": 0.7688888888888888, "broth": 0.6911111111111111, "steam": 0.8066666666666666, "composition": 0.8277777777777777 }, "reward": 0.7166111111111111, "runId": "anthropic/opus-5/max", "rewardStdDev": 0.039430787933251325, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.68, "bowl": 0.8633333333333333, "broth": 0.7, "steam": 0.8466666666666667, "composition": 0.8633333333333333 }, "reward": 0.726, "sampleStdDev": 0.0020000000000000018, "judgments": [ { "noodles": { "score": 0.5, "reason": "Individual golden strands and overlaps are readable, but the mound is dominated by stiff straight segments, angular bends and repeated wire-like loops rather than supple noodles." }, "toppings": { "score": 0.68, "reason": "Veined greens, scored mushroom caps, porous tofu, scallion rings and corn provide recognizable variety. The nori is an overly regular flat card, and the greens and mushrooms have simplified, rigid silhouettes." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, shaded inner cavity, rounded exterior and joined foot establish coherent depth and construction. Warm and cool edge highlights describe the material well; the lower support shadow is somewhat indistinct." }, "broth": { "score": 0.7, "reason": "Amber shading, small oil-like highlights and occlusion at the front rim read as soup. Ingredient contact is less convincing, with several toppings appearing placed on a smooth surface rather than visibly immersed." }, "steam": { "score": 0.84, "reason": "Normal time-ordered frames show changing translucent, irregular curls spreading above the soup without concealing the food. The broad cloud is somewhat oversized and diffuse, but its varying structure and fading edges are convincing." }, "composition": { "score": 0.86, "reason": "A single clearly readable dish anchors the dark scene, with restrained amber and blue lighting and no text or UI clutter. Desktop framing is balanced; mobile leaves a large empty upper region and brings the bowl close to the side edges." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Distinct golden strands have highlights and layered overlap, but the mound is dominated by stiff straight runs and angular, wire-like bends rather than supple noodles." }, "toppings": { "score": 0.68, "reason": "Veined greens, scored mushroom caps, porous tofu, scallion rings and corn provide readable variety. The nori is an overly regular card, and several vegetables have rigid, simplified silhouettes." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, deep inner wall, rounded body and joined foot establish convincing volume. Warm and cool edge highlights reinforce the ceramic form; the outer rim detailing is slightly overemphasized." }, "broth": { "score": 0.7, "reason": "Amber shading, small oil-like highlights and submerged ingredient edges suggest liquid depth. The surface remains somewhat uniformly opaque, with limited contact detail around the toppings." }, "steam": { "score": 0.86, "reason": "Normal frames show changing irregular translucent plumes extending from the soup and spreading into faint upper wisps. Food remains readable, though the broad cloud is somewhat dense and extends well beyond the bowl." }, "composition": { "score": 0.87, "reason": "The single dish is clearly framed against a restrained dark background, with warm food lighting and cool bowl edges creating separation. There is no distracting text or UI; the mobile layout leaves somewhat excessive space above the dish." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, scallions, corn, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Distinct golden strands have highlights and layered overlap, but the mound is dominated by stiff straight runs, angular bends and tight wire-like loops rather than supple noodles." }, "toppings": { "score": 0.68, "reason": "Veined greens, scored mushroom caps, porous tofu, scallion rings and corn provide readable variety. The nori is an overly regular flat rectangle, and the greens and mushrooms remain somewhat stiff and simplified." }, "bowl": { "score": 0.87, "reason": "A coherent elliptical rim, deep inner wall, rounded exterior and integrated foot establish convincing construction. Warm and cool edge highlights describe the ceramic volume; only the doubled side edging feels slightly overemphasized." }, "broth": { "score": 0.7, "reason": "Warm amber shading, scattered oil highlights and rim occlusion establish a readable liquid surface. Ingredient contact is less convincing, with little meniscus detail and several toppings appearing perched on the surface." }, "steam": { "score": 0.84, "reason": "Normal frames show changing irregular translucent tendrils rising from the bowl and spreading into diffuse upper wisps without hiding the food. The cloud is somewhat broad and smoky, but avoids identical mechanical plumes." }, "composition": { "score": 0.86, "reason": "The single dish is clearly framed against a restrained dark background, with warm food lighting balanced by cool rim highlights and ample space for steam. No distracting text or UI is visible; mobile framing leaves somewhat excessive empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6533333333333333, "toppings": 0.6466666666666667, "bowl": 0.6433333333333333, "broth": 0.66, "steam": 0.74, "composition": 0.7533333333333333 }, "reward": 0.6733333333333333, "sampleStdDev": 0.02829016319029169, "judgments": [ { "noodles": { "score": 0.62, "reason": "Noodles form a readable mound with overlapping curved strands and some volume. They look somewhat repetitive and flat, with stiff parallel loops, and they sit on top of the broth rather than blending into it." }, "toppings": { "score": 0.6, "reason": "Toppings are recognizable and varied: a leaf, bok choy stems, nori, bamboo shoots, shiitake with gill lines, tofu cubes and corn. The forms are simple and slightly flat, and the scallion rings look like repeated green dots." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent rim, inner wall and foot, with soft highlights. The outer handle-like rim lines look odd and the wall reads as a thin metallic outline, so the join between the rim and body looks slightly detached." }, "broth": { "score": 0.66, "reason": "The broth is a warm amber-brown with ripples, oil spots and some glow. Ingredients sit plausibly in it, though the surface is fairly flat and the contact edges are soft." }, "steam": { "score": 0.72, "reason": "Translucent wisps rise and change shape across the frames, spreading and fading. They are sometimes broad and hazy, but they do not hide the food." }, "composition": { "score": 0.74, "reason": "The bowl is a clear centered focal point in a dark, atmospheric scene with warm lighting and no distracting UI. The steam is large and the bowl sits slightly low in the frame." }, "vegan": { "score": 0.93, "reason": "Only plant ingredients are visible: tofu, mushrooms, nori, vegetables, corn and noodles. No animal products are seen." } }, { "noodles": { "score": 0.68, "reason": "Noodle mound has many supple, overlapping strands with believable bends and a readable dome volume. The strands are fairly uniform and slightly flat, with repeated loop patterns, and they sit somewhat on top of the broth." }, "toppings": { "score": 0.68, "reason": "Variety is readable: a veined leaf, nori sheet, bamboo shoots, two shiitake with gill lines, tofu cubes with sesame flecks, corn, and scallion rings. The nori is a rectangular block and the scallion rings look like uniform green donuts. Tofu cubes are a bit plain." }, "bowl": { "score": 0.65, "reason": "The bowl has a coherent rim, an orange inner wall, a foot and a grey glazed body with subtle highlights. The handle-like outline curves at the sides look odd, and the body reads slightly flat with some perspective mismatch between rim and body." }, "broth": { "score": 0.66, "reason": "The broth is a warm brown with chili-oil specks and faint ripples, with decent depth and contact at the noodles. It reads somewhat flat and matte, and the toppings sit partly on the surface rather than immersed." }, "steam": { "score": 0.78, "reason": "Translucent wispy steam rises above the bowl and changes shape across frames, drifting and fading softly. It is large and somewhat diffuse, and at times wide enough to feel like cloud backdrop, but it does not hide the food." }, "composition": { "score": 0.78, "reason": "A centered bowl sits in a dark atmospheric scene with warm lighting and a clear focal point. The palette is cohesive and there is no text or clutter. The bowl sits slightly low, and the steam is broad." }, "vegan": { "score": 0.93, "reason": "Tofu, shiitake, bamboo, leaf, nori, corn and scallions are all plant-based. No meat or egg is visible." } }, { "noodles": { "score": 0.66, "reason": "Noodles form a readable mound of wavy, overlapping strands with a soft tan colour and some depth. The strands are fairly uniform and flat-looking, the dome is stiff, and the scribbly tangle is somewhat repetitive." }, "toppings": { "score": 0.66, "reason": "Toppings are varied and recognizable: a veined leaf, dark nori, shiitake with gill lines, tofu cubes, corn and bamboo shoots. The tofu cubes are blocky, the nori is a rectangle, and the scallion rings look like uniform green eyes or donuts, so the set feels symbolic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, an orange inner wall, a tapering body and a foot with a rim highlight. The side handle-like lines are odd, and the rim edge and body join slightly, but the bowl still reads as solid ceramic." }, "broth": { "score": 0.66, "reason": "The broth is a warm brown with soft ripples and small oil spots, and the noodles and toppings sit in it plausibly. It is somewhat flat and murky, with little sheen or depth." }, "steam": { "score": 0.72, "reason": "Steam is translucent and wispy, and it changes shape and position across frames, rising and fading. It is quite large and diffuse, and in the first frame it is rather dense." }, "composition": { "score": 0.74, "reason": "A centered bowl on a dark atmospheric background has warm lighting and good colour harmony, and the food remains the focus. The steam cloud is big, and the bowl is cropped close to the bottom." }, "vegan": { "score": 0.93, "reason": "Only plant-based ingredients are visible: tofu, mushrooms, nori, vegetables, corn and noodles. There is no meat or egg." } } ], "diagnostics": { "vegan": 0.93 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6333333333333333, "toppings": 0.7233333333333333, "bowl": 0.8, "broth": 0.7133333333333333, "steam": 0.8333333333333334, "composition": 0.8666666666666667 }, "reward": 0.7505000000000001, "sampleStdDev": 0.006383572667401828, "judgments": [ { "noodles": { "score": 0.68, "reason": "Curved tan strands form a readable central mound with overlapping bends, drape over the pile and scallion rings on top; some strands look slightly wiry/flat-ribboned and a few ends fade indistinctly into the tangle." }, "toppings": { "score": 0.72, "reason": "Good variety: leafy green with vein detail, nori sheet, tofu cubes, bamboo shoots, corn kernels, scallion rings and cut shiitake-style caps; mushrooms read slightly cookie-like and corn cluster is a bit blobby, but shapes are organic and scaled plausibly." }, "bowl": { "score": 0.8, "reason": "Coherent wide rim ellipse, orange lacquer interior wall, dark glazed exterior with warm rim light on the right and cool accent on the left rim, plus a supported foot; perspective joins are consistent with only minor highlight ambiguity." }, "broth": { "score": 0.7, "reason": "Dark miso-toned liquid with warm central glow, chili-oil flecks and pale bubble specks; ingredients sit with contact shadows, though a few toppings (tofu cubes, corn) rest slightly above the surface rather than clearly immersed." }, "steam": { "score": 0.75, "reason": "Translucent wisps rise from the bowl and visibly reshape between desktop frames (V-plumes drift right under pointer-right, left under pointer-left) while reduced frames hold static, confirming animation; steam is faintly fog-like and dense high above but never hides the food." }, "composition": { "score": 0.85, "reason": "Centered focal bowl, warm orange interior against a dark atmospheric backdrop with subtle dust particles and balanced lighting; no text, UI or clutter competes with the dish." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, corn, scallions, bamboo shoots, leafy greens and nori in a dark broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.6, "reason": "Noodles form a coherent central mound with overlapping curved strands and a few draping over the left side, but they render as fairly uniform, flat ribbon-like strands with little thickness variation or volume \u2014 readable as a stylized noodle nest yet stiff and somewhat scribble-like rather than supple." }, "toppings": { "score": 0.7, "reason": "Good variety: bok choy with visible leaf veins and pale stems, two brown mushroom caps with radial gill patterns, cream tofu cubes, a dark nori sheet, bamboo shoot chunks, corn kernels and scattered scallion rings. Shapes are mostly organic and readable, though tofu cubes are plain, corn reads slightly bean-like, and mushroom caps are flat top-view disks rather than volumetric." }, "bowl": { "score": 0.8, "reason": "Coherent wide ceramic bowl: elliptical rim, orange inner wall, tapering dark body and a centered foot, with warm interior glow and exterior highlights. Perspective is consistent across all frames; the only minor flaw is the pale sliver on the left exterior that reads slightly like a detached highlight." }, "broth": { "score": 0.72, "reason": "Appetizing brown-orange miso-like surface with depth gradient, chili-flake and oil droplet specks, and believable contact shadows under tofu, mushrooms and corn. Faint concentric ripple arcs near the left edge look slightly mechanical, but immersion and occlusion are otherwise convincing." }, "steam": { "score": 0.9, "reason": "Steam clearly animates across frames: the V-shaped plumes at ~1s reshape and disperse by ~6s, sweeping right after the right-pointer frame and left after the left-pointer frame. Wisps are translucent, irregular, and never obscure the food \u2014 continuous, responsive, non-static behavior with no opaque fog." }, "composition": { "score": 0.9, "reason": "Well-balanced dark scene: the bowl is the clear focal point, warm backlight glow and subtle bokeh particles create depth, no text, UI or clutter, and the mobile viewport crops gracefully with the dish still centered and readable. Attractive restrained lighting supports rather than replaces the food." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: noodles, tofu, mushrooms, bok choy, corn, bamboo shoots, scallions, nori and vegetable-looking broth; no eggs, meat, fish or dairy appear anywhere in the frames." } }, { "noodles": { "score": 0.62, "reason": "Coherent pale noodle mound with shaded strands and visible ends, but the strands read fairly straight and rigid, more like a spaghetti pile than supple ramen, and they sit atop the broth with little believable immersion." }, "toppings": { "score": 0.75, "reason": "Good variety: bok choy leaf with veins and stalks, dark nori slab, scored shiitake domes, speckled tofu cubes, corn kernels, bamboo shoots and scallion rings. Most read as food, though mushrooms and corn are somewhat simplified blob/stylized forms." }, "bowl": { "score": 0.8, "reason": "Coherent ellipse rim with orange inner wall, consistent exterior body, footed base and grounded contact shadow; perspective holds across frames. Minor banding on the left rim highlight and slightly heavy edge lines." }, "broth": { "score": 0.72, "reason": "Warm brown liquid surface with oil specks, floating droplets, gentle ripple rings and occlusion shadows under toppings; it reads liquid but is somewhat flat and murky in the rear half." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps above the bowl that clearly change shape across desktop frames and lean right/left following the pointer frames; reduced frames freeze appropriately without opaque fog hiding the food." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a dark atmospheric scene with warm backlight, drifting embers and restrained palette; mobile crop scales cleanly with no text, UI or clutter distracting from the dish." }, "vegan": { "score": 1, "reason": "All visible toppings are plant-derived: bok choy, nori, mushrooms, tofu, corn, bamboo shoots and scallions in a dark plant-style broth; no animal ingredients visible." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9766666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 133.40000000000146, "frameIntervalP95Ms": 150, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "b9a6f65f7faf5251ed483bc190cc64e3624f16acfabe63501a239a4cc73c715c", "gradedAt": "2026-09-30T22:54:48.992826+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5233333333333333, "toppings": 0.62, "bowl": 0.7555555555555555, "broth": 0.6366666666666666, "steam": 0.6511111111111111, "composition": 0.7955555555555556 }, "reward": 0.6516666666666666, "runId": "anthropic/opus-5/medium", "rewardStdDev": 0.08528530549475297, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48666666666666664, "toppings": 0.61, "bowl": 0.8133333333333334, "broth": 0.5833333333333333, "steam": 0.62, "composition": 0.85 }, "reward": 0.6488333333333334, "sampleStdDev": 0.0070237691685684405, "judgments": [ { "noodles": { "score": 0.5, "reason": "Cream-colored strands are continuous and overlap toppings, but the sparse, nearly parallel broad arcs look like stiff ribbons rather than a supple noodle nest." }, "toppings": { "score": 0.61, "reason": "Veined green leaves, scored mushroom caps, tofu cubes and scallion rings provide readable variety. Textures remain schematic, and the straight-edged nori looks like a rigid card." }, "bowl": { "score": 0.83, "reason": "A coherent elliptical rim, visible lip thickness and smoothly shaded curved walls convincingly describe a dark lacquer bowl. The lower edge and contact with the surface are weakly defined." }, "broth": { "score": 0.59, "reason": "Warm brown liquid surrounds the ingredients with some oily patches and seasoning flecks. Limited surface reflections and immersion cues make it look fairly flat, with several garnishes appearing pasted on." }, "steam": { "score": 0.62, "reason": "Translucent haze changes shape and position above the soup across normal frames without substantially obscuring it. The broad blurry puffs lack distinct rising, tapering wisps and clearly readable dissipation." }, "composition": { "score": 0.85, "reason": "The centered dish remains the clear focus, with warm broth and green toppings contrasted against a restrained dark background and no UI or text. Mobile framing preserves the whole bowl; its very dark lower body slightly weakens separation." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, scallions and corn plausibly form vegan ramen; no animal-derived topping is discernible." } }, { "noodles": { "score": 0.48, "reason": "Continuous pale strands are readable, but the sparse, similarly arched strips look rigid and laid flat rather than supple, tangled noodles with volume." }, "toppings": { "score": 0.62, "reason": "Veined greens, scored mushroom caps, tofu cubes and scallion rings provide recognizable variety. Shapes remain simplified and smooth, especially the straight-edged nori cards and disk-like mushrooms." }, "bowl": { "score": 0.82, "reason": "Nested elliptical rim edges align coherently with the rounded walls, and warm highlights convey dark lacquer and thickness. The very dark underside leaves its base and surface contact weakly defined." }, "broth": { "score": 0.58, "reason": "Warm brown coloring and scattered oil-like patches suggest soup, but the surface is mostly flat and opaque, with limited sheen or immersion cues around the ingredients." }, "steam": { "score": 0.61, "reason": "Translucent haze changes shape and position above the soup across normal frames without obscuring the food. It reads more as broad blurred clouds than distinct rising, spreading wisps." }, "composition": { "score": 0.85, "reason": "A single centered dish remains the clear focus, with warm amber lighting, green accents and restrained dark surroundings free of text or UI. Mobile framing preserves the whole bowl; the lower bowl is slightly lost in shadow." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, scallions and corn plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.48, "reason": "Cream-colored strands are continuous and overlap toppings, but mostly form a sparse stack of broad, parallel arcs with little supple tangling or noodle volume." }, "toppings": { "score": 0.6, "reason": "Veined green leaves, scored mushroom caps, tofu cubes and scallion rings provide readable variety. However, mushrooms resemble smooth disks, tofu lacks texture, and nori is an overly straight, rigid card." }, "bowl": { "score": 0.79, "reason": "Nested elliptical rim, rounded walls and restrained warm highlights create coherent depth and a convincing dark ceramic silhouette. The bottom and surface contact are indistinct, giving slightly floating support." }, "broth": { "score": 0.58, "reason": "Warm brown color, scattered seasoning and subtle tonal patches suggest soup, but the surface is rather flat and opaque, with limited liquid sheen or convincing immersion around toppings." }, "steam": { "score": 0.63, "reason": "Normal frames show translucent patches changing shape and position above the soup without obscuring it. They read more as soft glowing clouds than irregular rising wisps, with limited visible upward development." }, "composition": { "score": 0.85, "reason": "A single well-framed dish remains the clear focus against a restrained dark background; warm amber light complements green and cream ingredients, and mobile framing stays intact. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn-like kernels, seaweed and noodles plausibly form vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4666666666666667, "toppings": 0.5333333333333333, "bowl": 0.6533333333333333, "broth": 0.5933333333333333, "steam": 0.55, "composition": 0.6866666666666666 }, "reward": 0.5678333333333333, "sampleStdDev": 0.022678918257565423, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are a few flat, pale ribbon-like strands fanned in one direction. They are readable as noodles but sparse, rigid and uniform, with little volume or overlap, and they barely sit in the broth." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake with cross scoring, a bok choy leaf, a nori sheet, scallion rings and corn-like dots are all recognizable and varied. The nori is a stiff dark rectangle and the tofu is blocky. The scallion and corn dots are small, repetitive and read as generic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved dark body with a warm rim highlight and a subtle band, and consistent perspective. It is very dark with little material detail and the base is lost in shadow, so it looks slightly flat." }, "broth": { "score": 0.6, "reason": "The broth is a warm reddish-brown with orange gradients and some depth, and the toppings sit reasonably in it. It is fairly flat, with little sheen or texture, and the immersion contact is weak." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and shift shape between frames, and in the pointer frames they drift sideways. They are a fairly diffuse blob-like haze and not very wispy, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "A single centered bowl in a dark, atmospheric scene with a warm glow and good color contrast from the greens. The food is the focus and there is no clutter. The bowl is somewhat dim and large empty dark areas surround it." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, bok choy, nori, scallions and corn. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.45, "reason": "Noodles are a few flat, smooth cream ribbons fanning across the soup. They are rigid and sparse, with little overlap or volume, and they read as stripes rather than supple strands." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, shiitake with cross-cuts, a bok choy leaf, nori, scallion rings and corn are all recognizable and varied. The nori is a stiff dark rectangle, the tofu is blocky, and the scallion and corn bits look like flat plastic dots, so the set stays generic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim with a double lip, a rounded body and a soft highlight band, and it sits on a plausible base. It is very dark and low in detail, with little visible material texture or ground contact." }, "broth": { "score": 0.58, "reason": "The broth is a warm brown miso with orange swirls and reads as liquid. It is fairly flat, and the toppings look somewhat pasted on top with weak immersion and few reflections." }, "steam": { "score": 0.55, "reason": "Soft translucent wisps rise above the bowl and change shape between frames. They are blurry and blobby, slightly static in form, and respond to the pointer." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point with warm glow and a dark atmospheric background and no clutter. The bowl body is murky, and the food occupies a small share of the frame." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, corn, nori and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.45, "reason": "Noodles are a few flat, rigid cream ribbons fanning out in parallel. They are smooth and readable, but too few, too uniform and lacking volume or tangle, and they look pasted on the surface." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake with cross scoring, a bok choy leaf, a nori sheet, scallion rings and corn kernels are all recognizable with some variety. The nori is a stiff dark rectangle and the scallions and corn are generic flat discs scattered evenly." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a curved body and a subtle highlight band, and the perspective is consistent. It is very dark and flat, with little visible ceramic material, and the base and contact shadow are barely readable." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown-orange with a reddish gradient and chili flecks, so it reads as miso-like liquid. It has little surface sheen or depth, and the ingredients sit on top more than in it." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps rise above the bowl and shift between frames, and they respond to the pointer. They look like blurry blobs, though, and the reduced-motion frames are static." }, "composition": { "score": 0.68, "reason": "The bowl is centered as the clear focal point against a dark, warm vignette with no clutter. The lower half of the bowl is murky and the food is fairly small in the frame." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, seaweed, corn and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6166666666666667, "toppings": 0.7166666666666667, "bowl": 0.8, "broth": 0.7333333333333333, "steam": 0.7833333333333333, "composition": 0.85 }, "reward": 0.7383333333333334, "sampleStdDev": 0.01876388374866288, "judgments": [ { "noodles": { "score": 0.65, "reason": "Pale cream strands with gentle bends and plausible thickness arc over the broth, but they read as flat tapered ribbons lying on the surface with no immersion shadow, and the noodle mass is sparse (about six visible strands) rather than a convincing clustered tangle." }, "toppings": { "score": 0.75, "reason": "Good variety: nori sheet with streak texture, mushroom slices with cut marks, leafy greens with veins, tofu cubes, scallion rings, corn, chili and sesame specks. Shapes are readable and appropriately scaled, though the tofu cubes are very crisp perfect geometry and the nori leans like a tent card \u2014 stylized but slightly card-like." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with inner-wall thickness, consistent perspective between rim and body, smooth curved walls to a rounded base, subtle sheen band and rim highlight, plus a soft contact shadow below. Minor shortcoming: the body is very dark and material detail is minimal, but no distortion or broken joins." }, "broth": { "score": 0.75, "reason": "Rich brown liquid surface with lighter swirl streaks, oil droplets, a meniscus-like lighter edge, and restrained sheen; nori base and mushrooms are occluded and toppings sit in the liquid. The swirl patterning is slightly uniform/decorative, but it reads as liquid, not solid." }, "steam": { "score": 0.8, "reason": "Frames show changing translucent wisps: faint haze at 1.0s, defined rising plumes at 2.8s and 4.2s with shifted shapes, dimmed on the right-pointer frame and concentrated leftward on the left-pointer frame \u2014 supported continuous rise/spread motion. Wisps stay irregular and never hide the food; only minor blobbiness in the diffuse haze." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark atmospheric backdrop with a warm glow, floating dust motes and vignette; balanced proportions and coherent lighting across desktop and mobile crops, with no text, UI or clutter to distract from the dish." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, nori, leafy greens, scallions, corn, chili and sesame in plant broth \u2014 all plausibly vegan with no animal-derived toppings; the small yellow garnish dots match the corn scattered elsewhere." } }, { "noodles": { "score": 0.65, "reason": "Pale curved strands are readable as noodles with gentle bends and some overlap, but they are sparse, lie flat on the surface, and repeat similar arc shapes with little volume or immersion into the broth." }, "toppings": { "score": 0.7, "reason": "Good variety: shaded tofu cubes, a folded nori sheet, mushroom slices with gill markings, bok-choy-like greens, scallion rings, corn and chili flakes. Shapes are recognizable though somewhat flat and evenly scattered; mushroom slices read slightly cookie-like." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a warm highlight ring, smooth tapering dark ceramic body and plausible base; perspective is consistent across frames. Only minor issue is low wall contrast against the dark background and a soft ground contact." }, "broth": { "score": 0.75, "reason": "Rich brown liquid surface with tonal variation, restrained edge sheen and scattered specks; ingredients mostly sit in the broth with subtle occlusion, though some toppings look lightly pasted on the surface." }, "steam": { "score": 0.75, "reason": "Across desktop-1/2/3 the translucent wisps change height and shape (short at 1.0s, taller plumes at 2.8s and 4.2s) and drift left in the pointer-left frame, indicating continuous rising motion that never obscures the food; wisps are slightly blobby at their tops." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a dark atmospheric scene with a warm back glow and faint particles, appetizing warm palette, no text or clutter; readable at both desktop and mobile sizes. Glow is restrained and does not substitute for food quality." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, nori, leafy greens, scallions, corn and chili in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.55, "reason": "A handful of cream-colored strands curve across the broth and emerge near the nori and greens, so they read as noodles, but they are few, flat ribbon-like arcs with uniform thickness and little volume or nesting; the surface still looks sparsely nought rather than a full noodle bed." }, "toppings": { "score": 0.7, "reason": "Recognizable variety: shaded tofu cubes, a dark nori sheet with texture lines, veined green leaves, brown mushroom slices with cross-hatch, scallion rings and corn. Weaknesses: the scallion/corn pieces read as repetitive scattered rings and dots, and shapes are simple flat vector forms rather than richly textured food." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a light inner lip, deep rounded dark ceramic body with soft highlight sweep and believable perspective; the bowl sits grounded with contact shadow. No visible distortion; only shortcoming is the base fading into darkness with no distinct foot." }, "broth": { "score": 0.7, "reason": "Brown liquid surface with swirls, oil droplets, chili flakes and sesame specks gives believable depth; toppings mostly sit in the surface with subtle shadows. Slight murkiness and a somewhat mottled, painted texture keep it from higher marks." }, "steam": { "score": 0.8, "reason": "Across the time-ordered desktop frames the steam changes clearly: small wisps at 1038ms grow into taller spreading plumes by 2780ms and 4202ms, then shift/dim in the pointer frames, showing continuous animated motion. It stays translucent and irregular without hiding the food; it is somewhat diffuse and blobby at times." }, "composition": { "score": 0.85, "reason": "Balanced centered bowl as clear focal point, warm rim lighting against an atmospheric dark background with faint particles, no clutter or UI. Mobile frame keeps the dish legible though it occupies a small share of the tall viewport." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, nori, mushrooms, greens, scallions and corn in plant broth; no animal-derived ingredients are depicted." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.399999999999636, "frameIntervalP95Ms": 66.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "3ce2070ce18f97fc765ea1dcaafe33823b65a18af149449416370fd36dc020b7", "gradedAt": "2026-09-30T22:53:51.080726+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.7133333333333334, "toppings": 0.6822222222222222, "bowl": 0.8155555555555556, "broth": 0.7022222222222222, "steam": 0.7555555555555555, "composition": 0.8288888888888889 }, "reward": 0.7427777777777778, "runId": "anthropic/opus-5/xhigh", "rewardStdDev": 0.05733365632983951, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.76, "toppings": 0.7133333333333333, "bowl": 0.9, "broth": 0.73, "steam": 0.7733333333333333, "composition": 0.8766666666666667 }, "reward": 0.787, "sampleStdDev": 0.0039051248379533806, "judgments": [ { "noodles": { "score": 0.76, "reason": "Continuous golden strands have varied bends, highlights and readable overlaps forming a mound. Some sharply bent, uniformly outlined strands feel stiff rather than supple." }, "toppings": { "score": 0.7, "reason": "Scored mushroom caps, veined greens, scallion rings and textured tofu provide readable variety. Leaves are rather flat and blade-like, and the gridded rectangular nori looks unusually rigid and regular." }, "bowl": { "score": 0.9, "reason": "Coherent elliptical rim, visible inner wall and rounded body establish convincing depth; warm rim highlights and a cool edge accent suggest dark ceramic. Ground shadow supports the base, with no material construction defect observed." }, "broth": { "score": 0.73, "reason": "Warm amber shading and scattered small oil-like rings read as soup, with ingredients contained within the surface. Limited meniscus and immersion cues make several toppings appear laid on top." }, "steam": { "score": 0.77, "reason": "Normal frames show changing translucent clouds that spread above the bowl without obscuring the food. The broad, softly glowing puffs lack finer rising wisps and have a slightly fog-like appearance." }, "composition": { "score": 0.87, "reason": "A single readable dish dominates the uncluttered dark scene, with coordinated amber highlights and green accents. No text or UI competes with the food; the mobile framing leaves somewhat excessive upper empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, scallions, corn and nori with noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.76, "reason": "Continuous golden strands have varied bends, highlights and readable overlaps forming a mound. Some angular turns and uniformly outlined strands feel slightly stiff rather than supple." }, "toppings": { "score": 0.72, "reason": "Scored mushroom caps, veined greens, scallion rings and textured tofu provide readable variety. Leaves are unusually smooth and spear-like, and the rectangular gridded nori looks more like a rigid panel than thin irregular seaweed." }, "bowl": { "score": 0.9, "reason": "Coherent elliptical rim, recessed cavity and smoothly rounded walls establish convincing depth. Warm rim reflections, a cool edge highlight and grounding shadow support the dark ceramic appearance; no material construction defect is observed." }, "broth": { "score": 0.74, "reason": "Amber shading, small oil-like glints and bubbles create an appetizing liquid surface. Ingredient contact is readable, but limited menisci and submerged detail make parts of the topping arrangement feel laid onto the surface." }, "steam": { "score": 0.77, "reason": "Normal frames show changing translucent clouds spreading above the bowl without concealing the food. The steam fades softly, but broad glowing patches dominate over finer connected wisps emerging directly from the soup." }, "composition": { "score": 0.88, "reason": "The single dish remains a clear focal point against a restrained dark background, with harmonious amber and green accents and coherent warm lighting. Both viewport sizes preserve the bowl without clutter; mobile has somewhat excessive upper negative space." }, "vegan": { "score": 1, "reason": "Visible noodles, mushrooms, greens, scallions, seaweed and tofu plausibly form vegan ramen, with no visibly animal-derived ingredients." } }, { "noodles": { "score": 0.76, "reason": "Continuous golden strands have varied bends, clear overlaps and shaded volume. Some parallel runs and blunt ends look stiff, and the mound has limited visible immersion." }, "toppings": { "score": 0.72, "reason": "Veined greens, scored mushroom caps, scallion rings and textured tofu provide readable variety. Leaves are unusually smooth and spear-like, while the nori is a regular rigid rectangle rather than irregular seaweed." }, "bowl": { "score": 0.9, "reason": "The elliptical rim, recessed cavity and rounded walls share coherent perspective, with convincing thickness, warm rim highlights and a cool outer edge. Ground shadow supports the base; no material construction defect is observed." }, "broth": { "score": 0.72, "reason": "Warm brown liquid has subtle sheen, small oil-like flecks and readable depth inside the rim. Ingredient contacts lack strong menisci or partial submersion, making some food look placed on the surface." }, "steam": { "score": 0.78, "reason": "Normal frames show changing translucent clouds spreading above the soup without obscuring the ingredients. The steam is broad and diffuse, with somewhat detached glowing patches rather than clearly traced rising wisps." }, "composition": { "score": 0.88, "reason": "A single readable dish anchors the uncluttered dark scene, with warm food highlights and contrasting green accents. Desktop and mobile retain the complete bowl; generous upper negative space slightly weakens the balance." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, mushrooms, tofu, scallions and seaweed plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.68, "toppings": 0.6333333333333333, "bowl": 0.73, "broth": 0.6266666666666667, "steam": 0.66, "composition": 0.76 }, "reward": 0.678, "sampleStdDev": 0.00793725393319378, "judgments": [ { "noodles": { "score": 0.68, "reason": "Noodles are supple, overlapping, curved strands with believable thickness and a few golden corn-like dots on top. The mass is somewhat tangled and loop-repetitive, and it sits a bit on top of the broth." }, "toppings": { "score": 0.62, "reason": "Shiitake with cross-scoring, bok choy leaves, tofu cubes, scallion rings and a nori sheet are all readable and varied. The tofu looks blocky, the nori is a stiff grid-textured card, and the leaves are somewhat vector-flat." }, "bowl": { "score": 0.75, "reason": "The bowl has a coherent elliptical rim, inner wall, rounded body and a copper rim highlight, with consistent perspective. The base is hardly visible and the wall is fairly dark and flat, but no distortion is seen." }, "broth": { "score": 0.66, "reason": "The broth is a warm amber-brown with small oil droplets and sesame flecks, and it reads as liquid. Immersion and contact with the ingredients are limited, and the surface is fairly flat." }, "steam": { "score": 0.68, "reason": "Soft translucent steam rises above the bowl, and it changes shape between frames and drifts with the pointer. It reads a bit as a diffuse glowing cloud rather than distinct wisps, but it does not hide the food." }, "composition": { "score": 0.78, "reason": "The bowl is centered in a dark, warm atmosphere with good color harmony and a clear focal dish. There is no clutter or text. The bowl sits somewhat low in the frame, leaving empty space above." }, "vegan": { "score": 0.93, "reason": "Visible ingredients are noodles, mushrooms, tofu, greens, scallions and nori, with no egg, meat or fish. The yellow dots are ambiguous but read as corn." } }, { "noodles": { "score": 0.68, "reason": "Noodles are supple, overlapping strands with believable bends and a readable volume in the broth. Thickness is fairly uniform and the loops look somewhat repetitive, so they read as a stylized tangle." }, "toppings": { "score": 0.66, "reason": "Shiitake with scored crosses, bok choy leaves, tofu cubes, a nori sheet, scallion rings and enoki-like nubs all read clearly and offer good variety. The nori looks like a grid-textured blue-gray card, the tofu is blocky, and the enoki bundle is ambiguous." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim with a copper highlight, a visible inner wall, and a rounded body with a subtle blue sheen. Perspective is consistent. The base is faint and dissolves into the dark background, so there is little visible contact or support." }, "broth": { "score": 0.62, "reason": "The warm brown broth has depth, small oil bubbles and sesame specks, and the ingredients sit in it reasonably. The surface is fairly flat and a bit murky, with little sheen variation." }, "steam": { "score": 0.62, "reason": "Soft translucent steam rises above the bowl and changes shape between frames, drifting and fading without hiding the food. It is a diffuse glow-like cloud rather than distinct wisps, and it is not clearly anchored to the soup surface." }, "composition": { "score": 0.75, "reason": "The bowl is centered as a clear focal dish in a dark, moody scene with warm lighting and no distracting text or clutter. The palette is cohesive. The bowl sits somewhat low, leaving large empty dark space above." }, "vegan": { "score": 0.93, "reason": "Visible ingredients are noodles, mushrooms, tofu, greens, nori and scallions, with no meat, egg or fish." } }, { "noodles": { "score": 0.68, "reason": "Noodles are supple, overlapping and curved with believable thickness, sitting in a coherent mound. They look somewhat tangled and uniform in color, with a few flat loops, and the enoki-like bundle at right is simple." }, "toppings": { "score": 0.62, "reason": "Shiitake with scored crosses, tofu cubes, leek leaves, scallion rings and a nori sheet are all readable. The nori reads as a stiff gridded card standing upright, the tofu is blocky, and the leaves are fairly flat. Variety is good." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, inner wall, rounded body and rim highlight, plus a soft contact shadow. It looks a little dark and flat, and the base is ambiguous, but no major distortion." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown with small oil droplets and a gradient toward the rim that gives some depth. It looks somewhat flat and opaque, and some toppings sit on top more than in it." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps rise above the bowl and change shape and position across frames, fading without hiding the food. They are fairly diffuse and blobby, and they look somewhat detached from the soup surface." }, "composition": { "score": 0.75, "reason": "The bowl is a centered focal point in a dark, warm atmosphere, with a pleasing palette and no clutter or text. The bowl is slightly low in the frame and there is a lot of empty space." }, "vegan": { "score": 0.93, "reason": "Only mushrooms, tofu, noodles, leek, scallions and nori are visible, with no meat or egg. The yellow dots are ambiguous but read as corn or oil." } } ], "diagnostics": { "vegan": 0.93 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7000000000000001, "toppings": 0.7, "bowl": 0.8166666666666667, "broth": 0.75, "steam": 0.8333333333333334, "composition": 0.85 }, "reward": 0.7633333333333333, "sampleStdDev": 0.040722639076235384, "judgments": [ { "noodles": { "score": 0.65, "reason": "Central tangle of pale yellow strands with smooth bends, overlaps and a few draping ends reads as noodles; however strands are flat ribbons of near-uniform width with a glossy uniform color and repetitive loop shapes, lacking broth-soaked depth." }, "toppings": { "score": 0.65, "reason": "Good variety: scored brown mushroom slices, tofu cubes, a standing nori sheet, green leaves, scallion rings, corn, sesame and a sprout/enoki bundle. Defects: leaves are glossy and plastic-like, the nori reads as a rigid blue-gray card with tile-like grid lines, and the mushroom discs are flat ovals." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: clean rim ellipse, visible inner wall band, smooth body and base with a soft contact shadow, plus a tasteful warm rim highlight. Perspective is consistent across all frames; only minor thinness of the near wall keeps it from higher." }, "broth": { "score": 0.7, "reason": "Amber-brown liquid surface with bubbles, oil flecks and small ripples reads as broth, and toppings sit at the surface with some contact shading. The surface is fairly flat in tone and the noodle mass floats on top without much immersion integration." }, "steam": { "score": 0.85, "reason": "Normal frames at 1106ms, 3583ms, 5871ms and pointer frames show translucent irregular wisps that rise, spread, drift left and fade, changing shape continuously without obscuring the food; plumes are slightly blobby at their bright cores." }, "composition": { "score": 0.85, "reason": "Centered dish as clear focal point against a dark atmospheric background with warm rim lighting; no text, UI or clutter; mobile crop keeps the bowl readable. Background glow supports but does not replace the food." }, "vegan": { "score": 0.9, "reason": "Visible toppings are tofu, mushrooms, nori, leafy greens, scallion, corn and sprouts in plant broth. The brown cross-hatched discs are ambiguous (could be read as fish cake) but most plausibly scored mushroom caps, so no clear animal ingredient." } }, { "noodles": { "score": 0.8, "reason": "Golden wavy strands heap convincingly in the center with overlapping volume, varied bends, highlight/shadow shading, and believable seating in the broth. Minor flaws: several strands repeat a similar S-curve so the mass reads slightly patterned, and the tan bundle at the right edge is ambiguous (noodle cluster vs enoki)." }, "toppings": { "score": 0.75, "reason": "Good variety and recognizability: a standing nori sheet with grid texture, leafy greens with veins, cross-scored cap-like discs, tofu cubes, corn kernels, scallion rings, sesame and an enoki-like bundle, all plausible in scale. Defects: leaves are fairly flat/graphic with a plasticky sheen, and the tofu cubes are somewhat rigid geometric blocks with box-like creases rather than soft organic curd." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic/lacquer bowl: clean elliptical rim with a copper inner line and highlight arc, consistent inner-wall band, tapered body with a soft curved highlight, and grounded contact shadow on the surface. Perspective between rim ellipse and body is consistent; no broken joins or distortion observed." }, "broth": { "score": 0.8, "reason": "Appetizing brown liquid surface with restrained gloss, floating oil specks, bubbles, chili flecks and scallion rings; ingredients show occlusion shadows and sit in the liquid rather than pasted above. Slight flatness in the mid-broth area and a somewhat uniform sheen keep it from being exceptional." }, "steam": { "score": 0.85, "reason": "Time-ordered desktop frames show the translucent warm plume changing shape, rising and drifting (leaning left in the pointer-right frame), so it is genuinely animated and continuously fades without hiding the food. Wisps are somewhat diffuse/foggy near the top, slightly less defined than ideal; reduced-motion frames correctly freeze." }, "composition": { "score": 0.85, "reason": "Balanced centered dish against a dark atmospheric backdrop with warm rim light and steam adding depth; harmonious brown/gold/green palette, no text, UI or clutter, and the mobile frame crops gracefully. Food remains the clear focal point." }, "vegan": { "score": 0.9, "reason": "Visible ingredients (noodles, nori, leafy greens, tofu, corn, scallions, sesame, enoki-like bundle, brown cross-scored caps) plausibly form a vegan bowl with no egg, meat or dairy; the brown cross-marked discs read most like scored shiitake caps, with slight ambiguity versus fish-cake, hence not full credit." } }, { "noodles": { "score": 0.65, "reason": "Golden strands form a coherent central tangle with curved bends and plausible placement on the broth, but many strands read as flat ribbons with repetitive wave loops and abrupt ends; a separate parallel-line bundle at the right reads more like stacked strokes than noodles." }, "toppings": { "score": 0.7, "reason": "Good variety: a readable nori sheet, a stamped tofu cube, scallion rings, corn kernels, sesame, and two leafy greens. Defects: the leaves are quite flat and lens/pod-like, and the two brown cross-patterned discs are ambiguous (mushroom vs naruto-style fish cake); everything is stylized but readable." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a lit lip, visible inner cavity and wall, dark ceramic body with a soft sheen, and grounded contact shadow. Perspective between rim, cavity and body is consistent; minor issue is the very dark exterior that nearly dissolves into the background and a slightly exaggerated rim band." }, "broth": { "score": 0.75, "reason": "Amber-brown liquid surface with edge sheen, floating scallion rings, sesame and small bubbles, and believable immersion/occlusion of tofu and leaves. Slightly murky and flat in the lit mid-zone, but it reads as liquid rather than solid." }, "steam": { "score": 0.8, "reason": "Translucent warm wisps rise from the bowl and clearly change shape across desktop-1 through pointer-left frames, indicating continuous animated rise/drift without hiding the food; slightly foggy/blobby in the mid-plume, and static in reduced frames as expected." }, "composition": { "score": 0.85, "reason": "Well-balanced dark atmospheric scene: single focal bowl, warm backlit steam, restrained glow, no text or clutter, and the layout holds in the mobile frame. Food remains the focus; the near-black bowl edges blend a bit heavily with the backdrop." }, "vegan": { "score": 0.5, "reason": "Toppings are otherwise plant-based (nori, tofu, corn, scallion, greens, sesame), but the two brown discs carry a spiral/cross pattern characteristic of narutomaki fish cakes; under the warm lighting they could also be scored shiitake caps, so appearance is ambiguous rather than unmistakable." } } ], "diagnostics": { "vegan": 0.7666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.898888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 133.29999999999927, "frameIntervalP95Ms": 133.29999999999927, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "972f338dce13e501439c53c2cf04b81d06bab4b2f014f72b55283209266a10ea", "gradedAt": "2026-09-30T22:47:58.176097+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6788888888888889, "toppings": 0.7166666666666667, "bowl": 0.8322222222222222, "broth": 0.7211111111111111, "steam": 0.6344444444444445, "composition": 0.8344444444444444 }, "reward": 0.7321111111111112, "runId": "anthropic/opus-5.5/high", "rewardStdDev": 0.039752474882191385, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6966666666666667, "toppings": 0.74, "bowl": 0.8933333333333333, "broth": 0.7233333333333333, "steam": 0.5266666666666666, "composition": 0.8666666666666667 }, "reward": 0.741, "sampleStdDev": 0.011947803145348561, "judgments": [ { "noodles": { "score": 0.7, "reason": "Continuous golden strands have readable bends, highlights and overlaps, but many share repetitive shallow waves and form a relatively flat layer rather than supple tangled volume." }, "toppings": { "score": 0.74, "reason": "Veined scalloped greens, scored mushroom caps, porous tofu and detailed scallion rings provide good variety. The nori resembles smooth rigid cards, and corn kernels look unusually glossy and bead-like." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, shaded inner cavity, curved lacquer-like walls and attached foot establish convincing depth and support. Fine gold bands and restrained highlights reinforce the material; no material construction defect is observed." }, "broth": { "score": 0.72, "reason": "Warm amber shading, scattered oil droplets and partially submerged strands suggest appetizing liquid. Limited contact menisci and uniformly surface-mounted toppings weaken immersion." }, "steam": { "score": 0.46, "reason": "The translucent haze changes subtly across normal frames and rises above the soup, but reads as a broad luminous fog bank rather than distinct irregular wisps with clearly legible rise and fade." }, "composition": { "score": 0.86, "reason": "A single readable dish dominates the dark warm scene, with diagonal chopsticks, restrained tabletop detail and coordinated green-gold accents. No distracting text or UI is present; mobile framing leaves substantial empty space above the bowl." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.72, "reason": "Continuous golden strands have readable overlaps and shaded thickness, but mostly follow similar horizontal waves with limited tangled volume." }, "toppings": { "score": 0.74, "reason": "Veined greens, scored mushroom caps, porous tofu and scallion rings provide readable variety. The nori looks like smooth rigid cards, and corn kernels are overly glossy." }, "bowl": { "score": 0.9, "reason": "Coherent elliptical rim, inner cavity, rounded lacquer-dark walls and attached foot establish convincing depth and support. Restrained rim highlights and fine gold bands reinforce the material; no material construction defect is observed." }, "broth": { "score": 0.73, "reason": "Warm amber gradients, small oil droplets and submerged noodle shadows suggest liquid depth. Some toppings appear laid on the surface with insufficient immersion or contact variation." }, "steam": { "score": 0.55, "reason": "Normal frames show subtle changes in the translucent haze above the soup, but it reads as a broad diffuse cloud rather than distinct rising, spreading wisps and washes over the rear toppings." }, "composition": { "score": 0.87, "reason": "The single dish remains the clear focus against a dark, uncluttered backdrop, with warm food colors and a strong chopstick diagonal. Mobile preserves the readable bowl, though its upper negative space is generous." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, corn, scallions and seaweed; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.67, "reason": "Continuous golden strands have readable highlights and overlaps, but many share similar shallow zigzag bends and lie in a relatively flat layer rather than a supple tangled volume." }, "toppings": { "score": 0.74, "reason": "Veined greens, scored mushroom caps, porous tofu and scallion rings provide clear variety and detail. The nori is unusually smooth and rectangular, and corn kernels look somewhat glossy and bead-like." }, "bowl": { "score": 0.9, "reason": "Coherent elliptical rim, visible inner walls, rounded dark body and attached foot establish convincing depth and support. Restrained highlights and fine gold bands suggest lacquer; no material construction defect is observed." }, "broth": { "score": 0.72, "reason": "Warm amber shading, small oil droplets and submerged noodle shadows suggest liquid depth. Several toppings look laid on the surface, with limited meniscus or contact detail." }, "steam": { "score": 0.57, "reason": "The normal frames show modest changes in translucent haze above the soup, supporting ambient motion. Broad diffuse clouds lack clearly resolved rising wisps and partially wash out the rear toppings." }, "composition": { "score": 0.87, "reason": "The single bowl remains the clear focus against a dark warm background, with a balanced diagonal from the chopsticks and readable green-gold food colors. No distracting text or UI is present; mobile framing leaves considerable empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings read as greens, mushrooms, tofu, corn, scallions and seaweed, with noodles and plant-compatible broth; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6066666666666667, "toppings": 0.6933333333333334, "bowl": 0.77, "broth": 0.6733333333333333, "steam": 0.6266666666666667, "composition": 0.7866666666666667 }, "reward": 0.6886666666666668, "sampleStdDev": 0.004725815626252613, "judgments": [ { "noodles": { "score": 0.6, "reason": "Wavy yellow strands overlap with some volume and sit in the broth, but they read as repetitive squiggles and are partly hidden under the toppings. They lack the supple continuous look of real noodles." }, "toppings": { "score": 0.68, "reason": "Variety is good: shiitake with star scoring, bok choy, tofu cubes, corn, scallion rings, nori sheets and menma. The details read clearly. The tofu is a bit blocky, the nori stands as stiff rectangles, and the scallion rings look like stamped repeats." }, "bowl": { "score": 0.75, "reason": "Coherent dark lacquer bowl with a lighter inner rim, a foot base, and a subtle highlight band and groove lines. The perspective is consistent, and the base sits on the table with a soft shadow. The wall is somewhat flat and dark." }, "broth": { "score": 0.68, "reason": "Warm orange-brown broth with chili oil droplets, bubbles and a soft glow. Toppings sit convincingly in it. The surface is somewhat flat and matte, with limited reflection." }, "steam": { "score": 0.66, "reason": "Soft, translucent, grainy steam rises above the bowl and shifts shape between frames. It is a bit foggy and diffuse, though it does not hide the food, and it lacks distinct wisps." }, "composition": { "score": 0.78, "reason": "Centered bowl with a dark moody backdrop, warm bokeh and a wood table. The chopsticks add a diagonal, and the food is the focus. There is no text or clutter. The nori stands a bit awkwardly." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, bok choy, corn, scallion, nori and noodles are visible. No animal products." } }, { "noodles": { "score": 0.62, "reason": "Wavy yellow strands overlap with some volume and sit in the broth. They read as instant-style noodles, but they are thin and somewhat repetitive, with flat squiggles and a slightly scribbly mass." }, "toppings": { "score": 0.7, "reason": "Varied and readable toppings: bok choy with white stems, shiitake with star scoring, tofu cubes, corn kernels, scallion rings, menma and nori. Illustrated style is consistent. The nori sheets are stiff upright rectangles and the scallion rings are somewhat uniform." }, "bowl": { "score": 0.78, "reason": "Dark lacquer bowl with a coherent elliptical rim, inner wall, highlight band, foot base and ground shadow. Perspective is consistent. The rim is a little wide and the body is plain." }, "broth": { "score": 0.66, "reason": "Warm orange-brown broth with chili oil droplets and a soft sheen. Ingredients sit in it with some contact shading. The surface is somewhat flat and opaque." }, "steam": { "score": 0.62, "reason": "Soft translucent haze rises above the bowl and shifts shape between frames. It is fairly diffuse and fog-like, partly veiling the nori, and lacks distinct wisps." }, "composition": { "score": 0.8, "reason": "Centered bowl in a dark atmospheric scene with warm bokeh and a wooden table. Colors are cohesive and the food is the focus. Chopsticks add balance. No text or clutter." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, corn, scallions, nori and noodles. No egg, meat or fish is visible." } }, { "noodles": { "score": 0.6, "reason": "Wavy, crinkled yellow strands overlap with believable volume and sit in the broth. They read as instant-style noodles. They are somewhat thin and repetitive, and partly buried under toppings, so the mass is a little indistinct." }, "toppings": { "score": 0.7, "reason": "Toppings are varied and readable: shiitake with star-scored caps, bok choy with a white stem, tofu cubes, corn kernels, scallion rings, and nori. Textures are decent. The nori looks like two stiff upright cards, and the scallion rings are a bit uniform and sticker-like." }, "bowl": { "score": 0.78, "reason": "Dark lacquer bowl with a coherent elliptical rim, inner wall, pedestal foot, and subtle highlight lines. It is grounded by a shadow on a wooden surface. The perspective is consistent. The inner rim is slightly flat, and the foot is a little murky." }, "broth": { "score": 0.68, "reason": "Warm orange-brown broth with oil droplets, chili specks, and a soft sheen. Ingredients sit in it with some occlusion and shadow. The surface is a little flat and uniform, and the noodle immersion is only lightly suggested." }, "steam": { "score": 0.6, "reason": "Soft translucent haze rises above the bowl and changes shape across frames. It is grainy and diffuse, and more of a fog cloud than distinct wisps. It partly veils the nori but does not hide the food." }, "composition": { "score": 0.78, "reason": "The bowl is centered as a clear focal point in a dark, warm atmosphere, with bokeh and a chopstick adding depth. The palette is cohesive and there is no text or clutter. The steam and nori top-heaviness is minor." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, bok choy, corn, scallions, nori, and noodles. There is no meat or egg, and the broth looks plant-based." } } ], "diagnostics": { "vegan": 0.93 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7333333333333333, "toppings": 0.7166666666666667, "bowl": 0.8333333333333334, "broth": 0.7666666666666667, "steam": 0.75, "composition": 0.85 }, "reward": 0.7666666666666667, "sampleStdDev": 0.026259918760981268, "judgments": [ { "noodles": { "score": 0.75, "reason": "Yellow curly strands are continuous with varied bends, highlights and shadow layering, sitting believably on the broth surface; the S-curl motif repeats noticeably in places, giving a slightly patterned look." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: bok choy with veined leaves, shiitake slices with cross-hatched caps, menma strips, scallion rings, chili threads and sesame. Weaker elements: the two nori sheets read as flat stiff rectangles in a slightly light sage color, and tofu cubes/corn are somewhat plain and glossy." }, "bowl": { "score": 0.85, "reason": "Coherent lacquer bowl with clean rim ellipse, cream inner band, gold body lines, well-joined pedestal foot with table contact shadow; chopsticks rest naturally across the rim. Only minor softness in material detail." }, "broth": { "score": 0.8, "reason": "Rich brown surface with subtle sheen, chili-oil droplets, small bubbles and edge darkening; toppings show occlusion shadows and appear immersed rather than pasted on. A few bubble rings read slightly graphic." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the wisps visibly change shape, height and spread while staying translucent and not obscuring the food; reduced-motion frames are static as expected. Slightly fog-like in density at times." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with warm bokeh, restrained dark palette, coherent warm lighting, no text or clutter; mobile framing keeps the bowl readable despite generous empty space above." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, corn, bok choy, scallion, bamboo shoots, nori and chili oil \u2014 all plant-derived; no animal toppings appear." } }, { "noodles": { "score": 0.75, "reason": "Wavy curly strands with continuous tan bodies and darker outlines coil believably across the broth, with visible strand ends and scallion rings resting on them; the squiggle pattern is somewhat uniform and slightly flat in places, but reads as coherent ramen noodles rather than scribbles." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: bok choy with veined leaves and pale stems, shiitake slices with radial star cuts, bamboo strips, sesame, chili threads, scallion rounds. Deductions for the corn rendered as glossy uniform yellow spheres, plain untextured tofu cubes, and the two nori sheets as flat gray-green featureless rectangles." }, "bowl": { "score": 0.8, "reason": "Dark lacquer-style bowl with a coherent elliptical rim, gold accent lines, believable wall thickness, warm interior light on the back rim, and a footed pedestal with a soft contact shadow on the wood. The base is a bit simplified and the foot is dark and low-contrast, but nothing is distorted or detached." }, "broth": { "score": 0.8, "reason": "Amber-toned liquid surface with speckled texture, small bubbles, red chili-oil dots, and a restrained back-rim highlight; noodles and bok choy dip into the surface with plausible occlusion, and no flat target discs or pasted-on ingredient look." }, "steam": { "score": 0.7, "reason": "The plume clearly changes shape and position across desktop-1/2/3 and both pointer frames, so it is animated, translucent, and irregular. However it forms one large dense fog mass that washes out the nori and back rim rather than distinct rising wisps, and near the chopsticks it veers opaque." }, "composition": { "score": 0.85, "reason": "Warm dark bokeh backdrop with a wood-grain table keeps the lit bowl as a clear focal point; balanced proportions, chopsticks resting on the rim as a natural accent, no text or clutter, and the mobile crop retains the framing. Minor loss of focal clarity where steam masks the nori." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: bok choy, shiitake, bamboo, tofu, corn, scallion, sesame, chili, and nori; no egg, meat, fish, or dairy appears in any frame." } }, { "noodles": { "score": 0.7, "reason": "Wavy tan strands form a coherent mass in the broth with plausible thickness and curl, though the squiggle pattern is somewhat repetitive and a few strands read as flat loops." }, "toppings": { "score": 0.65, "reason": "Good variety: shiitake slices with cross-cut stars, bok choy leaves with pale stalks, bamboo strips, corn, tofu cubes, chili threads and standing nori. Defects: scallion rings scattered like confetti on open broth, the nori reads as two stiff gray card rectangles, and the tofu cubes look glossy/blocky." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer bowl with consistent elliptical rim, visible wall thickness, gold pinstripe detail, believable pedestal foot, and chopsticks resting convincingly across the rim; only minor stiffness in the foot join." }, "broth": { "score": 0.7, "reason": "Appetizing miso-toned liquid with oil droplets, bubbles near the rim and gentle sheen; most toppings sit with believable occlusion, though isolated scallion rings and a light ring near the inner edge look slightly pasted on." }, "steam": { "score": 0.75, "reason": "Frames at 1100/4619/7923/16836/26441 ms show the translucent plumes changing shape, height and spread each time \u2014 clearly animated wisps that rise and thin without obscuring the food; puffs read slightly clustered rather than fully continuous." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point, warm dark bokeh backdrop, unified lighting, no text or clutter, and the mobile frame (390x844) renders cleanly without overflow or cropping issues." }, "vegan": { "score": 1, "reason": "Visible ingredients are all plant-derived: noodles, mushrooms, bok choy, scallions, tofu, corn, bamboo shoots, nori and chili oil; no egg, meat or fish appears." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9766666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 325.0, "frameIntervalP95Ms": 333.2999999999993, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5bb8a440f8267d48b6ec0125663728e16966d53239b1ee38029d70c4a679aae3", "gradedAt": "2026-09-30T22:48:43.630368+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4922222222222222, "toppings": 0.5888888888888889, "bowl": 0.758888888888889, "broth": 0.5655555555555556, "steam": 0.5644444444444444, "composition": 0.76 }, "reward": 0.6146666666666667, "runId": "anthropic/opus-5.5/low", "rewardStdDev": 0.07989732995121505, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.41000000000000003, "toppings": 0.5, "bowl": 0.78, "broth": 0.48, "steam": 0.44333333333333336, "composition": 0.7933333333333333 }, "reward": 0.5568333333333334, "sampleStdDev": 0.0060069404303133705, "judgments": [ { "noodles": { "score": 0.4, "reason": "Clearly separated yellow strands have consistent thickness and some overlap, but mostly form rigid, nearly straight stacked sticks rather than supple noodles with convincing bends and volume." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, speckled tofu, scallion rings and corn provide readable variety. Large smooth oval leaves and perfectly rectangular nori remain stiff, simplified food symbols with little organic texture." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inner cavity, curved dark walls and attached foot form a coherent vessel. Subtle highlights and decorative bands suggest lacquer, though the lower body and support are very dark and somewhat flattened." }, "broth": { "score": 0.48, "reason": "Warm brown coloring and scattered oil-like spots identify the soup surface, but it has little liquid sheen or depth. Ingredients largely appear layered on top rather than visibly immersed." }, "steam": { "score": 0.45, "reason": "Translucent hazy patches change position across normal frames without obscuring the food. They mostly resemble detached background clouds, lacking clearly readable emergence from the soup and tapering rising wisps." }, "composition": { "score": 0.8, "reason": "A single readable dish anchors the warm, dark scene with a restrained palette and no text or UI clutter. Diagonal chopsticks add balance, although the desktop bowl sits low and mobile presentation leaves considerable empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, nori, corn and scallions plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.4, "reason": "Numerous distinct yellow strands have consistent thickness and some overlap, but mostly resemble straight, cut sticks arranged in a flat pile rather than supple continuous noodles." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, speckled tofu, scallion rings and greens provide readable variety. However, leaves are smooth oval panels, nori is rigid rectangular cards, and most textures remain schematic." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, shaded curved walls, decorative bands and attached foot form a coherent vessel with convincing depth. Very dark lower shading and weak contact grounding limit material definition." }, "broth": { "score": 0.48, "reason": "Warm brown liquid and scattered oil-like spots are readable, but the surface is largely flat and matte, with little immersion or liquid contact around the ingredients." }, "steam": { "score": 0.42, "reason": "Translucent haze changes position across normal frames and leaves the food visible, but appears mainly as broad disconnected blurred patches rather than wisps visibly emerging from the soup." }, "composition": { "score": 0.79, "reason": "A single clear focal bowl, restrained warm dark palette and diagonal chopsticks create an uncluttered presentation. Desktop placement is bottom-heavy, while the mobile dish is relatively small amid substantial empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.43, "reason": "Numerous clearly separated yellow strands have consistent thickness, but mostly straight, parallel lengths resemble rigid sticks rather than supple noodles, with little curled volume." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, speckled tofu, scallion rings and leafy greens provide readable variety. However, oversized smooth oval leaves and rigid rectangular nori dominate, with limited organic texture." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved dark walls, decorative bands and attached foot form a coherent vessel with convincing depth. Subdued highlights and weak ground contact make the material and support less convincing." }, "broth": { "score": 0.48, "reason": "Warm brown broth and scattered oil spots are recognizable, but the surface is largely matte and flat; noodles and toppings appear layered above it with little visible immersion." }, "steam": { "score": 0.46, "reason": "Soft translucent patches change position across normal frames without obscuring the food. They mostly resemble detached blurred clouds, with little readable connection to the soup or continuous rising wisps." }, "composition": { "score": 0.79, "reason": "A single readable bowl, diagonal chopsticks and restrained warm background create a clean dark presentation without text or clutter. The desktop dish sits low, while the mobile dish is relatively small amid broad empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions and seaweed plausibly form vegan ramen; no material animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6, "bowl": 0.68, "broth": 0.5166666666666667, "steam": 0.5, "composition": 0.6866666666666666 }, "reward": 0.5813333333333334, "sampleStdDev": 0.002516611478423644, "judgments": [ { "noodles": { "score": 0.5, "reason": "Many thin yellow strands with overlap and some volume, but they are short, straight and rigid, looking like scattered sticks rather than supple noodles. Little bending or coiling." }, "toppings": { "score": 0.6, "reason": "Readable variety: shiitake with scored caps, tofu cubes with sesame, nori sheets, bok choy, corn and scallion rings. Shapes are somewhat flat and vector-like; the bok choy leaves are simple ovals and nori stands as rigid cards." }, "bowl": { "score": 0.68, "reason": "Dark lacquer bowl with a coherent elliptical rim, red bands, a foot and a soft rim highlight. Perspective is consistent. The base is faint and the body is somewhat flat." }, "broth": { "score": 0.55, "reason": "Warm orange-brown surface with a darker edge gradient and fat droplets, appetizing in color. Fairly flat, with little sheen or immersion detail; noodles and toppings sit on top." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps drift and change between frames, and they don't hide the food. They read as blurry blobs and bokeh with weak connection to the soup surface." }, "composition": { "score": 0.68, "reason": "Centered bowl with a dark warm vignette and a clear focal point; chopsticks add a diagonal. The toppings are somewhat cluttered, and the bowl base fades into darkness." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, nori, corn and scallions with no visible animal products." } }, { "noodles": { "score": 0.5, "reason": "Many thin yellow strands with overlap and volume read as noodles, but they are short, straight and rigid, like cut sticks, with few natural bends or continuous loops. Little of them appears submerged in the broth." }, "toppings": { "score": 0.6, "reason": "Shiitake with scored cross marks, tofu cubes with speckles, nori sheets, bok choy leaves, corn and scallion rings give good variety and are readable. The bok choy is a smooth, plasticky blob, the nori stands as stiff cards, and the scallion rings are uniform." }, "bowl": { "score": 0.68, "reason": "The dark lacquer bowl has a coherent elliptical rim, a tapered body, red accent lines and a rim highlight. The base is faint and merges into the dark background, so contact and support are weak. Overall perspective is consistent." }, "broth": { "score": 0.5, "reason": "The orange-brown broth has a gradient and small oil droplets, which gives an appetizing color. It looks fairly flat, and the toppings and noodles sit on top rather than being immersed, with little visible sheen or depth." }, "steam": { "score": 0.5, "reason": "Soft translucent blurry wisps drift and change between frames, and they do not hide the food. They read as diffuse blobs rather than wisps emerging from the soup, and they are fairly faint." }, "composition": { "score": 0.7, "reason": "The bowl is a centered, dominant focal point on a dark warm vignette, with pleasing colors and no text or clutter. The chopsticks add a diagonal, and the lower half is slightly empty." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, bok choy, nori, corn and scallion, with no visible meat, egg or fish. The broth appears plant-based." } }, { "noodles": { "score": 0.5, "reason": "Many pale yellow strands with some overlap and volume, but they are mostly straight, rigid and stacked like short sticks; few bends and they read as flat, repetitive lines lying on the surface." }, "toppings": { "score": 0.6, "reason": "Variety is readable: shiitake with scored caps, tofu cubes, torn nori sheets, bok choy, corn, scallion rings and sesame. The nori looks like stiff rectangles and the bok choy leaves are smooth blobs, so they feel vector-flat rather than organic." }, "bowl": { "score": 0.68, "reason": "A coherent dark bowl with an elliptical rim, tapering walls, red accent lines and a base, with consistent perspective. The interior looks shallow, the foot is barely visible and it fades into the background, and the highlights are muted." }, "broth": { "score": 0.5, "reason": "Warm orange-brown broth with oil-droplet dots and a darker edge gradient gives some depth. The surface is fairly flat, and the noodles and toppings sit on top rather than being immersed." }, "steam": { "score": 0.5, "reason": "Soft translucent blurry wisps drift and change between frames without hiding the food. They read as diffuse blobs and bokeh rather than distinct wisps rising from the broth, and they are faint." }, "composition": { "score": 0.68, "reason": "A centered bowl in a dark, warm atmosphere with the food as the focus, and chopsticks add a diagonal. The bowl is cropped a little low, and the chopsticks and nori overlap awkwardly; there is no clutter." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn, scallion and noodles are all plant-based, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6666666666666666, "bowl": 0.8166666666666667, "broth": 0.7, "steam": 0.75, "composition": 0.8 }, "reward": 0.7058333333333333, "sampleStdDev": 0.030034702151566804, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles read as a fanned bundle of thin, uniform parallel strands with repeated similar arcs; they lack volume, tangle and coherent overlap, looking flat and somewhat rigid, and are confined to the left half of the bowl rather than filling it believably." }, "toppings": { "score": 0.7, "reason": "Good variety and readable ingredients: scored shiitake caps with white cross-cuts, tofu cubes with sesame specks, scallion rings, corn, chili threads and standing nori sheets. Defects: the large glossy green leaf on the right is smooth and plasticky in shape, tofu cubes are plain featureless blocks, and nori reads as stiff flat cards." }, "bowl": { "score": 0.8, "reason": "Coherent black lacquer bowl with consistent elliptical rim, tapering walls, red accent rings, subtle rim highlight and a visible foot providing support; contact with the surface reads correctly. It is very dark and merges into the background, slightly reducing rim/body separation." }, "broth": { "score": 0.7, "reason": "Warm orange-brown liquid surface with small oil droplets and restrained sheen; ingredients sit on and overlap the surface with plausible contact. The broth is somewhat flat and uniform in texture with limited depth, reading more as flat fill than liquid." }, "steam": { "score": 0.75, "reason": "Across desktop-1/2/3 the translucent wisps shift position and shape above the bowl, supporting animated rising steam; wisps are soft, irregular and never obscure the food. They are quite faint and blobby, and nearly vanish in the reduced-motion frames." }, "composition": { "score": 0.8, "reason": "Centered focal dish on a warm dark atmospheric background with a coherent glow, chopsticks as an accent, no text or clutter, and appetizing color contrast between broth and toppings. Slight imbalance from large empty upper space, and the mobile layout pushes the bowl into the lower third." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, shiitake mushrooms, corn, scallions, chili threads, nori and a leafy green in a plant-toned broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.55, "reason": "Yellow strands are continuous with plausible thickness and read as noodles in broth, but they are drawn as largely straight, flat parallel ribbons with minimal bends or volume, giving a rigid bundled appearance." }, "toppings": { "score": 0.6, "reason": "Good variety: scored shiitake caps, tofu cubes, standing nori sheets, corn, chili threads and scallion rings are recognizable. However scallion rings are scattered pasted-looking ovals floating over mushrooms, and the glossy white/green vegetable on the right reads as an abstract plastic-like form rather than crisp bok choy." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a clean rim ellipse, visible interior wall depth, subtle body ridges, a foot ring, and chopsticks resting naturally across the rim. Slightly stylized and flat-shaded, but no broken joins or perspective conflicts." }, "broth": { "score": 0.7, "reason": "Orange-brown surface has soft gradient, oil-droplet speckles and restrained sheen; tofu and mushrooms sit in contact with the surface. Noodles and some scallions read slightly pasted on top, and the liquid is a bit flat/uniform." }, "steam": { "score": 0.7, "reason": "Translucent blurred wisps rise above the bowl and shift position/shape across desktop frames 1-3, without hiding the food; reduced-motion frames are appropriately static. The wisps are somewhat amorphous blob-like rather than finely curled plumes." }, "composition": { "score": 0.8, "reason": "Centered focal dish on a warm dark atmospheric background, balanced chopstick diagonal, no text or clutter, and the mobile viewport scales cleanly without overflow. Background steam blobs are slightly diffuse but not distracting." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, shiitake mushrooms, nori, scallions, corn and chili threads over plant broth \u2014 all plausibly vegan with no animal-derived ingredients." } }, { "noodles": { "score": 0.6, "reason": "Numerous pale yellow strands form a coherent bed with plausible thickness, but they read as mostly flat, near-parallel straight lines with limited curvature, overlap, or volume rather than supple wavy noodles." }, "toppings": { "score": 0.7, "reason": "Good variety: shiitake caps with white cross cuts and sesame, tofu cubes, two standing nori sheets with speckle texture, corn, chili threads, and scallion rings. The large green bok-choy-like halves are recognizable but slightly glossy/cartoonish, and scallion rings are scattered a bit uniformly." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with clean rim ellipse, tapering body, subtle red accent stripes, soft highlights, and a foot with contact against the surface; no broken joins or perspective mismatch observed." }, "broth": { "score": 0.7, "reason": "Warm orange-brown surface with specks and restrained sheen, and toppings sit with visible edges/shadows; however most of the surface is covered by the flat noodle bed, leaving limited readable liquid depth and immersion cues." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and the pointer frames the translucent gray wisps shift position and shape over the soup, rising and dispersing without opaque fog or hiding the food; wisps are somewhat blobby but clearly dynamic." }, "composition": { "score": 0.8, "reason": "Balanced desktop framing with the dish as clear focal point, warm dark atmospheric background, chopsticks adding diagonal interest, and no text/UI clutter; on mobile the bowl is small with much empty space, slightly weakening presentation." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, nori, corn, scallions, bok choy, and chili in plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.29999999999927, "frameIntervalP95Ms": 33.400000000001455, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a168adef2e60a2d729c9a8e78b1170ccb011e4632f014c5433ea139859b503bc", "gradedAt": "2026-09-30T22:48:26.308634+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6477777777777778, "toppings": 0.7044444444444444, "bowl": 0.8211111111111111, "broth": 0.7011111111111111, "steam": 0.7177777777777777, "composition": 0.8333333333333334 }, "reward": 0.731, "runId": "anthropic/opus-5.5/medium", "rewardStdDev": 0.0639363833127204, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.62, "toppings": 0.6833333333333333, "bowl": 0.88, "broth": 0.6433333333333333, "steam": 0.7066666666666667, "composition": 0.88 }, "reward": 0.7291666666666667, "sampleStdDev": 0.0020207259421636875, "judgments": [ { "noodles": { "score": 0.62, "reason": "Continuous cream-colored strands have rounded bends and readable overlaps, but mostly form a flat, similarly oriented wavy layer with limited tangled volume." }, "toppings": { "score": 0.68, "reason": "Mushroom gills, leaf veins, tofu pores, corn and scallion rings provide readable variety. Shapes remain schematic, and the nori is an unnaturally straight rectangular sheet." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, shaded cavity, curved lacquer-like walls, fine red bands and an attached foot create convincing depth and support. No material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm amber shading and small surface bubbles suggest appetizing liquid, but the surface lacks nuanced reflections and ingredients appear largely laid atop it rather than partially immersed." }, "steam": { "score": 0.72, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the food. Steam spreads softly, but reads more as diffuse billowing haze than distinct rising, tapering wisps." }, "composition": { "score": 0.88, "reason": "The single dish remains clearly framed on desktop and mobile, with warm food colors against a restrained dark wooden setting and no distracting UI. Generous negative space and soft lighting maintain focus; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, scallions, seaweed and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.62, "reason": "Continuous cream-colored strands have gentle bends and readable overlaps, but mostly run in flat parallel waves with limited tangled volume and immersion." }, "toppings": { "score": 0.69, "reason": "Mushroom gills, tofu pores, leaf veins, corn kernels and scallion rings provide readable variety. Shapes remain diagrammatic, especially the straight rectangular nori and uniformly outlined mushrooms." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, shaded inner wall, rounded lacquer body, fine red bands and attached foot establish convincing depth and support. No material construction defect is observed." }, "broth": { "score": 0.64, "reason": "Warm amber shading and small surface bubbles suggest appetizing liquid, but the surface is very smooth and ingredients mostly appear laid on top rather than partially submerged." }, "steam": { "score": 0.7, "reason": "Translucent haze changes position and density across normal frames while leaving the food visible. It spreads above the soup, but reads more as broad diffuse clouds than distinct rising, tapering wisps." }, "composition": { "score": 0.88, "reason": "A single clearly framed dish, warm food colors against dark wood, restrained atmospheric light and uncluttered presentation create a strong focal hierarchy on desktop and mobile. No material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.62, "reason": "Continuous cream-colored strands have rounded bends and readable overlaps, but mostly form a flat, horizontally layered patch with limited tangled volume." }, "toppings": { "score": 0.68, "reason": "Mushroom gills, tofu pores, leaf veins, corn and scallion rings provide recognizable variety. Shapes remain diagrammatic, especially the straight rectangular nori and uniformly outlined mushrooms." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, shaded cavity, curved dark walls, fine red bands and a grounded foot convincingly depict lacquerware. No material construction defect is observed." }, "broth": { "score": 0.64, "reason": "Warm amber shading and scattered small surface bubbles suggest appetizing liquid, but weak immersion cues make much of the food appear laid over the surface." }, "steam": { "score": 0.7, "reason": "Translucent vapor changes distribution across the normal frames and spreads above the soup without hiding it. It reads more as broad diffuse haze than distinct rising, tapering wisps." }, "composition": { "score": 0.88, "reason": "A single clearly framed dish, restrained warm lighting, dark wood surface and uncluttered background create a balanced atmospheric presentation that remains readable on mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6066666666666667, "toppings": 0.6466666666666667, "bowl": 0.75, "broth": 0.6766666666666666, "steam": 0.6133333333333333, "composition": 0.7533333333333333 }, "reward": 0.668, "sampleStdDev": 0.008888194417315597, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles are wavy, cream-colored strands that overlap and read as ramen with some volume. They are clustered in one mound and look a bit flat and repetitive, with little strand definition." }, "toppings": { "score": 0.66, "reason": "Variety is good: shiitake caps, tofu cubes, bok choy leaves with veins, corn, scallion rings, sesame and a nori sheet. The nori is a rigid rectangle, the scallion rings are thin outlines, and the shapes are small and flat. They read clearly but look like clip-art." }, "bowl": { "score": 0.75, "reason": "Dark lacquer bowl with a coherent elliptical rim, a red accent band, a foot and a soft contact shadow. Perspective is consistent and the highlights are subtle. The only small flaw is a tiny odd notch on the left side." }, "broth": { "score": 0.65, "reason": "Warm orange-brown broth with a gradient, small bubbles and a rim reflection, which looks appetizing. It reads somewhat flat, and the toppings sit on it with limited immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Soft translucent haze and wisps rise above the bowl and shift between frames, and they do not hide the food. The steam is diffuse and broad, more like fog than distinct wisps, and its motion is modest." }, "composition": { "score": 0.75, "reason": "The bowl is centered as the clear focal point in a dark, moody wooden-table scene with warm colors and a vignette. There is no text or clutter. The nori and the bowl are well balanced, though the scene is a bit empty." }, "vegan": { "score": 0.95, "reason": "Visible ingredients are mushrooms, tofu, bok choy, corn, scallions, nori and noodles. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.6, "reason": "Wavy cream strands with varied bends and overlap read as ramen noodles, but they are small, somewhat flat and clustered in one patch with repeated similar curves." }, "toppings": { "score": 0.62, "reason": "Shiitake caps, tofu cubes, bok choy leaves, corn, scallion rings and nori are all recognizable and plant-based. Nori is a clean rectangle, scallions are thin outline rings, and tofu blocks are simple, so the depiction is stylized but varied." }, "bowl": { "score": 0.75, "reason": "Dark lacquer bowl with a coherent elliptical rim, red accent band, foot and shadow on the wood table. Perspective is consistent and the highlights are subtle. There is a slight odd notch near the left rim and the wall reads a bit flat." }, "broth": { "score": 0.68, "reason": "Warm orange-brown broth with a gradient, small bubbles and edge shading looks appetizing and deep. Ingredients sit in it fairly well, though the surface is fairly flat with little sheen texture." }, "steam": { "score": 0.62, "reason": "Soft translucent wisps rise above the bowl and change shape across frames, drifting and fading without hiding the food. They are diffuse and somewhat fog-like rather than distinct wisps." }, "composition": { "score": 0.75, "reason": "Centered bowl in a dark, moody wooden-table scene with warm lighting and a clear focal point. No text or clutter. The bowl is somewhat small in the frame and the bokeh background is plain." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, corn, bok choy, scallion, nori and noodles with no visible meat, egg or fish." } }, { "noodles": { "score": 0.62, "reason": "Wavy cream noodle strands overlap with believable volume and sit in the broth. They are fairly small, with some repeated flat curves, and the mass looks a bit busy." }, "toppings": { "score": 0.66, "reason": "Shiitake, tofu cubes, bok choy leaves, scallion rings, corn and nori are all readable and varied. Some pieces look flat or vector-like, and the nori is a rigid upright rectangle." }, "bowl": { "score": 0.75, "reason": "The dark lacquer bowl has a coherent rim, a red accent band, a foot and a soft contact shadow. Perspective is consistent. The rim and body are slightly plain, and there is a small odd notch at the left." }, "broth": { "score": 0.7, "reason": "The warm orange-brown broth has depth, a soft gradient, bubbles and a sheen, and the noodles and toppings sit in it. It is somewhat flat and smooth in places." }, "steam": { "score": 0.62, "reason": "Translucent soft wisps rise and change shape between frames, without hiding the food. They are diffuse and hazy, more like a fog cloud than distinct wisps." }, "composition": { "score": 0.76, "reason": "The bowl is a centered focal point in a dark, moody, warm wood setting with no distracting UI. The food is small relative to the frame, and the background is a bit empty." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, vegetables, corn, seaweed and noodles are visible. There are no animal products." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.7833333333333333, "bowl": 0.8333333333333334, "broth": 0.7833333333333333, "steam": 0.8333333333333334, "composition": 0.8666666666666667 }, "reward": 0.7958333333333334, "sampleStdDev": 0.005204164998665327, "judgments": [ { "noodles": { "score": 0.7, "reason": "Wavy cream-colored strands form a coherent mass with varied bends, outlines and slight shadow volume, but they read as a fairly uniform squiggle layer pasted across the surface rather than strands with depth and believable submersion." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: shiitake slices with gills, sesame-speckled tofu cubes, scallion rings, corn kernels, chili threads, bamboo-strip pieces, leafy greens and a large textured nori sheet. Scallion rings look scattered on top and one diagonal white-green stalk piece is ambiguous, keeping it from higher." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer-style bowl with consistent rim ellipse, visible inner wall, red decorative band, tapering body and footed base sitting on the table with a subtle contact shadow; no broken joins or perspective mismatch, only minor stiffness in the glaze highlights." }, "broth": { "score": 0.8, "reason": "Warm amber tonkotsu-like surface with darker edge gradient, restrained sheen and scattered oil bubbles; tofu, mushrooms and leaves sit in the liquid with occlusion, though the noodle layer floats flatly on top rather than emerging from it." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise and drift across the desktop frames (patterns clearly change between 1.0s, 3.3s, 5.7s and pointer frames), spreading and thinning without obscuring the dish; slightly billowy/foggy in places but continuous, not static blobs." }, "composition": { "score": 0.9, "reason": "Centered focal bowl on a wooden table with warm rim light, dark atmospheric bokeh background and restrained palette; food stays the focus with no clutter or UI. The oversized nori sheet slightly dominates the back edge." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: mushrooms, tofu, bok choy/leek, scallions, corn, chili threads, bamboo strips, sesame and nori; no egg, meat, fish or dairy forms are present." } }, { "noodles": { "score": 0.75, "reason": "Continuous wavy pale-yellow strands with coherent overlap and believable layering in the broth; wave pattern is somewhat regular and strands lie flat on the surface with limited volume/heaping, but no maze scribbles or broken ribbons." }, "toppings": { "score": 0.8, "reason": "Readable shiitake caps with stems, speckled tofu cubes, corn cluster, scallion rings, chili threads, sesame, bok-choy-style leaves and a standing nori sheet; good variety and organic shapes, with the nori and tofu slightly flat/card-like but appropriate for the style." }, "bowl": { "score": 0.8, "reason": "Coherent rim ellipse, inner lip highlight, dark lacquer body with red band detail, visible foot and contact shadow on the wooden table; perspective of rim, body and base is consistent with the elevated 3/4 view with no visible joins breaking." }, "broth": { "score": 0.75, "reason": "Appetizing orange miso-toned surface with edge bubbles, oil specks and a tare ring near the rim; toppings sit with some occlusion, though noodles rest mostly on top of the liquid rather than visibly dipping in, and surface texture is fairly uniform." }, "steam": { "score": 0.8, "reason": "Time-ordered desktop frames show translucent wisps that change position and shape between 1056ms, 3359ms and 5758ms, rising and spreading above the bowl without hiding the food; wisps stay somewhat diffuse/cloudy but are clearly animated, not static plumes." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark wood table with warm atmospheric glow, restrained bokeh and dust particles, coherent lighting and no distracting text or clutter; mobile crop leaves generous dark space but the dish stays readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, corn, scallion, leafy greens, nori, chili and sesame in an orange plant-based broth; no meat, fish, egg or dairy-like toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Pale wavy strands with consistent thickness, gentle bends and overlap read clearly as noodles sitting in the broth. Shortcoming: they are arranged as fairly uniform side-by-side squiggles with limited tangling or depth, giving a slightly patterned, flat mass rather than a natural tangle." }, "toppings": { "score": 0.75, "reason": "Good variety with recognizable shiitake slices (gill lines visible), speckled 3D tofu cubes, a textured standing nori sheet, bok choy leaves with pale stems, corn kernels, bamboo strips, sesame and chili threads. Minor defects: scallion rings look like small uniform hoops, chili threads are slightly scribbly, and mushrooms read a touch flat." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer bowl: smooth rim ellipse, visible wall thickness, red accent band, tapering body with foot and base, consistent slightly-elevated perspective, body highlight and plausible contact with the wood surface. No broken joins or perspective clash observed; only very minor softness in the foot area." }, "broth": { "score": 0.8, "reason": "Amber-orange surface with a warm gradient, restrained sheen, small oil bubbles near the rim and a darker meniscus edge. Tofu, mushrooms, greens and noodles sit within the surface with believable contact rather than floating above it. Slightly flat in the mid-right area but appetizing and readable." }, "steam": { "score": 0.9, "reason": "Translucent, irregular wisps that clearly evolve across the time-ordered desktop frames: a plume rises over the nori in desktop-2, shifts and spreads rightward by desktop-3, and disperses differently in the pointer frames. Wisps rise, spread and fade without becoming opaque fog or obscuring the food." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with a warm amber/dark-lacquer palette on a wood-grain table, subtle bokeh and a dark atmospheric background; no text, UI or clutter competes with the dish. Mobile framing leaves generous empty dark space above/below, slightly reducing balance, but presentation remains clean." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu cubes, shiitake mushrooms, nori, scallions, corn, bok choy, bamboo strips, sesame and chili threads in plant broth \u2014 no meat, fish, egg or dairy present." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9788888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 150, "frameIntervalP95Ms": 150, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "59a9cc7e3e8529d53f8789778152366a310028ea4c6f5ae2250e05e158c47525", "gradedAt": "2026-09-30T22:46:48.770977+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6422222222222222, "toppings": 0.78, "bowl": 0.8233333333333334, "broth": 0.7266666666666667, "steam": 0.8233333333333334, "composition": 0.8511111111111112 }, "reward": 0.7693888888888889, "runId": "anthropic/opus-5.5/xhigh", "rewardStdDev": 0.020989635890477306, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.7633333333333333, "bowl": 0.9, "broth": 0.7066666666666667, "steam": 0.87, "composition": 0.8866666666666667 }, "reward": 0.7606666666666667, "sampleStdDev": 0.0022546248764114206, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable thickness and highlights, but the dominant parallel, tightly repeated waves form a flat patterned mat rather than a supple overlapping pile." }, "toppings": { "score": 0.76, "reason": "Veined greens, scored mushroom caps, sprouts, scallion rings and corn provide readable variety and detail. Tofu is rather sharply cubical, and the upright nori sheets look overly straight and card-like." }, "bowl": { "score": 0.9, "reason": "The elliptical rim, recessed cavity, curved dark walls and grounded foot form coherent perspective and depth. Fine surface streaks and restrained rim and wall highlights convincingly suggest lacquer; no material construction defect is observed." }, "broth": { "score": 0.7, "reason": "Warm amber shading, small oil droplets and localized sheen establish an appetizing soup surface. Ingredient contact is somewhat diagrammatic, with limited immersion cues around the tofu and noodle mat." }, "steam": { "score": 0.87, "reason": "Normal frames show changing translucent, irregular wisps rising above the soup and spreading into softer haze without obscuring the food. Some long bands are slightly stripe-like, but the varied shapes and fading edges are convincing." }, "composition": { "score": 0.88, "reason": "A clearly isolated focal bowl, warm food colors, dark wood and subdued background lights create a cohesive atmospheric presentation without UI or clutter. Mobile retains the whole dish, though its generous upper negative space slightly weakens balance." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushrooms, leafy vegetables, sprouts, corn, scallions and seaweed; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable thickness and highlights, but predominantly parallel, repetitive waves form a flat patterned mat rather than supple, overlapping noodle volume." }, "toppings": { "score": 0.77, "reason": "Veined leafy greens, scored shaded mushroom caps, scallion rings, sprouts and corn provide readable variety. Tofu has some surface texture, but remains sharply boxlike; nori sheets are unusually straight and rigid." }, "bowl": { "score": 0.9, "reason": "A coherent elliptical rim, deep inner wall, rounded tapering body and attached foot establish convincing construction. Warm rim highlights and subtle exterior streaks suggest glazed material; table contact is grounded. No material construction defect is observed." }, "broth": { "score": 0.72, "reason": "Warm amber shading, scattered oil droplets and small highlights suggest appetizing liquid. Ingredient contact is readable, though the surface is somewhat uniformly opaque and several toppings appear laid onto it with limited immersion." }, "steam": { "score": 0.87, "reason": "Normal frames show changing translucent wisps and irregular curls above the soup, spreading upward without obscuring ingredients. Fine streaks mix with softer haze convincingly, though the broad upper plumes are slightly smoky and their emergence is faint." }, "composition": { "score": 0.88, "reason": "The single dish remains prominent against a restrained dark wooden setting, with warm food colors, coherent highlights and uncluttered framing. Mobile preserves the entire bowl. The tall steam region adds slightly excessive empty upper space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, sprouts, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have readable thickness and highlights, but predominantly parallel, repeated waves form a flat patterned mat rather than supple overlapping noodle volume." }, "toppings": { "score": 0.76, "reason": "Veined leafy greens, scored mushroom caps, sprouts, scallion rings and corn provide clear variety and detail. Tofu is rather sharply geometric, and nori sheets are unusually straight and card-like." }, "bowl": { "score": 0.9, "reason": "A coherent elliptical rim, recessed cavity, curved lacquer-like walls, fine surface streaks and restrained highlights convincingly establish depth and material. The base meets the tabletop plausibly; no material construction defect is observed." }, "broth": { "score": 0.7, "reason": "Warm amber shading, small oil droplets and localized glints suggest appetizing broth. Ingredient contact is readable, but the surface is somewhat opaque and flat, with limited convincing partial immersion." }, "steam": { "score": 0.87, "reason": "Normal frames show changing translucent, irregular wisps rising above the bowl and spreading without obscuring the food. Fine trails and diffuse fading edges are convincing, though the upper cloud is somewhat broad and dense." }, "composition": { "score": 0.9, "reason": "The single dish remains clearly focal through warm food colors, subtle rim lighting and a restrained dark wooden setting. Both desktop and mobile preserve the whole bowl without distracting UI; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible noodles, tofu-like cubes, mushrooms, leafy greens, seaweed, sprouts and corn plausibly form vegan ramen; no visibly animal-derived topping is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6933333333333334, "toppings": 0.7933333333333333, "bowl": 0.77, "broth": 0.7066666666666667, "steam": 0.75, "composition": 0.8 }, "reward": 0.7541666666666668, "sampleStdDev": 0.0109696551146029, "judgments": [ { "noodles": { "score": 0.68, "reason": "Wavy, overlapping ramen strands with layered highlights read as real noodles with a believable tangle. They are small and compressed into a central cluster, and the strands look a bit uniform and flat." }, "toppings": { "score": 0.78, "reason": "Varied and clearly readable toppings: shiitake with star scoring, bok choy with veined leaves, golden tofu cubes, bean sprouts, corn, peas, scallions, chili threads, and nori with torn edges. Nori reads slightly like stiff cards standing upright, and the tofu cubes are somewhat geometric." }, "bowl": { "score": 0.75, "reason": "The bowl has a coherent elliptical rim, a lacquer-like gradient on the walls, a foot ring, and a specular highlight. Perspective is consistent. The base looks slightly thin and the contact shadow with the table is soft." }, "broth": { "score": 0.68, "reason": "The warm miso-orange broth has droplets, oil specks, and chili flecks, and the ingredients sit in it believably. The surface is fairly flat and matte, with little depth or reflection." }, "steam": { "score": 0.75, "reason": "Translucent wispy steam curls upward and changes shape across frames, including after the pointer moved, and it does not hide the food. It looks a bit streaky and is not strongly anchored to the bowl surface." }, "composition": { "score": 0.8, "reason": "A centered bowl on a dark wooden table with soft bokeh gives a moody, cohesive scene with warm colors. The food is the clear focus and there is no clutter or text. The bowl is slightly low in the frame." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, vegetables, seaweed, corn, and noodles are visible, with no egg, meat, or fish." } }, { "noodles": { "score": 0.7, "reason": "Wavy, overlapping ramen strands with believable crimp and volume, sitting in the broth. The mass is fairly compact and slightly repetitive in pattern, and the strands are small, but it reads as noodles." }, "toppings": { "score": 0.8, "reason": "Varied, clearly readable toppings: shiitake with star scoring, bok choy with veined leaves, fried tofu cubes, corn, peas, bean sprouts, scallion rings, chili threads. Nori has irregular torn edges. Nori sheets are a bit flat and rigid, and the tofu cubes look slightly geometric." }, "bowl": { "score": 0.78, "reason": "Coherent elliptical rim, lacquer-like brown walls with rim highlights, a visible foot and a contact shadow. The perspective is consistent. The base is slightly dark and indistinct against the table." }, "broth": { "score": 0.72, "reason": "Warm miso-orange surface with oil droplets, sheen and ripples, and the toppings sit in it. The surface is somewhat flat and uniform, with limited depth." }, "steam": { "score": 0.75, "reason": "Translucent wispy plumes rise and change shape across frames and fade into the dark background without hiding the food. They are somewhat wide and faint, but the motion is continuous and organic." }, "composition": { "score": 0.8, "reason": "A centered, well-lit bowl is the clear focus against a dark wood table with soft bokeh. The palette is warm and cohesive, and there is no clutter or text. Some empty space remains at the bottom." }, "vegan": { "score": 0.97, "reason": "Tofu, mushrooms, vegetables, seaweed, corn and noodles are all plant-based. No eggs, meat or fish are visible." } }, { "noodles": { "score": 0.7, "reason": "Wavy, layered noodle strands with believable ruffle and overlap, sitting in the broth under toppings. The mass is compact and somewhat repetitive, and individual strands are small and not very distinct." }, "toppings": { "score": 0.8, "reason": "Varied and readable: shiitake with star-cut tops, bok choy with veined leaves, torn-edged nori, bean sprouts, corn, peas, scallions, chili threads and golden tofu cubes. Tofu cubes are a bit blocky and the nori is slightly flat, but the textures are organic." }, "bowl": { "score": 0.78, "reason": "Coherent elliptical rim, dark lacquer walls, copper rim highlight, visible foot and a grounded contact with the table. Perspective is consistent. The base is slightly dim and the rim is thin in places, but no material defect." }, "broth": { "score": 0.72, "reason": "Warm miso-orange broth with oil droplets, subtle sheen and ripples. Ingredients sit in it with some occlusion. The surface is fairly flat and uniform, with limited depth." }, "steam": { "score": 0.75, "reason": "Translucent, irregular wisps rise from the bowl and change shape across frames, and they respond to the pointer. They drift a bit far from the soup and sit in the dark upper region, but they do not hide the food." }, "composition": { "score": 0.8, "reason": "Centered bowl as a clear focal point, warm food against a dark wood table and soft bokeh, with no text or clutter. Lighting is coherent and the food is the focus. The upper area is somewhat empty." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, vegetables, seaweed, corn, peas and noodles are visible, with no egg, meat or fish." } } ], "diagnostics": { "vegan": 0.9566666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7333333333333333, "toppings": 0.7833333333333333, "bowl": 0.8, "broth": 0.7666666666666667, "steam": 0.85, "composition": 0.8666666666666667 }, "reward": 0.7933333333333333, "sampleStdDev": 0.0014433756729740337, "judgments": [ { "noodles": { "score": 0.8, "reason": "Golden crimped noodles read clearly with wavy texture, layered overlap and shading that suggests volume; the white sprout strands on the left are slightly scribbly and the crimp pattern is somewhat repetitive, but the mass is coherent and sits believably in the broth." }, "toppings": { "score": 0.75, "reason": "Good variety: two nori sheets, veined leafy greens, bok choy halves, shiitake mushrooms, tofu cubes, scallion rings, corn, edamame and chili threads. Leaves and scallions read well; weaknesses are the plasticky faceted tofu cubes, the graphic star pattern stamped on mushroom caps, and bok choy pieces leaning without clear contact." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a continuous elliptical rim, visible wall thickness, tapering body, foot and table contact with shadow; rim highlight and interior reflections add material feel. Slight render-like stiffness but no broken joins or perspective mismatch." }, "broth": { "score": 0.75, "reason": "Warm miso-brown surface with a light gradient, scattered red oil droplets and pale specks; ingredients mostly appear seated with some occlusion. The liquid reads a bit flat and uniform, and immersion shading around the tofu and nori bases is shallow." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rise from the soup and spread without hiding the food. Comparing desktop-1/2/3 and pointer frames shows clearly changing plume shapes over time, indicating continuous motion rather than a static plume." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point on a dark wooden table with restrained bokeh, warm appetizing palette, no text or clutter, and steam providing vertical balance. Mobile frame leaves a large empty dark region above, a minor framing weakness." }, "vegan": { "score": 1.0, "reason": "All visible toppings are plant-based: nori, leafy greens, bok choy, shiitake, tofu, bean sprouts, scallions, corn, edamame and chili threads; no egg, meat or seafood appears." } }, { "noodles": { "score": 0.7, "reason": "Golden crinkled noodle mass with wavy strands, believable thickness and a slight domed volume; however the crinkle pattern is fairly uniform and reads a bit like a flat textured patch, with the left edge blending into the bean sprout tangle." }, "toppings": { "score": 0.8, "reason": "Rich variety with readable forms: two standing nori sheets, leafy greens with veins, diagonal-cut leek halves, three shiitake with star cuts, fried tofu cubes, bean sprouts, scallion rings, corn and edamame. Minor flaws: greens slightly leaf-generic, tofu cubes a touch blocky/glossy, corn cluster blobby." }, "bowl": { "score": 0.8, "reason": "Coherent dark lacquer bowl with elliptical copper rim, visible interior wall streaks, curved exterior, foot and soft contact shadow; rim highlights on left and lower right. Rim edge thins slightly on the right and the glossy streak there is a bit conspicuous." }, "broth": { "score": 0.75, "reason": "Appetizing miso-toned surface with chili-oil flecks, sesame, bubbles at the rim and a lighter elliptical sheen; most toppings sit with some occlusion, though mushrooms and tofu rest a little flat on the surface with weak contact shadows." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rising from the bowl; plume shapes clearly morph and drift across desktop-1/2/3 and pointer frames, spreading and fading without obscuring the food. Slightly broad smoke-like spread beyond the bowl." }, "composition": { "score": 0.9, "reason": "Centered focal dish on dark wood with warm bokeh, coherent amber/green palette, atmospheric lighting and no text or clutter; presentation is balanced and the food stays the focus." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: nori, shiitake, tofu, leek, bean sprouts, scallions, corn, edamame, greens and chili in a miso-like broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Golden curly noodle mass sits coherently in the broth with wavy strand texture and believable volume, but the pattern is somewhat repetitive/mechanical in places and reads more as a unified instant-noodle block than distinct supple strands." }, "toppings": { "score": 0.8, "reason": "Strong variety and readability: shiitake caps with star-score cuts, bok choy halves with veined leaves, golden tofu cubes with browned tops, corn, edamame, scallion rings, chili threads, enoki sprouts, and nori sheets with torn edges. Minor shortcomings: tofu cubes are slightly uniform/glossy and nori reads a bit stiff, but forms are clearly food, not plastic stand-ins." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with visible wall thickness on the lit left edge, believable interior depth, dark bronze/ceramic material with a soft sheen, and a centered foot base contacting the table. Perspective is consistent; only minor softness where the nori meets the back rim." }, "broth": { "score": 0.8, "reason": "Warm brown surface with a gentle back-glow, scattered oil droplets and chili flecks; mushrooms, noodles and toppings show contact shading and occlusion, so they read as sitting in the liquid rather than pasted on. Restrained sheen, appetizing and readable." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps rise from the bowl and clearly change shape across the time-ordered desktop frames (desktop-1/2/3 and pointer frames show different plume configurations), rising, spreading and fading without obscuring the food; the reduced-motion frames freeze it as expected. Slightly dense in the upper area but convincingly animated." }, "composition": { "score": 0.85, "reason": "Bowl is a clear focal point on a dark wood table with tasteful bokeh, warm coherent lighting, and no text or clutter; mobile crop keeps the dish well-framed. Background atmosphere supports rather than replaces food quality." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, bok choy/leafy greens, corn, edamame, scallions, chili, enoki, nori and noodles in a brown broth \u2014 all plausibly plant-based, with no animal-derived toppings." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9855555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "10f744351317188ecb13c940307c227b36b524d27c88389f6061f3c84d4cbb2a", "gradedAt": "2026-09-30T22:44:45.520492+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6577777777777778, "toppings": 0.7077777777777778, "bowl": 0.8255555555555556, "broth": 0.6855555555555556, "steam": 0.7411111111111112, "composition": 0.8122222222222222 }, "reward": 0.7345555555555556, "runId": "anthropic/sonnet-5.5/high", "rewardStdDev": 0.0747433262225787, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.65, "bowl": 0.87, "broth": 0.66, "steam": 0.68, "composition": 0.85 }, "reward": 0.6895, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0.5, "reason": "Numerous continuous cream strands have clear overlaps and shading, but angular, closely stacked ribbon-like bends dominate a rather flat mound." }, "toppings": { "score": 0.65, "reason": "Scallion rings, corn, split-pattern mushroom caps and green leaves provide readable variety. Leaves are thick and uniformly oval, while the vertically striped nori resembles a rigid card rather than thin seaweed." }, "bowl": { "score": 0.87, "reason": "Coherent elliptical rim, visible inner wall, curved body and integrated foot convey convincing depth. Gold edging, a red band and subtle scalloped decoration enrich the lacquer finish; ground contact is slightly weak." }, "broth": { "score": 0.66, "reason": "Warm brown liquid has a restrained gradient, scattered glints and small surface rings. Ingredient bases are partly occluded, though broad untextured areas and limited contact effects make the soup somewhat flat." }, "steam": { "score": 0.68, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the food. The steam spreads irregularly, but reads as soft blurry puffs rather than fine rising wisps; continuous flow cannot be established from these samples." }, "composition": { "score": 0.85, "reason": "A single readable dish remains the focus against a restrained dark amber background, with harmonious green, cream and red accents and no text or UI. Mobile framing retains the full bowl; desktop placement is somewhat low with ample unused space above." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, scallions, corn, mushrooms, seaweed and golden rectangular toppings plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Numerous readable cream strands overlap, but angular bends and tightly stacked, ribbon-like paths make the mound look rigid rather than supple." }, "toppings": { "score": 0.65, "reason": "Scallion rings, corn, leafy greens and scored mushroom caps provide clear variety. Leaves are overly smooth and oval, mushrooms are schematic, and the nori resembles a stiff striped card." }, "bowl": { "score": 0.87, "reason": "Coherent elliptical rim, inset cavity, tapered walls and integrated foot establish convincing depth. Gold edging, red band and subtle scalloped decoration enrich the lacquer finish; ground contact is slightly weak." }, "broth": { "score": 0.66, "reason": "Warm brown shading, small highlights and scattered surface rings suggest liquid. Limited meniscus and immersion cues make several toppings appear placed on a smooth surface rather than nestled into soup." }, "steam": { "score": 0.68, "reason": "Normal frames show changing translucent vapor rising above the soup without hiding the food. Broad blurry puffs read more as a cloud than fine, tapering wisps; continuous fluidity cannot be established from these snapshots." }, "composition": { "score": 0.85, "reason": "A single unobstructed bowl anchors the dark scene, with warm amber lighting and green accents creating clear focus on desktop and mobile. No text or UI competes; the desktop placement leaves somewhat excessive space above." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, scallions, corn, mushrooms and seaweed plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Cream-colored strands have readable overlap and shading, but the mound is dominated by angular, flattened ribbon loops rather than supple bends." }, "toppings": { "score": 0.65, "reason": "Mushroom cap cracks, scallion rings, corn and leafy greens provide recognizable variety. Leaves are overly smooth and oval, while the nori resembles a rigid striped card." }, "bowl": { "score": 0.87, "reason": "Nested elliptical rims, a shaded inner cavity, curved patterned walls and a connected foot establish coherent depth. Gold edging and restrained lacquer highlights are polished; ground contact is slightly faint." }, "broth": { "score": 0.66, "reason": "Warm brown shading, small surface rings and scattered highlights suggest liquid. Limited meniscus and immersion cues make several ingredients appear placed on top of a flat surface." }, "steam": { "score": 0.68, "reason": "Normal frames show changing translucent clouds rising above the soup without obscuring the food. The broad blurred puffs are less convincing than fine, tapering irregular wisps." }, "composition": { "score": 0.85, "reason": "A single clearly readable bowl anchors a restrained dark scene, with harmonious gold, green and brown colors and no distracting interface or text. Desktop placement is somewhat low with excessive upper empty space." }, "vegan": { "score": 1, "reason": "Visible mushrooms, greens, scallions, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived topping is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6733333333333333, "toppings": 0.6733333333333333, "bowl": 0.7566666666666667, "broth": 0.6133333333333333, "steam": 0.6933333333333334, "composition": 0.7366666666666667 }, "reward": 0.6933333333333334, "sampleStdDev": 0.013051181300301274, "judgments": [ { "noodles": { "score": 0.68, "reason": "Noodle mound has many wavy, overlapping cream strands with readable volume and shading. The strands look somewhat uniform and flat, and the mound sits on top of the broth like a pasted nest rather than blending in." }, "toppings": { "score": 0.68, "reason": "Shiitake with cracked caps, bok choy leaves, corn kernels, scallion rings, sesame seeds, and a jagged nori sheet are all recognizable. The nori is a tall rigid slab, the bok choy is glossy and cartoonish, and the tofu/bamboo is barely legible." }, "bowl": { "score": 0.75, "reason": "Dark lacquer bowl with a red band, gold trim, a scale pattern, an inner rim, and a foot. The perspective is consistent and the highlights are convincing. The base is slightly dark and blends into the background." }, "broth": { "score": 0.62, "reason": "Warm brown broth with a slight sheen and small oil specks. It reads as a liquid, but it is fairly flat, and some toppings look placed on top rather than immersed." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps rise and shift shape between frames, and they drift with the pointer. The plume is somewhat blobby and cloudy, and it partly overlaps the nori." }, "composition": { "score": 0.72, "reason": "The bowl is centered as a clear focal point in a dark, warm atmosphere with subtle embers. The bowl sits low in the frame, leaving a lot of empty space above it." }, "vegan": { "score": 0.93, "reason": "Only mushrooms, bok choy, corn, scallions, nori, noodles, and possibly tofu or bamboo are visible. There is no egg or meat." } }, { "noodles": { "score": 0.66, "reason": "The noodle mound has wavy cream strands with overlap and volume, and it reads as ramen. The strands are fairly uniform and small, the pile looks a bit flat, and the scallion bits sit on top like stickers." }, "toppings": { "score": 0.66, "reason": "Bok choy, shiitake with cracked caps, corn kernels, scallion rings and a tall nori sheet are all recognizable and varied. The nori is a rigid upright slab with a clean zigzag top edge, the bok choy leaves look glossy and plastic, and the tofu or bamboo piece is ambiguous." }, "bowl": { "score": 0.74, "reason": "The dark lacquer bowl has a coherent elliptical rim, a red and gold band, a scale pattern, highlights and a foot. Perspective is consistent and the base sits convincingly on the surface. The body looks a little shallow and the lower part is dark and low in detail." }, "broth": { "score": 0.6, "reason": "The broth is warm brown with a glossy sheen, and the toppings sit in it with small ripples. It is fairly flat and opaque, and the depth and immersion contact are weak." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps rise from the bowl and change shape and drift across frames, with no hard edges. The plume is a bit blobby and cloud-like, and it partly overlaps the nori." }, "composition": { "score": 0.74, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm lighting and subtle embers. There is no clutter. The bowl sits somewhat low and the background is fairly empty." }, "vegan": { "score": 0.92, "reason": "Only vegetables, mushrooms, corn, nori and noodles are visible, with no meat or egg. The pale tofu or bamboo piece is slightly ambiguous." } }, { "noodles": { "score": 0.68, "reason": "The noodle mound has supple, overlapping wavy strands with cream color and plausible volume. It reads a bit flat and uniform, and the strands sit on top of the broth rather than dipping into it." }, "toppings": { "score": 0.68, "reason": "The toppings are varied and recognizable: bok choy, shiitake with cracked caps, corn, scallion rings, sesame, a tall nori sheet and what looks like tofu or bamboo. The nori is a rigid, card-like slab with straight vertical streaks, and the bok choy reads slightly glossy and cartoonish." }, "bowl": { "score": 0.78, "reason": "The bowl has a coherent elliptical rim, a dark lacquer body with a red band and gold lines, a scale pattern and a footed base. The highlights and perspective are consistent. The rim is slightly doubled, and the base contact shadow is faint." }, "broth": { "score": 0.62, "reason": "The broth is a warm brown with some gradient and a few small oil ripples. It looks fairly flat and opaque, and the ingredients sit on the surface with little immersion or reflection." }, "steam": { "score": 0.72, "reason": "Soft translucent wisps rise above the bowl and change shape across frames, drifting and fading. The plume is somewhat blobby and sits in a wide cloud that partly overlaps the nori." }, "composition": { "score": 0.75, "reason": "A centered bowl sits against a dark, warm atmospheric background with subtle embers, and the food is the clear focal point. The bowl sits low in the frame, leaving a lot of empty space above it. There is no clutter or text." }, "vegan": { "score": 0.95, "reason": "Only plant ingredients are visible: mushrooms, bok choy, corn, scallions, nori, noodles and possibly tofu. No meat or egg is visible." } } ], "diagnostics": { "vegan": 0.9333333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.8, "toppings": 0.8, "bowl": 0.85, "broth": 0.7833333333333333, "steam": 0.85, "composition": 0.85 }, "reward": 0.8208333333333334, "sampleStdDev": 0.0028867513459481316, "judgments": [ { "noodles": { "score": 0.8, "reason": "Central nest of cream-colored strands with wavy bends, outlined detail and shaded volume; reads convincingly stylized, though some strands look repetitive and the mound edge is slightly flat." }, "toppings": { "score": 0.8, "reason": "Good variety: bok choy with white stems, two cross-hatched shiitake slices, scallion rings, corn kernels, sesame, and a standing nori sheet with fold lines. Nori is a bit card-like and the beige tofu/menma block is partly occluded, but shapes are organic and readable." }, "bowl": { "score": 0.85, "reason": "Coherent ellipse rim with visible wall thickness, navy body, red band, gold pinstripe, seigaiha pattern, footed base with contact shadow and side highlight; perspective stays consistent across frames with subtle pointer parallax." }, "broth": { "score": 0.8, "reason": "Amber liquid surface with warm gradient, light sheen near the noodles, scattered oil droplets and specks; ingredients appear seated with contact rather than pasted on, though surface texture is slightly simplified." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps rise from the soup and clearly change shape, height and density between desktop-1, desktop-2/3 and pointer frames; food stays visible and the plume is irregular rather than a mechanical column." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark atmospheric background with restrained warm bokeh and a subtle ground line; no text, UI or clutter, and the food remains the emphasis." }, "vegan": { "score": 1.0, "reason": "Visible toppings are bok choy, shiitake, scallion, corn, sesame, nori and a tofu/menma-style block with plant broth; no meat, fish, egg or dairy appears anywhere in the scene." } }, { "noodles": { "score": 0.8, "reason": "A cohesive pale noodle nest with wavy, varied strands, shading gaps and scallion garnish; strands drape and dip toward the broth at the edges. Slightly matted/pancake-like mass with some repetitive wave shapes keeps it from the top tier." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: two shiitake slices with star-cut gills, bok choy-style leaves with veins and stalks, corn clusters, scattered scallion rings, sesame specks and menma-like bamboo pieces. Defect: the nori sheet stands as a stiff, card-like panel with rigid vertical ridges and a hard wavy top edge." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: smooth elliptical rim, tan inner wall, red and gold bands, subtle seigaiha pattern on the body, visible foot with contact shadow and a plausible exterior highlight; perspective stays consistent across all frames. No material construction defect observed." }, "broth": { "score": 0.75, "reason": "Brown broth reads as liquid with tonal variation, a lighter oil/reflection patch, sesame specks, and ingredients sitting at the surface with contact. Somewhat uniform and flat in areas, and the noodle mound sits on top rather than clearly immersed." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rise from the soup and clearly evolve across the time-ordered desktop and pointer frames (drifting and reshaping between 1.0s, 3.7s, 6.3s, 11s, 15.6s), fading without obscuring the food; not opaque fog or identical plumes." }, "composition": { "score": 0.85, "reason": "Centered, well-proportioned focal bowl on a warm dark atmospheric ground with restrained ember bokeh, coherent lighting, no text/UI/clutter, and no horizontal overflow on desktop or mobile per measurements." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, scallions, shiitake mushrooms, leafy greens, corn, nori and bamboo-style pieces; no egg, meat, fish or dairy appears." } }, { "noodles": { "score": 0.8, "reason": "Cream-colored strands are continuous, supple and individually drawn with varied bends forming a coherent mound; slight uniform repetition of S-curves and mild tangling keeps it below reference quality." }, "toppings": { "score": 0.8, "reason": "Cross-cut shiitake slices, leafy bok choy with pale stems, scallion rings, corn, sesame and a striped torn-edge nori sheet are all recognizable with organic shapes; corn and leaves have a slightly glossy plastic sheen and the nori is a touch oversized." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with visible inner wall, tapered dark body with red band and wave pattern, supported footed base with contact shadow; construction is consistent with no broken joins, only mild stylized flattening." }, "broth": { "score": 0.8, "reason": "Amber liquid surface with soft sheen gradient and subtle highlight swirls; mushrooms and vegetables sit with believable contact and occlusion, though the surface reads slightly flat and uniform in places." }, "steam": { "score": 0.85, "reason": "Across the normal desktop frames the translucent grey wisps change shape, rise and dissipate continuously without hiding the food; not opaque fog or a static plume, though wisps get slightly blobby mid-plume." }, "composition": { "score": 0.85, "reason": "Centered focal dish in a restrained dark atmospheric scene with warm ember particles and coherent lighting; balance is pleasant, bowl slightly small relative to frame but no clutter, text or competing elements." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, bok choy, mushrooms, scallions, corn, sesame, nori and what appears to be menma; no meat, fish, egg or dairy forms are present." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9777777777777777, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 133.35000000000036, "frameIntervalP95Ms": 150, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "496f05463ca742a6cabee848cd64a453360ccb70bb4b87f35bfa9ded0dec91cc", "gradedAt": "2026-09-30T22:44:45.344701+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5711111111111111, "toppings": 0.6766666666666666, "bowl": 0.7544444444444445, "broth": 0.6466666666666667, "steam": 0.6544444444444445, "composition": 0.7855555555555556 }, "reward": 0.6756666666666666, "runId": "anthropic/sonnet-5.5/low", "rewardStdDev": 0.08644571193015368, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.64, "bowl": 0.8466666666666667, "broth": 0.6066666666666667, "steam": 0.5633333333333334, "composition": 0.8366666666666667 }, "reward": 0.6541666666666667, "sampleStdDev": 0.004193248541803063, "judgments": [ { "noodles": { "score": 0.48, "reason": "Clearly outlined golden strands overlap beneath toppings, but angular zigzags, blunt ends and mostly flat paths look more like rigid tubing than supple noodles." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, speckled tofu, green leaves and scallion rings provide readable variety. Shapes remain simplified, with a perfectly rectangular striped nori card and little organic surface texture." }, "bowl": { "score": 0.84, "reason": "A continuous elliptical rim, coherent curved walls, subtle lacquer highlights and attached foot establish convincing depth and support. The dark body loses some material detail, but no material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm amber shading and scattered oil droplets suggest appetizing soup, but the surface is fairly flat and opaque, with weak immersion cues around the noodles and toppings." }, "steam": { "score": 0.56, "reason": "Normal frames show translucent haze increasing and spreading above the soup without obscuring the dish. It reads as broad blurred clouds rather than distinct irregular rising wisps; continuous fading is not established." }, "composition": { "score": 0.83, "reason": "The single dish is clearly framed against a restrained dark background, with warm broth and bright greens creating a strong focal contrast and no clutter. Mobile framing leaves substantial empty space above the smaller bowl." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Clearly outlined strands overlap beneath toppings, but angular zigzags, blunt ends and flat parallel runs read more as rigid piping than supple noodles." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, speckled tofu, leafy greens and scallion rings provide readable variety. Shapes remain simplified, and the perfectly rectangular, ruled nori looks like a rigid card." }, "bowl": { "score": 0.85, "reason": "A continuous elliptical rim, coherent curved walls, subtle lacquer highlights and grounded foot establish convincing depth and support. No material construction defect is observed; material shading is somewhat restrained." }, "broth": { "score": 0.6, "reason": "Warm amber coloring, scattered oil droplets and subtle surface arcs suggest soup, but the surface is fairly flat and ingredient edges show little immersion or liquid contact." }, "steam": { "score": 0.55, "reason": "Normal frames show translucent haze expanding above the soup without obscuring it. It reads as broad soft glow or fog rather than distinct irregular rising wisps, and the captures do not establish continuous rise and fading." }, "composition": { "score": 0.83, "reason": "The single dish is clearly framed against a restrained dark background, with warm broth and green toppings creating a readable focal contrast and no distracting text. Mobile framing leaves considerable empty space above the dish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy greens, scallions, corn and seaweed; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.48, "reason": "Clearly outlined golden strands overlap beneath toppings, but mostly form flat, angular zigzags with blunt ends rather than supple, volumetric noodles." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, speckled tofu, green leaves and corn provide readable variety. Scallions are repetitive disks, leaves have limited texture, and the perfectly rectangular striped nori resembles a rigid card." }, "bowl": { "score": 0.85, "reason": "A continuous elliptical rim, coherent curved walls, attached foot and grounding shadow convincingly establish a supported vessel. Restrained highlights suggest dark lacquer; the front wall is somewhat uniformly shaded." }, "broth": { "score": 0.61, "reason": "Warm amber shading and scattered oil droplets suggest appetizing soup, but the surface feels relatively flat and opaque, with weak immersion cues around noodles and toppings." }, "steam": { "score": 0.58, "reason": "Normal frames show translucent haze expanding upward from the rear of the soup without concealing the dish. It reads more as broad diffuse glowing fog than distinct irregular steam wisps; continuous rise and fading are not clearly established." }, "composition": { "score": 0.85, "reason": "The single dish is clearly framed against a restrained dark brown setting, with warm broth and green toppings creating a strong focal contrast and no distracting UI. Mobile framing preserves the bowl but leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy greens, scallions, corn and seaweed alongside noodles. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6066666666666667, "bowl": 0.6, "broth": 0.5833333333333334, "steam": 0.6, "composition": 0.6866666666666666 }, "reward": 0.602, "sampleStdDev": 0.011532562594670805, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles are cream strands with outlines that bend and overlap, and they shift between frames. They look a bit flat and outlined, with some repeated short curved segments, and the mass is partly obscured by toppings." }, "toppings": { "score": 0.62, "reason": "Shiitake with slits, tofu cubes, bok choy leaf, scallion rings, corn and a nori sheet are all recognizable and varied. The nori is a rigid, straight-edged rectangle, the scallion rings are flat discs, and the overall style is vector-clean rather than organic." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark wall and a foot, and the perspective is consistent. The walls are murky and flat, the base is barely readable, and a slight rim/body mismatch shows near the left and right edges." }, "broth": { "score": 0.6, "reason": "The orange-brown broth has a warm gradient and a chili oil speckle texture. It looks somewhat flat and opaque, and the ingredients sit on it with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Soft translucent steam rises from the bowl and grows larger and drifts across the frames, and it reacts to the pointer. It is diffuse and hazy, close to a blurry glow, and lacks distinct wisps." }, "composition": { "score": 0.7, "reason": "The bowl is centered as the clear focal point against a dark vignette, with warm, appealing colors. The nori sticks out awkwardly, and the lower part of the bowl merges into the dark background." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens, corn, nori and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.55, "reason": "Noodles are cream strands with outlines and some organic bends, and they shift between frames. They read as short flat ribbons with repeated similar curves, and several segments look clipped. The mass is not very voluminous." }, "toppings": { "score": 0.6, "reason": "Shiitake with radial cuts, tofu cubes, a bok choy leaf, scallion rings, corn and a nori sheet are all readable and vegan. The nori is a rigid rectangle standing upright, the scallions are flat disc rings, and the corn is glossy. Variety is good but several items look generic." }, "bowl": { "score": 0.6, "reason": "The rim, wall and foot are coherent with a dark lacquer feel and a consistent perspective. The body is very dark and flat, the base is a thin plate, and the wall has little highlight detail, so depth reads weakly." }, "broth": { "score": 0.55, "reason": "Warm orange broth with a glow and chili oil spots looks appetizing. It is fairly flat, with little ripple or depth, and the toppings sit on it without much immersion or contact." }, "steam": { "score": 0.6, "reason": "Soft translucent steam rises and drifts in the later frames and grows over time, and it does not hide the food. It is a diffuse glow rather than defined wisps, and it is faint in the early frames." }, "composition": { "score": 0.68, "reason": "The bowl is centered and dominant against a dark atmospheric background, with warm lighting and no clutter. The bowl is cropped a little low, and the upright nori competes for attention." }, "vegan": { "score": 0.95, "reason": "The toppings are mushrooms, tofu, vegetables, corn and nori, with no visible animal products." } }, { "noodles": { "score": 0.55, "reason": "Noodles are tan strands with outlines that overlap and bend in places, and they read as ramen. Many look like short flat, stubby ribbons with uniform outlines, and the mass is somewhat tangled and indistinct." }, "toppings": { "score": 0.6, "reason": "The toppings are varied and recognizable: shiitake with cut marks, tofu cubes, scallion rings, a bok choy leaf, corn and nori. The nori is a rigid rectangular card standing upright, the scallions look like flat rings, and the tofu is blocky, so the set feels somewhat generic." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark lacquer body, a base, and some highlights. The wall is somewhat flat and the base looks slightly detached, with a strange dark foot and halo beneath it. Overall it is solid but plain." }, "broth": { "score": 0.6, "reason": "The broth is an orange-brown surface with chili oil spots and warm glow, and the ingredients sit in it believably. It looks fairly flat and opaque, with limited depth and little immersion detail." }, "steam": { "score": 0.6, "reason": "Soft translucent haze rises from the bowl and grows across the frames, drifting with the pointer. It is diffuse and somewhat foggy, without defined wisps, but it does not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with warm lighting and good color contrast. The food is the focus and there is no clutter. The nori and the bowl's dark base are slightly awkward, and the top of the frame is empty." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, bok choy, scallions, corn, nori and noodles. There is no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.7833333333333333, "bowl": 0.8166666666666667, "broth": 0.75, "steam": 0.8, "composition": 0.8333333333333334 }, "reward": 0.7708333333333334, "sampleStdDev": 0.02982588361362213, "judgments": [ { "noodles": { "score": 0.65, "reason": "Cream strands curve, overlap and bend plausibly in the broth with consistent thickness, but they read as many short rounded squiggle segments with visible gaps rather than a continuous coherent noodle mass." }, "toppings": { "score": 0.75, "reason": "Good variety and readable forms: shiitake caps with gill marks, a leafy green with sesame seeds, scallion rings, speckled tofu slabs, corn kernels and a striped nori sheet. Defects: the nori stands as a stiff rectangular card, and the mushrooms/tofu are very smooth and simplified." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, deep tapering dark ceramic body with a subtle side sheen, consistent elevated perspective, and a small foot contacting the surface. The foot is a slightly blocky rectangle and the body is quite plain, but nothing is detached or distorted." }, "broth": { "score": 0.7, "reason": "Orange-brown liquid surface with a warm center gradient, scattered oil droplets and believable ingredient immersion; toppings sit in the broth rather than floating above. The surface is somewhat flat and uniform toward the edges with limited texture." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps that clearly change across the normal frames: minimal at ~1s, rising plumes on the left by ~3-5s, and taller drifting plumes in the pointer frames. They never hide the food. Wisps are slightly blobby but fade and spread believably; reduced-motion frames are appropriately static." }, "composition": { "score": 0.8, "reason": "Warm glowing dish centered in a dark atmospheric scene, appetizing color palette, no text or clutter, and the food stays the focus. The mobile layout leaves much empty dark space with a small dish, a minor balance issue." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-based: mushrooms, leafy greens, scallions, tofu, corn, nori and sesame; no animal-derived toppings appear." } }, { "noodles": { "score": 0.65, "reason": "Noodles are curvy, supple cream strands of plausible thickness with organic bends, but they read as somewhat flat squiggle segments lying on the surface; several strands end abruptly and there is little coherent overlap or volume/nest structure." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: scored shiitake caps, scallion rings with white cores, tofu blocks, corn kernels, sesame, chili oil dots, a leafy green, and an upright striped nori sheet. Shapes are stylized but recognizable; corn beads are slightly bead-like and the leaf is simple." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer bowl with a consistent rim ellipse, subtle rim highlight, tapering body, visible foot, and grounded contact shadow; the 3/4 perspective holds together with no broken joins. Interior wall thickness is a bit thin." }, "broth": { "score": 0.75, "reason": "Warm orange-brown broth with a believable radial glow, oil droplets, and sesame; ingredients mostly sit in the liquid with reasonable contact. Surface texture is limited and the lit center looks slightly like a projected glow rather than liquid depth." }, "steam": { "score": 0.8, "reason": "Across time-ordered desktop frames the wisps clearly evolve: faint at 1s, rising plumes by 3s, spreading leftward and drifting through pointer frames. Steam is translucent, irregular, and never obscures the food; motion is supported by the frames." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric presentation with the bowl as clear focal point, warm rim lighting against deep brown background, no clutter or competing elements. Mobile framing places the bowl low with ample empty headroom but keeps it readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, scallions, corn, leafy green, nori, sesame and chili oil in plant broth; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.7, "reason": "Stylized pale strands curve with plausible bends and coherent overlaps across the broth, but they are uniformly thick flat ribbons with rounded caps and repeated squiggle shapes rather than supple volumetric strands." }, "toppings": { "score": 0.8, "reason": "Good variety: scored shiitake caps, a bok choy leaf with vein, tofu blocks, scallion rings, corn kernels, sesame and chili oil all read clearly; the standing nori is slightly rigid/card-like and scallions are flat rings." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer bowl with consistent rim ellipse, visible interior wall, tapering body, foot and contact shadow; rim highlight reads as ceramic/lacquer. No material distortion observed." }, "broth": { "score": 0.8, "reason": "Warm orange-brown liquid surface with soft gradients, floating oil droplets and sesame; toppings sit in and are partially occluded by the broth. The bright central hotspot reads as lighting rather than a broth defect." }, "steam": { "score": 0.8, "reason": "Translucent wisps rise and shift across desktop-1\u21923 and the pointer frames, growing and drifting left/right without hiding food; slightly blobby and occasionally detached, but clearly time-varying, not static." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark warm backdrop with restrained glow; no text or clutter. Mobile frame leaves large empty dark space above the small bowl, slightly reducing balance." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, bok choy, tofu, scallions, corn, nori, sesame and chili oil in plant broth; no animal-derived ingredients apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100.0, "frameIntervalP95Ms": 116.59999999999854, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "882ebcd22ff853c115b0e0de8f8cf6bd4252bb0fa23414447f90c3537abacb37", "gradedAt": "2026-09-30T22:44:43.648473+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6466666666666667, "toppings": 0.6644444444444445, "bowl": 0.8166666666666667, "broth": 0.5855555555555556, "steam": 0.6788888888888889, "composition": 0.8055555555555556 }, "reward": 0.6997222222222222, "runId": "anthropic/sonnet-5.5/medium", "rewardStdDev": 0.0688017145565859, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6066666666666667, "bowl": 0.8533333333333333, "broth": 0.54, "steam": 0.6666666666666667, "composition": 0.8333333333333333 }, "reward": 0.6596666666666666, "sampleStdDev": 0.0037859388972001857, "judgments": [ { "noodles": { "score": 0.5, "reason": "Golden strands have clear thickness and curved paths, but repeated flat elliptical loops dominate, with little convincing overlap or submerged volume." }, "toppings": { "score": 0.61, "reason": "Veined greens, scored mushroom caps, tofu cubes and scallion rings are readable and varied. Shapes remain schematic, especially the perfectly rectangular striped nori and smooth disk-like mushrooms." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, curved walls, continuous red bands and small attached foot form coherent depth and perspective. Soft highlights convey dark lacquer; grounding beneath the foot is weak." }, "broth": { "score": 0.52, "reason": "Warm amber liquid is visible around the strands with small surface accents, but limited sheen and immersion make the food look arranged on a flat surface rather than sitting in soup." }, "steam": { "score": 0.67, "reason": "Normal frames show changing translucent haze above the soup without obscuring the food. The vapor is broad and cloud-like, with weakly defined rising wisps and emergence points." }, "composition": { "score": 0.83, "reason": "A clear single-dish focal point, warm gold and green accents, restrained dark background and absence of interface clutter create an attractive presentation. Desktop placement is somewhat bottom-heavy, with substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy greens, seaweed, scallions and corn alongside noodles; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have rounded bends and highlights, but repeated broad, flat loops dominate, with little convincing piled volume or submerged depth." }, "toppings": { "score": 0.61, "reason": "Veined greens, scored mushroom caps, scallion rings and shaded tofu provide readable variety. Shapes remain schematic, especially the perfectly rectangular, rigid nori and smooth disk-like mushrooms." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, substantial walls, curved red bands and smoothly shaded lacquer body establish convincing depth. The small foot joins the body coherently; ground contact is weakly defined." }, "broth": { "score": 0.55, "reason": "Warm amber liquid is visible around the strands with small reflective accents, but its surface is mostly flat and ingredients show limited immersion or liquid contact." }, "steam": { "score": 0.67, "reason": "Translucent vapor changes position and density across normal frames and spreads above the soup without obscuring it. It reads more as broad diffuse cloud patches than distinct rising, tapering wisps." }, "composition": { "score": 0.84, "reason": "The uncluttered dish remains readable against a dark warm background, with restrained red, green and gold accents and consistent lighting. Mobile framing preserves the whole bowl; desktop placement is somewhat bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict greens, mushrooms, tofu, scallions, corn and seaweed, with noodles and plant-compatible broth. No visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable thickness and highlights, but repeated broad flat loops dominate, with little convincing piled volume or varied supple overlap." }, "toppings": { "score": 0.6, "reason": "Veined greens, scored mushroom caps, tofu cubes and scallion rings are recognizable and varied. Shapes remain schematic, especially the perfectly rectangular striped nori and smooth, disk-like mushrooms." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, inner cavity, curved walls and small integrated foot form coherent geometry. Soft lacquer highlights and red bands describe the curvature well; the base contact is somewhat understated." }, "broth": { "score": 0.55, "reason": "Warm amber liquid is visible around the strands with small sheen marks, but the surface reads mostly flat and ingredients show limited immersion or liquid contact." }, "steam": { "score": 0.66, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the dish. The broad blurred patches suggest heat, but lack clearly articulated rising wisps and read partly as ambient fog." }, "composition": { "score": 0.83, "reason": "A single readable dish anchors the dark scene, with restrained warm lighting, green accents and no distracting UI or text. Mobile preserves the full bowl, though both layouts leave considerable empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings read as mushrooms, tofu, leafy greens, scallions, corn and seaweed; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.68, "toppings": 0.6266666666666667, "bowl": 0.7566666666666667, "broth": 0.48333333333333334, "steam": 0.62, "composition": 0.75 }, "reward": 0.6603333333333334, "sampleStdDev": 0.017925772879665017, "judgments": [ { "noodles": { "score": 0.68, "reason": "Dense tangle of curved strands with varied overlap and highlights reads as noodles. Strands are a bit uniform and flat, with some repeated loops, and the mass lacks depth." }, "toppings": { "score": 0.62, "reason": "Bok choy leaves, scored shiitake, tofu cubes, scallion rings, chili flecks and corn are all recognizable. The nori is a stiff rectangular card standing upright and the tofu cubes are small and blocky, so it looks somewhat artificial." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, curved walls, red lacquer bands, highlights and a base with a shadow. The bowl is a bit dark and the lower base looks slightly cut off, but there is no major distortion." }, "broth": { "score": 0.45, "reason": "Little broth is visible because noodles cover nearly the whole surface. There is only a brown edge near the rim, and some faint ripple rings appear in one frame. Toppings sit on the noodles rather than in liquid." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape across frames. The effect is somewhat diffuse and foggy, and it is partly detached from the bowl surface." }, "composition": { "score": 0.75, "reason": "Centered bowl with a dark moody background, warm bokeh and good color harmony. The food is the clear focus and there is no clutter, though the bowl sits somewhat low." }, "vegan": { "score": 0.93, "reason": "Only vegetables, mushrooms, tofu, nori and noodles are visible, with no animal products. The corn-like yellow bits are plant-based." } }, { "noodles": { "score": 0.68, "reason": "Dense, overlapping golden strands with soft highlights and plausible swirl volume. Strands are fairly uniform and look like repeated loops, and there is little depth of immersion." }, "toppings": { "score": 0.66, "reason": "Bok choy, scored shiitake, tofu cubes, nori, scallion and chili are all readable, and the bok choy leaves have veins. The nori is a stiff flat rectangle standing upright, the tofu cubes are plain, and the mushrooms are slightly flat discs." }, "bowl": { "score": 0.78, "reason": "Dark lacquer bowl with red bands, a coherent elliptical rim, a curved body and a base shadow, plus a soft highlight. Perspective is consistent. The base is slightly dark and indistinct, and the rim could be thicker." }, "broth": { "score": 0.5, "reason": "Little broth is visible because noodles cover the surface. There is a brown edge near the rim but no real sheen or immersion, and toppings sit on top of the noodles." }, "steam": { "score": 0.66, "reason": "Soft translucent wisps drift above the bowl and change shape between frames. They are fairly diffuse and foggy, and partly blend into the bokeh background." }, "composition": { "score": 0.75, "reason": "The bowl is centered as the clear focal point against a dark, warm atmosphere with restrained ember bokeh. Colors are appetizing and there is no text or clutter. The bowl is a little low in the frame and the toppings are off-balance." }, "vegan": { "score": 0.93, "reason": "Only noodles, mushrooms, tofu, bok choy, nori, scallion, chili and corn are visible. No egg, meat or fish." } }, { "noodles": { "score": 0.68, "reason": "The noodles form a dense swirl of many overlapping, slightly curved strands with highlights, which reads as a convincing noodle mass. They are somewhat uniform and flat-looking, and they lack depth and individual lift." }, "toppings": { "score": 0.6, "reason": "The bok choy leaves are organic and readable. The shiitake caps have a scored cross pattern. The tofu cubes are recognizable, and there are scallion rings, chili flecks and corn. The nori is a stiff rectangular card standing upright, which looks rigid. The mushrooms and tofu are a bit simple and flat." }, "bowl": { "score": 0.74, "reason": "The dark lacquer bowl has a coherent elliptical rim, red bands, a curved body with a soft highlight and a base with a contact shadow. There is no material perspective break. The lower body fades into darkness and the foot is barely defined." }, "broth": { "score": 0.5, "reason": "Little broth is visible because the noodles cover nearly the whole surface. Only a brown edge near the rim shows, with a faint sheen. The toppings sit on top of the noodles rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape across frames, drifting and fading without hiding the food. They are somewhat diffuse and foggy, and they are not clearly tied to the bowl surface." }, "composition": { "score": 0.75, "reason": "A centered bowl sits in a dark, warm atmosphere with subtle ember bokeh, and it is clearly the focal point. The colors are harmonious and there is no distracting text or UI. The nori and bok choy stick out a bit awkwardly." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are noodles, mushrooms, tofu, bok choy, nori, scallions and corn. There are no egg, meat or fish items." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.76, "toppings": 0.76, "bowl": 0.84, "broth": 0.7333333333333333, "steam": 0.75, "composition": 0.8333333333333334 }, "reward": 0.7791666666666667, "sampleStdDev": 0.021577380131362917, "judgments": [ { "bowl": { "reason": "Coherent lacquer-style bowl: elliptical rim with visible inner edge, smoothly curved dark body with red stripe bands following the curvature, glossy highlight, and an attached foot with a grounding shadow. Perspective stays consistent across frames, including the slight parallax tilt in pointer frames; no broken joins or distortion observed.", "score": 0.82 }, "broth": { "reason": "Warm golden broth reads as liquid at the edges with amber pooling near the rim and subtle sheen; mushrooms and tofu carry small contact shadows so they sit in the surface rather than floating. The surface is mostly covered by the noodle carpet, so liquid depth is only thinly demonstrated, keeping it competent rather than excellent.", "score": 0.75 }, "composition": { "reason": "Well-balanced dark scene: the glowing bowl is the clear focal point against a warm brown backdrop with restrained ember particles and soft steam. Harmonious warm palette with green/red accents, no text or clutter, and the layout adapts cleanly to the narrow mobile viewport.", "score": 0.85 }, "noodles": { "reason": "Continuous tan strands with plausible thickness and curly bends, swirling coherently across the surface with slight edge shadowing that suggests volume at the broth line. Within the flat illustration style they read as noodles, though the swirl pattern is somewhat repetitive and uniform in tone, lacking strand separation variety in places.", "score": 0.8 }, "steam": { "reason": "Time-ordered frames (1039/3231/5253 ms) show the translucent gray wisps visibly shifting shape and position above the bowl, confirming animated rise-and-spread rather than static smoke. Wisps stay translucent and mostly irregular, though they lean toward puffy cloud blobs and slightly crowd the nori top.", "score": 0.75 }, "toppings": { "reason": "Good variety of readable vegan ingredients: a bok choy with layered leaves and pale base, two scored shiitake caps, tofu cubes with simple shading, a striped standing nori sheet, scattered scallion rings, chili threads and corn. Shapes are stylized and somewhat flat (nori is a plain rectangle, tofu is basic boxes) but none are generic plastic stand-ins.", "score": 0.8 }, "vegan": { "reason": "Visible toppings are bok choy, mushrooms, tofu, nori, scallions, chili and corn in plant broth; no egg, meat, fish or dairy anywhere.", "score": 1.0 } }, { "bowl": { "reason": "Coherent dark lacquer bowl: elliptical rim with visible inner wall, tapered body with red bands, foot and contact shadow, consistent perspective; only minor simplification in the foot.", "score": 0.85 }, "broth": { "reason": "Golden-brown surface with darker edge ring and restrained sheen; toppings sit on/in the liquid plausibly, though most of the surface is covered by the noodle mass so immersion cues are limited.", "score": 0.75 }, "composition": { "reason": "Centered, readable focal dish on an atmospheric dark backdrop with warm bokeh motes; no text or clutter; steam plume is somewhat dominant but the food stays the focus; mobile crop also holds together.", "score": 0.85 }, "noodles": { "reason": "Continuous wavy cream-colored strands form a believable tangled nest with coherent overlaps and some strand-end detail; slightly uniform wiggle pattern and a few repetitive swirl loops keep it below top marks.", "score": 0.78 }, "steam": { "reason": "Brownish translucent plume rises above the bowl and its blob configuration changes across desktop-1/2/3, so it animates; however it is fairly dense and fog-like in the core with blobby edges, though it never hides the food.", "score": 0.7 }, "toppings": { "reason": "Good variety: bok choy with leaf veins, scored shiitake caps, tofu cubes, striped nori sheet, scallion rings, chili strips and corn; shapes are recognizable but simplified, with a stiff rectangular nori card and uniform ring scallions.", "score": 0.78 }, "vegan": { "reason": "Visible toppings are bok choy, mushrooms, tofu, nori, scallion, chili and corn in plant broth; no animal-derived ingredients apparent.", "score": 1 } }, { "bowl": { "reason": "Coherent elliptical rim with visible thickness, smoothly curved lacquer-dark body with red stripe bands following the curvature, glossy highlight, and a small attached foot; perspective between rim and body is consistent across all frames with no detached parts.", "score": 0.85 }, "broth": { "reason": "Golden-tan broth reads as a rich soup with a darker ring near the rim; noodles and toppings sit on the surface with subtle shadows and sheen rather than floating, though the mostly noodle-covered surface leaves little distinct liquid texture.", "score": 0.7 }, "composition": { "reason": "Centered bowl on an atmospheric dark ember background with warm palette; steam fills the upper space without clutter, no text or UI, and the mobile crop keeps the dish focal. Slightly generous empty headroom above the bowl.", "score": 0.8 }, "noodles": { "reason": "Continuous golden strands with plausible thinness and bends forming a coherent nest across the surface; they lie fairly flat with a mildly decorative swirl pattern and some uniform repeated arcs, but read clearly as noodles, not scribbles or rigid ribbons.", "score": 0.7 }, "steam": { "reason": "Translucent gray wisps rise from the bowl and visibly change shape and position across desktop-1/2/3, spreading and fading without hiding the food; reduced-motion frames are correctly static. Wisps are somewhat blobby and fog-like rather than finely wisped.", "score": 0.8 }, "toppings": { "reason": "Good variety: bok choy leaves with white stems, two cross-cut shiitake, tofu cubes, standing nori sheet, scallion rings and chili flakes. Shapes are recognizable but stylized; the nori reads as a stiff striped card with perfectly straight edges and the mushrooms are simple flat disks.", "score": 0.7 }, "vegan": { "reason": "All visible toppings are plant-plausible: greens, shiitake, tofu, nori, scallion, chili; the small yellow cluster most likely reads as corn but is small enough to be ambiguous, so slight reservation applied.", "score": 0.85 } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9622222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.30000000000109, "frameIntervalP95Ms": 100.10000000000036, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a9999a07eff2feb8ab64fa727f569552ed6578461b55710bb327526ced440bf7", "gradedAt": "2026-09-30T22:44:44.940932+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6, "toppings": 0.6711111111111111, "bowl": 0.8122222222222222, "broth": 0.6944444444444444, "steam": 0.6744444444444444, "composition": 0.7766666666666666 }, "reward": 0.6985, "runId": "anthropic/sonnet-5.5/xhigh", "rewardStdDev": 0.07686803265626381, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6633333333333333, "bowl": 0.8533333333333333, "broth": 0.7, "steam": 0.6566666666666667, "composition": 0.7633333333333333 }, "reward": 0.6813333333333333, "sampleStdDev": 0.0012583057392117633, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin golden strands have readable overlaps and some curved ends, but the mound is dominated by short angular, rigid-looking segments rather than supple continuous noodles." }, "toppings": { "score": 0.66, "reason": "Scallion rings, scored mushroom caps, browned tofu, corn and veined greens provide recognizable variety. Greens are conspicuously polygonal and upright, and the large nori sheets look overly stiff and smooth." }, "bowl": { "score": 0.85, "reason": "The coherent elliptical rim, orange inner wall, rounded dark exterior and perspective-following gold decoration establish convincing depth. Highlights and a contact shadow support the vessel; the lower body is somewhat indistinct." }, "broth": { "score": 0.7, "reason": "Warm brown liquid has small oil rings, restrained highlights and a clear boundary inside the bowl. Some ingredients appear layered on its surface rather than convincingly immersed, limiting depth." }, "steam": { "score": 0.66, "reason": "Normal frames show changing translucent pale patches above the soup without obscuring the food. They suggest drifting steam, but resemble diffuse cloudy blobs more than connected, tapering wisps." }, "composition": { "score": 0.76, "reason": "A single readable bowl remains the focus against a restrained dark brown background, with complementary green and gold accents and no UI clutter. Heavy all-over grain and a low placement with substantial empty space reduce polish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, scallions, corn and seaweed with noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Individual pale strands and overlaps are readable, but the mound is dominated by short angular, tightly zigzagging segments rather than supple continuous noodles." }, "toppings": { "score": 0.67, "reason": "Scored mushroom caps, browned tofu, scallion rings and corn provide readable variety. Veined greens are overly angular and papery, while the large nori sheets look stiff and uniformly planar." }, "bowl": { "score": 0.86, "reason": "The elliptical rim, orange inner wall, curved dark body and grounded base form coherent depth and perspective. Gold decoration follows the curvature and restrained highlights suggest lacquer; no material construction defect is observed." }, "broth": { "score": 0.7, "reason": "Warm brown liquid has small oil rings, surface highlights and credible containment. Ingredient edges have limited immersion cues, and the surface shading feels somewhat flat." }, "steam": { "score": 0.64, "reason": "Normal frames show changing translucent vapor above the soup without obscuring the food. The vapor is mostly broad blurry patches rather than clearly connected, tapering wisps, and the sampled frames cannot establish continuous flow." }, "composition": { "score": 0.76, "reason": "A single readable dish, warm food colors and a restrained dark setting create a clear focus across desktop and mobile. Heavy image-wide grain reduces clarity, and the low placement leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, scallions, corn and nori with noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Numerous individually outlined strands give readable overlap, but short angular bends and dense maze-like tangles dominate rather than supple continuous noodles." }, "toppings": { "score": 0.66, "reason": "Scored mushroom caps, browned tofu, corn and scallion rings provide recognizable variety. Veined greens are conspicuously angular and paper-like, while nori sheets are overly regular and upright." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim, warm inner wall, curved dark body and integrated base establish convincing depth. The gold border follows the bowl perspective and restrained highlights suggest lacquer; the lower body is somewhat indistinct." }, "broth": { "score": 0.7, "reason": "Warm brown liquid has subtle sheen and scattered oil rings, with ingredients contained inside its surface. Limited translucency and immersion cues make the central food mound feel partly laid on top." }, "steam": { "score": 0.67, "reason": "Normal frames show changing translucent vapor above the soup without obscuring the food. Broad blurred patches suggest heat but lack clearly articulated rising, tapering wisps." }, "composition": { "score": 0.77, "reason": "A single readable dish, restrained warm palette and uncluttered dark setting create a balanced focus across desktop and mobile. Heavy grain and excess upper negative space weaken clarity and polish." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, nori, corn, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5833333333333334, "toppings": 0.5666666666666667, "bowl": 0.75, "broth": 0.6, "steam": 0.6, "composition": 0.7333333333333333 }, "reward": 0.6316666666666667, "sampleStdDev": 0.01376892636821524, "judgments": [ { "noodles": { "score": 0.6, "reason": "Fine, thin noodle strands form a tangled mound with plausible overlap and volume. They read as a fairly uniform texture, and the strands are small and somewhat indistinct." }, "toppings": { "score": 0.6, "reason": "Bok choy leaf, translucent nori, tofu blocks, shiitake caps and green scallion dots are all recognizable. The nori looks like a flat translucent card and the scallions are glossy dots. The tofu is simple, so variety is decent but the execution is stylized and flat." }, "bowl": { "score": 0.75, "reason": "The dark bowl has a coherent rim, a red lacquer interior, a meander pattern band and a foot ring with soft shadow contact. The perspective is consistent. Minor flaws include a slightly odd vertical highlight and a somewhat flat wall." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown with small oil and chili flecks and reads as liquid. It is somewhat murky and flat, and the nori and leaf sit awkwardly on top with only modest immersion cues." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise and shift between frames without hiding the food. They are blurry and somewhat blobby, and a little faint." }, "composition": { "score": 0.75, "reason": "A centered bowl in a dark, moody scene keeps the food as the focal point. The warm palette is cohesive and there is no clutter. The bowl is small in the frame and the grain is heavy." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens, seaweed and noodles are visible, with no animal products. The corn-like yellow bits are plant-based." } }, { "noodles": { "score": 0.6, "reason": "Fine, thin noodle strands form a dense, believable mass with overlaps and variation. They read somewhat as fried or shredded noodles, and the individual strands are too small and uniform to show supple bends." }, "toppings": { "score": 0.55, "reason": "Bok choy, nori, tofu cubes, shiitake and scallion dots are all recognizable. The nori is a translucent flat rectangle, the bok choy is faceted and card-like, and the scallion dots look like glossy beads, so the forms are fairly generic." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with a red-lacquer interior, a Greek-key band, a foot ring, and a soft shadow for grounding. Perspective is consistent and the rim has thickness. Highlights are a little muted." }, "broth": { "score": 0.6, "reason": "Warm brown broth with texture and depth, and the noodles sit in it plausibly. The surface is somewhat murky and flat, and the nori and bok choy look pasted on top with weak immersion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape across frames without hiding the food. They are faint and blobby, and the motion is gentle." }, "composition": { "score": 0.75, "reason": "Centered bowl in a dark, moody scene with a warm palette and vignette. The food is the focus and there is no clutter. The bowl is small in frame and the image is very dark." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, nori, scallions and noodles all look plant-based. No animal products are visible." } }, { "noodles": { "score": 0.55, "reason": "Fine, curly noodle strands form a readable mass with plausible overlap, but they are tiny and scribbly, and the clump looks flat and texture-like with little individual strand definition." }, "toppings": { "score": 0.55, "reason": "Bok choy leaf, nori sheet, tofu cubes, shiitake caps, corn and scallion dots are all recognizable. The nori is a translucent, rigid rectangle, the bok choy is faceted and card-like, and the green dots look like plastic beads." }, "bowl": { "score": 0.75, "reason": "Dark ceramic bowl with a coherent elliptical rim, red-lacquer interior, Greek-key band, foot ring and soft shadow. Perspective is consistent with a subtle highlight. A slightly odd vertical highlight streak is a minor defect." }, "broth": { "score": 0.6, "reason": "Warm brown broth with speckles and oil spots reads as miso-like soup, and the noodles sit in it. The surface is fairly flat and murky, and the nori and leaf look pasted above rather than immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and shift shape between frames, without hiding the food. They are fairly faint and blobby, and their movement reads as gentle but somewhat repetitive." }, "composition": { "score": 0.7, "reason": "Centered bowl with a dark moody background, a warm palette and film grain keeps the food as the focus. The bowl is small in the frame and the scene is quite dark, with lots of empty space." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn and scallions with noodles in a dark broth. There are no visible animal products." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.7833333333333333, "bowl": 0.8333333333333334, "broth": 0.7833333333333333, "steam": 0.7666666666666666, "composition": 0.8333333333333334 }, "reward": 0.7825, "sampleStdDev": 0.047696960070847325, "judgments": [ { "noodles": { "score": 0.75, "reason": "Pale strands form a coherent tangled mound with varied curves, consistent thin thickness, and draping over the mound edges into the broth; slightly uniform squiggle character and flat color keep it below top marks, but no rigid ribbons or maze scribbles." }, "toppings": { "score": 0.8, "reason": "Good variety with readable details: bok choy with veins, torn-edge nori sheet, seared tofu blocks, shiitake slices with gill patterning, corn kernels, and scattered scallion rings; scallions read slightly bead-like and uniformly strewn, and the leaf has a mild artificial glow." }, "bowl": { "score": 0.85, "reason": "Coherent dark lacquer bowl with a crisp elliptical rim, red interior glaze band, gold meander pattern that follows the wall curvature, tapering body to a supported pedestal foot with ground shadow; consistent perspective and subtle highlights, no broken joins." }, "broth": { "score": 0.8, "reason": "Dark miso-like surface with restrained glossy highlights, oil/chili specks, and believable immersion: mushrooms and corn sit at the surface, scallions float, and the noodle mound contacts the liquid; slightly opaque but plausible for the style." }, "steam": { "score": 0.85, "reason": "Across desktop-1 through desktop-3 and pointer frames the wisps change height and shape, rising, spreading and diffusing over time; translucent irregular plumes that never obscure the food, with only mild blobbiness in the denser cloud." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a warm dark atmospheric setting with vignette and grain texture; balanced proportions, coherent lighting, no text, UI or clutter; mobile framing keeps the dish fully readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, scallions, tofu, corn, mushrooms, bok choy and nori; no meat, fish, egg or dairy forms appear." } }, { "noodles": { "score": 0.6, "reason": "Thin pale strands form a coherent mound sitting in the broth with some layering, but they are very uniform in thickness and squiggle pattern, reading somewhat flat and repetitive rather than supple individual strands with volume." }, "toppings": { "score": 0.75, "reason": "Good variety: a bok choy leaf with veins, two dark green nori sheets with irregular torn edges, scallion rings, striped tofu/aburaage slabs, corn kernels and scored mushroom caps; shapes are stylized but readable, though the tofu blocks and scallion dots are somewhat simplified." }, "bowl": { "score": 0.8, "reason": "Coherent black ceramic bowl with consistent rim ellipse, red inner band, tapering walls, footed base, soft ground shadow and highlight streak; the gold meander band is a plausible lacquer decoration, only slightly graphic in execution." }, "broth": { "score": 0.75, "reason": "Dark reddish-brown liquid surface with oil specks and restrained sheen; toppings show contact shadows and sit in the liquid believably, though parts of the surface read slightly murky at the edges." }, "steam": { "score": 0.7, "reason": "A translucent gray plume rises from the bowl and clearly changes shape across desktop frames (tighter in frame 1, more diffuse and spread in later frames), so it is animated, not static; however it is somewhat smoky/blobby rather than crisply wisp-structured." }, "composition": { "score": 0.8, "reason": "Centered focal dish with warm rim-lit dark background, restrained palette and no text or clutter; heavy film grain is a consistent stylistic choice though it slightly mutes fine detail." }, "vegan": { "score": 1.0, "reason": "Visible toppings are bok choy, nori, scallions, tofu/aburaage, corn and mushrooms in plant broth; no egg, meat, fish or dairy is depicted." } }, { "noodles": { "score": 0.8, "reason": "Thin, consistently drawn wavy strands form a coherent tangled mound sitting in the broth, with believable bends and overlap; slight repetitiveness in strand length and a few indistinct clumps keep it from the top tier." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: shiitake caps with cross-hatching, scored fried tofu, corn kernels, scallion rings, sesame, torn-edged nori sheets and a veined bok choy leaf. The leaf looks slightly translucent/glowing and the nori edges are a bit rigid, minor flaws only." }, "bowl": { "score": 0.85, "reason": "Coherent lacquer-style bowl: consistent rim ellipse, red-orange inner glaze band, gold meander frieze wrapping correctly, footed base with contact shadow and a believable highlight streak; it even tilts plausibly in pointer frames. No broken joins observed." }, "broth": { "score": 0.8, "reason": "Rich dark broth with restrained sheen, oil droplets and chili flecks; toppings sit in the liquid with visible occlusion and shading around the noodle mound rather than floating above it. Depth is slightly murky at the edges." }, "steam": { "score": 0.75, "reason": "Normal frames (desktop-1/2/3, pointer frames) show translucent wisps changing shape and position over time, rising and diffusing without obscuring the food. Plumes are somewhat diffuse and blob-like rather than finely wisped, but motion is genuinely supported across frames." }, "composition": { "score": 0.85, "reason": "Balanced centered focal bowl in a warm dark atmospheric setting, coherent lighting and vignette, no text, UI or competing elements; the heavy film-grain texture slightly softens clarity but reads as intentional styling." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, shiitake mushrooms, corn, scallion, bok choy, nori, sesame and noodles in plant broth; nothing resembling meat, fish, egg or dairy is present." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9788888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 216.59999999999854, "frameIntervalP95Ms": 216.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "9321649382f629154a4c0554a435598b3cb9feb24a6cdeb085d8ea54f439c8ed", "gradedAt": "2026-09-30T23:23:50.777828+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4711111111111111, "toppings": 0.4888888888888889, "bowl": 0.6422222222222222, "broth": 0.5055555555555555, "steam": 0.58, "composition": 0.7122222222222222 }, "reward": 0.5536666666666666, "runId": "google/gemini-3.7-flash/low", "rewardStdDev": 0.0710969603975991, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.43333333333333335, "bowl": 0.66, "broth": 0.45, "steam": 0.5066666666666667, "composition": 0.7366666666666667 }, "reward": 0.531, "sampleStdDev": 0.0034999999999999714, "judgments": [ { "noodles": { "score": 0.48, "reason": "Thin continuous cream strands are readable, but mostly form parallel shallow arcs with exposed blunt ends, little tangled volume, and limited immersion." }, "toppings": { "score": 0.43, "reason": "Scallion rings and varied plant-colored toppings are recognizable, but rigid rectangular seaweed, glossy spherical corn, crosshatched disks, and an untextured leaf dominate the depiction." }, "bowl": { "score": 0.66, "reason": "The elliptical rim and curved front wall join coherently with consistent perspective. However, the vessel reads as a shallow platter, with weak cavity depth and poorly resolved base support." }, "broth": { "score": 0.46, "reason": "Warm amber shading suggests soup, but the surface is largely a solid-looking gradient with minimal liquid sheen or contact effects; ingredients appear layered above it rather than immersed." }, "steam": { "score": 0.52, "reason": "The translucent central haze visibly changes shape and intensity across normal frames, but reads as broad glowing fog rather than distinct rising wisps and substantially veils the food on mobile." }, "composition": { "score": 0.74, "reason": "A centered dish, restrained dark background, warm palette, and diagonal chopsticks create a clear focal arrangement without text or UI. Mobile proportions leave considerable empty space and reduce food readability." }, "vegan": { "score": 1, "reason": "Visible noodles, seaweed, green vegetables, corn-like kernels, and grilled plant-like toppings plausibly form vegan ramen; no visibly animal-derived ingredient is evident." } }, { "noodles": { "score": 0.48, "reason": "Thin continuous cream strands are readable, but mostly form parallel shallow arcs with exposed blunt ends and little tangled volume or convincing immersion." }, "toppings": { "score": 0.44, "reason": "Scallion rings, grilled pieces, greens and seaweed provide variety, but rigid rectangular nori, crosshatched disks, spherical corn and a smooth untextured leaf dominate the depiction." }, "bowl": { "score": 0.66, "reason": "The elliptical rim and attached dark front wall have coherent perspective and subtle decoration. The vessel is unusually shallow, with little readable cavity or base and weak surface contact." }, "broth": { "score": 0.43, "reason": "Warm amber shading suggests soup, but the surface looks like a smooth solid plane; ingredients sit above it without convincing liquid contact, menisci or partial submersion." }, "steam": { "score": 0.48, "reason": "Normal frames show changing translucent pale clouds above the soup, but these are broad blurred columns rather than irregular wisps. The dense central glow obscures toppings, especially on mobile." }, "composition": { "score": 0.74, "reason": "A centered single dish, restrained dark palette and diagonal chopsticks create a clear atmospheric presentation without text or UI. Mobile framing gives excessive visual weight to the towering steam." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, seaweed, corn and grilled vegetable or tofu-like pieces plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.48, "reason": "Thin cream strands are continuous and gently curved, but form a sparse stack of nearly parallel arcs with little tangled volume or convincing immersion." }, "toppings": { "score": 0.43, "reason": "Scallion rings, grilled blocks, seaweed and greens provide variety, but rigid rectangular seaweed, crosshatched disks and glossy spherical kernels dominate. The large green leaf lacks veins and reads as a smooth paddle." }, "bowl": { "score": 0.66, "reason": "The elliptical rim and connected dark front wall have coherent perspective and decorative shading. The vessel appears unusually shallow, with little readable lower body or base and weak contact definition." }, "broth": { "score": 0.46, "reason": "Warm amber gradients suggest soup, but the surface is largely matte and solid-looking. Ingredients appear arranged above it with few convincing submerged edges or liquid contact cues." }, "steam": { "score": 0.52, "reason": "Broad translucent plumes change position and density across normal frames, supporting ambient movement. They resemble diffuse glowing fog rather than irregular rising wisps and substantially veil the food, especially on mobile." }, "composition": { "score": 0.73, "reason": "A centered dish, restrained dark background and diagonal chopsticks create a clear focal arrangement without text or UI. Warm colors are cohesive, though the mobile presentation gives excessive space and prominence to the steam." }, "vegan": { "score": 0.94, "reason": "Visible toppings plausibly depict grilled tofu, mushrooms, seaweed, greens, scallions and corn. No clearly animal-derived ingredient is observed, though some simplified topping forms remain ambiguous." } } ], "diagnostics": { "vegan": 0.98 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.5, "bowl": 0.5166666666666667, "broth": 0.4666666666666667, "steam": 0.5333333333333333, "composition": 0.6166666666666667 }, "reward": 0.4966666666666667, "sampleStdDev": 0.01421560175769333, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are a few thin flat cream lines lying on the surface. They read as noodles but are sparse and wire-like, with little volume or overlap. They do not look immersed in the broth." }, "toppings": { "score": 0.5, "reason": "Toppings are recognizable, with nori sheets, grilled tofu slabs, bok choy, corn, scallions, and bamboo shoots. Several read as stiff cards or glossy blobs. The bok choy head looks like a plastic green sphere, and the lattice discs are odd. There is decent variety but plasticky rendering." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent elliptical rim and a dark lacquer-like wall with a foot ring. The lower body dissolves into darkness and the base and support are unclear. The perspective is mostly consistent, though the rim is very wide and the body looks shallow." }, "broth": { "score": 0.5, "reason": "The broth is a warm orange gradient that looks appetizing. It is fairly flat and opaque, with scattered chili dots. Toppings sit on the surface with weak immersion and few ripples or contact cues." }, "steam": { "score": 0.55, "reason": "Soft translucent steam rises above the bowl and shifts between frames. It is hazy and blobby, and it partially veils the nori. Its motion looks plausible but it is not very wispy." }, "composition": { "score": 0.65, "reason": "The dark atmospheric scene has a centered, readable focal bowl with warm lighting and restrained colors. The chopsticks are cut off at the edge and the lower bowl merges into the darkness. There is no clutter or text." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are tofu, nori, bok choy, corn, scallions, bamboo shoots, and noodles. No meat, egg, or fish is visible. The lattice discs are ambiguous but most likely plant-based, such as mushrooms." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, pale, flat curved lines lying on the surface. They are sparse, read as rigid ribbons with little volume, and are not clearly immersed in the broth." }, "toppings": { "score": 0.5, "reason": "Tofu with grill marks, nori sheets, a bok choy, corn, scallions, menma and a lattice-patterned item are all recognizable and varied. The bok choy head is a glossy smooth blob, the nori are stiff flat cards, and the lattice disks look generic, so the set is artificial." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim, cavity and dark body with some ridges, but the lower body merges into a dark mass and the base looks odd. The shape is wide and flattened with weak contact and support." }, "broth": { "score": 0.45, "reason": "The broth is a flat orange disc with a smooth gradient and scattered chili beads. It has little depth or sheen, and the toppings look placed on top rather than immersed." }, "steam": { "score": 0.55, "reason": "Soft translucent haze rises above the bowl and shifts between frames. It is fairly diffuse and blobby, but it does not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is a centered focal point in a dark atmospheric scene with warm lighting. The toppings are cluttered and the lower half of the frame is a large dark empty area." }, "vegan": { "score": 0.85, "reason": "Tofu, nori, bok choy, corn, scallions and menma all look plant-based. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin pale flat curved lines lying on the surface. They read as a few strands, lack volume and overlap, and look like rigid ribbons rather than a supple noodle mass." }, "toppings": { "score": 0.5, "reason": "Nori sheets, grilled tofu slabs, a bok choy, corn kernels, scallions and menma are all recognizable. However, the bok choy head is a glossy green blob, the tofu is a rigid block, and the lattice discs are odd. The variety is good but the depiction is artificial." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim, a dark body and ridges at the base, but the body merges into a dark shadow and the base is unclear. The rim perspective is slightly inconsistent with the lower body, and the chopsticks look detached." }, "broth": { "score": 0.45, "reason": "The broth is a flat orange disc with a smooth gradient and few ripples. Ingredients sit on top more than in it, and the surface has little depth." }, "steam": { "score": 0.5, "reason": "Soft translucent haze rises and shifts between frames, so it reads as plausible steam. It is fairly diffuse and blobby, and partly obscures the nori." }, "composition": { "score": 0.6, "reason": "The bowl is centered and dominant in a dark, moody scene with warm lighting and no UI clutter. The bowl's lower body is lost in shadow, and the toppings are crowded." }, "vegan": { "score": 0.85, "reason": "Tofu, nori, bok choy, corn, scallions and menma all look plant-based. No meat or egg is visible." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5333333333333333, "toppings": 0.5333333333333333, "bowl": 0.75, "broth": 0.6, "steam": 0.7, "composition": 0.7833333333333333 }, "reward": 0.6333333333333333, "sampleStdDev": 0.02602082499332669, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are readable pale strands fanning across the front of the bowl, but they are uniformly spaced, flat, parallel repeated curves with identical sweep \u2014 a stylized fan rather than supple, individually bent strands. Continuous and coherent, but rigid and repetitive." }, "toppings": { "score": 0.5, "reason": "Good variety: striped tofu/aburaage slabs with grill marks, nori sheets, scallion-ring cluster, speckled menma cylinders, red chili threads. However, glossy plastic-like forms dominate much of the arrangement: a lollipop-like green sphere on a pale stem, beaded glossy corn spheres, a crosshatched ball that reads more like a woven ball than a mushroom, and stiff rectangular nori cards." }, "bowl": { "score": 0.8, "reason": "Coherent shallow donburi-style bowl: elliptical rim, visible wall thickness, tapered dark exterior, patterned foot ring, and grounded contact shadow. Perspective between rim, cavity and base is consistent; chopsticks rest believably on the rim. Minor shortcomings in wall depth and rim material richness." }, "broth": { "score": 0.55, "reason": "Broth is a smooth orange-tan surface with a soft radial glow and scattered oil/seasoning specks, but it reads fairly flat and solid with little texture or sheen. Toppings sit mostly on top with minimal occlusion or immersion shading at their bases." }, "steam": { "score": 0.7, "reason": "Steam is clearly animated: the translucent plume changes shape across the normal frames (broad plume shifting left/right and spreading between desktop-1/2/3 and pointer frames) and is fully absent in the reduced-motion frames. It is translucent and irregular, though somewhat foggy and milky over the toppings rather than distinct wisps." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric presentation: centered focal bowl, warm appetizing palette, subtle wood-texture background, floating dust particles and vignette, no text or UI clutter. Chopsticks add diagonal interest. Steam partially veils the food and the mobile crop leaves the bowl small with much empty space." }, "vegan": { "score": 1.0, "reason": "Visible toppings \u2014 nori, tofu/tofu-pouch slabs, green vegetable, scallions, corn-like spheres, menma, chili threads and plant broth \u2014 are all plausibly plant-derived; no meat, fish, egg or dairy imagery is present." } }, { "noodles": { "score": 0.55, "reason": "Cream-colored strands curve continuously through the lower broth with gentle bends, but they are sparse flat lines with little volume or bundle cohesion, reading more as decorative curves than a substantial noodle nest." }, "toppings": { "score": 0.55, "reason": "Good variety: readable nori sheets, striped tofu slabs, ridged menma cylinders, sprout cluster and chili threads. However glossy plastic-looking spheres dominate parts of it \u2014 the large shiny green ball, beaded corn kernels and scattered red droplets \u2014 and the crosshatched brown balls are ambiguous." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, dark ceramic body with a subtle wave-pattern base and believable shadow/contact, with chopsticks resting on the rim. Defects: the bowl is unusually shallow and flat for ramen and the broth fills nearly to the rim with an abrupt wall transition." }, "broth": { "score": 0.65, "reason": "Smooth orange-brown liquid surface with a soft radial gradient giving restrained depth; ingredients generally sit in contact with it. Weaknesses: the surface lacks oil/texture detail and the scattered glossy droplets read as floating beads rather than broth sheen." }, "steam": { "score": 0.6, "reason": "Steam clearly changes shape and intensity across the time-ordered frames (dim in desktop-1/3, brighter and wider in desktop-2/pointer frames), so it is animated. But it reads as a soft fog-like cloud rather than defined rising wisps, and on mobile it hazes over much of the dish." }, "composition": { "score": 0.7, "reason": "Dark atmospheric scene with the dish as clear focal point, warm palette, subtle background texture and no distracting UI or clutter. The desktop framing is balanced; the mobile view leaves the dish small with large empty dark areas." }, "vegan": { "score": 0.6, "reason": "Visible toppings (nori, tofu-like slabs, sprouts, corn, menma, chili) plausibly form a vegan bowl, but the crosshatched brown spheres could be read as meatballs, so appearance is ambiguous rather than clearly plant-based." } }, { "noodles": { "score": 0.55, "reason": "Pale cream strands fan out beneath the toppings with gentle bends and readable continuity, but they read as flat, thin parallel lines in a repetitive radial fan rather than supple strands with volume or believable immersion in broth." }, "toppings": { "score": 0.55, "reason": "Good variety and some convincing pieces: speckled nori sheets, grill-marked tofu blocks with gloss highlights, scallion-ring cluster and chili threads. However, the large focal green ball on a stalk is a glossy plastic-sphere form, the two crosshatched brown spheres read as generic textured balls rather than clear mushrooms, and the corn kernels are shiny beads, which weakens food believability." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, dark lacquer-like body with a wood-grain base pattern, consistent interior/exterior perspective, contact shadow beneath, and chopsticks resting plausibly on the right rim. The bowl is stylistically shallow, but geometry is intact with no distortion or detached parts." }, "broth": { "score": 0.6, "reason": "Warm amber surface with restrained sheen, scattered chili-oil droplets and sesame specks that read as liquid texture, and a soft central glow. However the surface is essentially a flat gradient with little depth, and toppings sit on top with weak occlusion/immersion shadows." }, "steam": { "score": 0.8, "reason": "Across the desktop frames the translucent wisp plume changes shape, position and spread continuously, rising from the soup and fading without becoming opaque fog or hiding the dish; reduced-motion frames show it correctly removed. Slightly blobby/cloud-like in places." }, "composition": { "score": 0.85, "reason": "Well-centered glowing focal dish against a dark atmospheric background with subtle particles, balanced chopsticks on the right, coherent warm lighting, no text or clutter, and the mobile viewport crops gracefully. Background glow does not compensate for topping weaknesses but the presentation itself is attractive." }, "vegan": { "score": 0.9, "reason": "Visible ingredients plausibly form vegan ramen: nori, grilled tofu, greens, corn, scallion rings, chili threads, sesame and noodles. The brown crosshatched spheres and rolled cylinders are slightly ambiguous in identity, but nothing is clearly animal-derived." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8877777777777778, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100.0, "frameIntervalP95Ms": 116.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "035ba53b9d39f1671e92211a2af533f878a1c85f634f0f25a06176e6167f8d64", "gradedAt": "2026-09-30T23:22:21.954003+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.38666666666666666, "toppings": 0.48444444444444446, "bowl": 0.5711111111111111, "broth": 0.5488888888888889, "steam": 0.5222222222222223, "composition": 0.5811111111111111 }, "reward": 0.0, "runId": "google/gemini-3.7-flash/medium", "rewardStdDev": 0.08724693563545817, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.42666666666666664, "toppings": 0.45333333333333337, "bowl": 0.63, "broth": 0.48, "steam": 0.4166666666666667, "composition": 0.6266666666666667 }, "reward": 0.49783333333333335, "sampleStdDev": 0.004804511768466529, "judgments": [ { "noodles": { "score": 0.42, "reason": "Thin golden strands have readable curved bends, but appear as sparse isolated squiggles rather than a coherent overlapping noodle mass with volume." }, "toppings": { "score": 0.44, "reason": "Textured nori, grill markings and small garnish details add variety, but the rigid rectangular seaweed, smooth leaf, thick disks and glossy egg-like form remain strongly artificial." }, "bowl": { "score": 0.63, "reason": "A dark rim, curved highlight and cast shadow establish a supported vessel. The angular upper contour and weakly articulated outer walls limit convincing ceramic depth." }, "broth": { "score": 0.48, "reason": "Warm amber shading and scattered surface flecks suggest soup, but the surface lacks liquid sheen and convincing immersion; large toppings look layered onto it." }, "steam": { "score": 0.43, "reason": "Normal frames show changing translucent haze around the soup, supporting motion. Broad blurred patches obscure ingredients and lack distinct rising, spreading wisps." }, "composition": { "score": 0.61, "reason": "Warm food colors contrast clearly with the restrained dark tabletop, without text or UI clutter. The mobile dish is substantially clipped on the left, and dense topping overlap weakens readability." }, "vegan": { "score": 0, "reason": "A prominent egg half with a pale white surrounding an orange yolk is visibly animal-derived." } }, { "noodles": { "score": 0.43, "reason": "Thin golden strands have curved bends and highlights, but appear as sparse short squiggles rather than a coherent overlapping noodle portion with volume." }, "toppings": { "score": 0.46, "reason": "Textured nori, a veined leaf and small garnish rings provide variety. However, the large grilled block, crosshatched disks and glossy oval forms look rigid and diagrammatic rather than organically textured food." }, "bowl": { "score": 0.62, "reason": "A dark enclosing rim, curved highlight and cast shadow establish a supported vessel. The angular upper contour and minimally articulated walls make the cavity and ceramic thickness less convincing." }, "broth": { "score": 0.48, "reason": "Warm amber shading and scattered surface flecks suggest soup, but the surface is largely a smooth gradient and major toppings look layered above it without convincing immersion or liquid contact." }, "steam": { "score": 0.4, "reason": "Normal desktop frames show translucent haze changing position and intensity. It reads mostly as broad blurred patches drifting over the food, not irregular rising wisps with clearly articulated spreading and fading." }, "composition": { "score": 0.64, "reason": "The warm dish contrasts attractively with the restrained dark tabletop, without text or UI clutter. Desktop framing is readable, but the mobile bowl is substantially cropped on the left and steam washes out its focal toppings." }, "vegan": { "score": 0, "reason": "The prominent cream-colored oval with an orange yolk unmistakably depicts an egg half, despite the surrounding plant-like toppings." } }, { "noodles": { "score": 0.43, "reason": "Thin yellow strands have curved bends and highlights, but appear as sparse isolated squiggles rather than a coherent overlapping noodle volume." }, "toppings": { "score": 0.46, "reason": "Textured nori, grill markings, leaf veins and small garnish details provide variety. However, the rigid rectangular seaweed, smooth leaf, crosshatched disks and glossy egg-like oval remain predominantly geometric food symbols." }, "bowl": { "score": 0.64, "reason": "The dark rim encloses a readable cavity with restrained highlights and a supporting shadow. Its angular upper contour and minimally described exterior make the vessel feel somewhat flat and faceted." }, "broth": { "score": 0.48, "reason": "Warm amber shading and scattered flecks suggest broth, but the surface lacks convincing liquid sheen and immersion; large toppings appear layered over a smooth gradient." }, "steam": { "score": 0.42, "reason": "Normal desktop frames show changing translucent haze over the soup, supporting motion. It appears as broad blurred patches rather than rising irregular wisps, with little readable spreading or fading structure." }, "composition": { "score": 0.63, "reason": "Warm food colors stand out against a restrained dark tabletop without text or UI clutter. The desktop dish is readable, but the mobile framing cuts off much of its left side and steam partially washes out the food." }, "vegan": { "score": 0, "reason": "A prominent halved egg with a pale white and orange yolk is visibly depicted." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.4, "bowl": 0.4166666666666667, "broth": 0.5166666666666667, "steam": 0.5, "composition": 0.5 }, "reward": 0.42000000000000004, "sampleStdDev": 0.008660254037844393, "judgments": [ { "noodles": { "score": 0.3, "reason": "Only a few short, squiggly cream strands are visible on the broth; there is no real noodle mass or volume, and the strands read as flat scribbles." }, "toppings": { "score": 0.4, "reason": "The toppings are varied (nori, lotus root, a tofu-like block, a leaf, scallions, corn-like beads), but many are generic or plastic-looking. The crosshatch mushrooms and the rectangular card are crude. The egg half is an obvious depiction of an animal product." }, "bowl": { "score": 0.4, "reason": "The bowl is a dark circle with an odd rim arc and a large offset shadow. The rim and body are inconsistent, the silhouette is lumpy, and the wall thickness and base are unclear. The bowl also shifts position between frames." }, "broth": { "score": 0.5, "reason": "The warm brown gradient looks appetizing and has some depth. It is fairly flat, though, and the toppings look pasted on top rather than sitting in the liquid." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps drift up and to the right and change between frames. They are blobby and mostly off to one side, but they do not hide the food." }, "composition": { "score": 0.5, "reason": "The dark moody background keeps the bowl as the focal point. The bowl is off-center, the large shadow is distracting, and the layout shifts between frames. The mobile view crops the bowl." }, "vegan": { "score": 0.2, "reason": "A marinated egg half with a bright orange yolk is clearly visible, which is an animal product. The remaining ingredients are plant-based." } }, { "noodles": { "score": 0.3, "reason": "Only a few short, scattered squiggles of noodle are visible at the broth edges. There is no noodle mass or volume, and the strands are thin and flat, so they read as sparse scribbles." }, "toppings": { "score": 0.4, "reason": "The toppings are varied: nori, a green leaf, lotus root, a fried tofu-like block, mushroom-like cross-hatched disks, scallion rings and corn. Several read as generic flat shapes or cards, and the red starburst scallion is odd. The egg half is prominent and clearly not vegan." }, "bowl": { "score": 0.4, "reason": "The dark bowl has a thin rim highlight, but the rim shape is distorted and lopsided. The rim highlight arc is offset, and the bowl wall and broth circle don't align in perspective. The shadow is heavy, and the bowl shifts position between frames." }, "broth": { "score": 0.5, "reason": "The warm brown gradient looks appetizing and has some depth. The surface is fairly flat, and the toppings sit on it rather than in it, with little immersion or contact." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps drift off the bowl and change between frames. They are blobby and mostly appear off to the side, but they do not hide the food." }, "composition": { "score": 0.5, "reason": "The bowl is a clear focal point against a dark, moody background. It is off-center and the frame shifts, and the nori sticks out over the rim awkwardly. The mobile view crops the bowl." }, "vegan": { "score": 0.15, "reason": "A marinated egg half with an orange yolk is clearly visible, which is an animal-derived ingredient. The other toppings are plant-based." } }, { "noodles": { "score": 0.3, "reason": "Only a few short, squiggly yellow strands are visible at the edges of the broth. There is no real noodle mass, and the strands read as flat scribbles with no volume or overlap." }, "toppings": { "score": 0.4, "reason": "Nori, lotus root, a leaf, a mushroom-like cross-hatched disc and a menma-like plank are all recognizable. Several read as generic flat shapes: the nori is a rigid rectangle, the hatched discs look like grills, and the leaf is a plain vector. The egg half is also present." }, "bowl": { "score": 0.45, "reason": "The bowl reads as a dark round vessel with a light rim arc and a cast shadow. The rim and body geometry are inconsistent and look distorted, and the dark body blends into the background. The bowl also shifts position and size between frames." }, "broth": { "score": 0.55, "reason": "The warm brown gradient broth looks appetizing and has some depth, with small oil droplets on the surface. It is fairly flat, and the toppings sit pasted on top rather than immersed." }, "steam": { "score": 0.5, "reason": "Soft translucent wisps drift up and to the right and change between frames. They are blobby, and they stay mostly at the edges rather than rising from the soup." }, "composition": { "score": 0.5, "reason": "The dark atmosphere is moody and the bowl is the clear focal point. The layout is cluttered and the toppings overlap awkwardly. The bowl is off-center and in the mobile view it is cropped." }, "vegan": { "score": 0.2, "reason": "A marinated egg half with a orange yolk is clearly visible, which is an animal-derived ingredient. The rest of the bowl is plant-based." } } ], "diagnostics": { "vegan": 0.18333333333333335 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.43333333333333335, "toppings": 0.6, "bowl": 0.6666666666666666, "broth": 0.65, "steam": 0.65, "composition": 0.6166666666666667 }, "reward": 0.5941666666666667, "sampleStdDev": 0.03165569985537094, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles appear only as sparse, flat yellow squiggle strokes and a small bundle of parallel lines peeking over the tofu; no continuous supple strands with volume or coherent noodle mass in the broth." }, "toppings": { "score": 0.55, "reason": "Recognizable nori sheet, egg half, lotus-root slice, leaf and a grill-lined tofu slab, but several elements are abstract or plastic-like: a red/green starburst that reads as nothing edible, glossy orange bead clusters, and a plain flat green blob leaf." }, "bowl": { "score": 0.65, "reason": "Coherent top-down dark vessel with a wide rim, thin left highlight arc, interior depth and a soft drop shadow; the outer silhouette is slightly lumpy/notched in places (top-right bulge), a minor flaw." }, "broth": { "score": 0.65, "reason": "Warm brown liquid with radial sheen, scattered oil droplets and chili flecks; toppings mostly sit on the surface with soft shading, though contact/occlusion is shallow and the center reads slightly like a flat gradient." }, "steam": { "score": 0.65, "reason": "Frames show moving translucent whitish wisps: faint haze at the right in desktop-1 grows and shifts by desktop-2/3, and pointer-left shows plumes rising above the rim; some haze is blobby rather than wispy, but motion is clearly supported." }, "composition": { "score": 0.6, "reason": "Desktop presents a clean, restrained dark scene with the bowl as clear focal point, but the mobile frame crops the bowl heavily off the left edge, and the dish sits left-of-center with large empty background." }, "vegan": { "score": 0, "reason": "An unmistakable soft-boiled egg half with a large orange yolk is prominently placed in the bowl, a clearly animal-derived topping." } }, { "noodles": { "score": 0.45, "reason": "Noodles appear only as sparse, thin cream squiggle strands scattered near the broth rim; no coherent noodle nest, volume, or believable mass under the toppings \u2014 reads as decorative flat squiggles." }, "toppings": { "score": 0.7, "reason": "Good variety with a convincing soft-boiled egg (glossy yolk, specked white), textured nori sheet, ridged menma block, lotus-root slice and dark wood-ear; the large green leaf and crosshatched brown discs are simplistic and the pink/green starburst garnish is ambiguous." }, "bowl": { "score": 0.65, "reason": "Dark lacquer-style bowl with a cream rim highlight, inner rim and supporting shadow ellipse; form is coherent but low-contrast against the dark background, the silhouette is slightly irregular and subtly wobbles between frames." }, "broth": { "score": 0.7, "reason": "Warm brown radial-gradient surface with chili-oil droplets and darker edge; toppings sit on it with soft shadows and immersion reads believably, though the surface is somewhat flat with little texture or sheen variation." }, "steam": { "score": 0.7, "reason": "Translucent wisps clearly change position and shape across the timed frames (right-side plumes in frames 2\u20133, bright puff near the egg in pointer-right, rising plumes above the bowl in pointer-left), and are suppressed in reduced-motion frames; wisps are slightly blob-like but do not hide the food." }, "composition": { "score": 0.6, "reason": "Attractive dark atmospheric setting with dust specks and a focused dish, but the bowl sits left of center leaving empty space on the right, and the mobile view badly crops the bowl off-screen at the left edge." }, "vegan": { "score": 0, "reason": "An unmistakable soft-boiled egg half with a bright orange yolk sits in the bowl; the small orange bead cluster also resembles roe. Clearly animal-derived toppings earn 0." } }, { "noodles": { "score": 0.45, "reason": "Noodles appear only as sparse, thin yellow squiggles scattered on the surface rather than a coherent bed or volume under the toppings; the short disconnected flat strands read more like decorative squiggles than a believable noodle mass." }, "toppings": { "score": 0.55, "reason": "Some toppings read well in the flat-vector style: a convincing soft egg with glossy yolk, a lotus-root slice with holes, a textured nori sheet, and a leaf. Others are ambiguous or weak: two flat crosshatched brown disks, a striped tan block, an abstract green/red starburst, and a plain green ellipse. The nori is a rigid rectangle that overhangs and appears to float past the rim." }, "bowl": { "score": 0.7, "reason": "Coherent top-down bowl: dark circular rim with a thin ceramic highlight arc, recessed inner cavity, consistent perspective, and an elliptical contact shadow beneath. Simple and somewhat featureless, but no broken geometry or detached parts (the overhang issue belongs to the nori topping)." }, "broth": { "score": 0.6, "reason": "Warm brown radial-gradient surface with a restrained sheen, oil/condiment droplets, and soft shadows that seat the toppings. However the liquid is a smooth flat gradient with no visible texture, ripples, or depth, looking somewhat solid." }, "steam": { "score": 0.6, "reason": "Steam is animated: faint haze over the nori in early frames grows, wisps shift, and by the pointer-left frame distinct translucent plumes rise above the upper rim. It never hides the food, though in desktop-2/3 the right-side wisps break into a few discrete floating blobs rather than continuous plumes." }, "composition": { "score": 0.65, "reason": "Desktop framing is attractive: dark atmospheric background with subtle line texture and dust motes, warm glowing dish as a clear focal point, balanced placement, no text or clutter. The mobile viewport is a real defect, with the bowl clipped off the left edge and roughly half out of frame." }, "vegan": { "score": 0.0, "reason": "A clearly recognizable soft-boiled egg half with a distinct orange yolk sits prominently in the bowl, an unmistakable animal-derived topping; vegan compliance fails regardless of the plant-based vegetables, nori, and noodles present." } } ], "diagnostics": { "vegan": 0.0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.061111111111111116, "veganFailed": true }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 66.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "6825e394f9934628fa420452049300b2e7971f7d18f887517991ca4085691f28", "gradedAt": "2026-09-30T23:22:09.947904+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5566666666666666, "toppings": 0.6055555555555556, "bowl": 0.7544444444444445, "broth": 0.4611111111111111, "steam": 0.4544444444444445, "composition": 0.7611111111111112 }, "reward": 0.6040000000000001, "runId": "google/gemini-3.7-flash/high", "rewardStdDev": 0.09403914669493293, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.42, "toppings": 0.5, "bowl": 0.75, "broth": 0.35, "steam": 0.38, "composition": 0.77 }, "reward": 0.528, "sampleStdDev": 0.0020000000000000018, "judgments": [ { "noodles": { "score": 0.42, "reason": "Thin cream strands are readable, but dense parallel waves form a flat, comb-like mat with many abrupt ends and little convincing overlap or volume." }, "toppings": { "score": 0.5, "reason": "Mushroom scoring, leaf veins, lotus-root holes and garnish provide variety and recognizable details. However, perfectly round mushroom caps, glossy spherical corn and heavily crosshatched oval slices look diagrammatic rather than organic." }, "bowl": { "score": 0.73, "reason": "Nested elliptical rims are coherent, with readable thickness, restrained highlights and a grounded silhouette. The nearly black exterior obscures the wall curvature and base, weakening the sense of ceramic depth." }, "broth": { "score": 0.35, "reason": "An amber surface surrounds the noodles, but it reads as a flat solid fill with little liquid sheen or depth. Ingredients mostly appear layered above it rather than immersed." }, "steam": { "score": 0.4, "reason": "Normal frames show changing translucent haze over the soup and drifting patches above it. The vapor is mostly diffuse disconnected blobs, without clearly readable wisps rising and spreading from the liquid." }, "composition": { "score": 0.76, "reason": "The centered dish, warm wood, dark vignette and restrained chopstick placement create a clear, uncluttered focal scene. Mobile retains the full arrangement but leaves the dish relatively small amid substantial empty space." }, "vegan": { "score": 0.72, "reason": "Mushrooms, greens, seaweed, corn and lotus root suggest plant-based ramen. The large grilled brown oval slices are visually ambiguous and could resemble meat or a plant-based substitute; no unmistakable animal ingredient is visible." } }, { "noodles": { "score": 0.42, "reason": "Fine strands are clearly visible, but densely repeated parallel waves form a flat, fabric-like mat with abrupt ends and little convincing overlap or volume." }, "toppings": { "score": 0.5, "reason": "Mushroom scoring, leaf veins, lotus-root holes and scallion rings provide readable variety. However, perfectly circular mushrooms, glossy spherical corn and rigid gridded slices make much of the food look diagrammatic." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, nested cavity and consistent highlights establish a coherent dark bowl. Its exterior is largely lost in a broad black silhouette, limiting wall curvature, material detail and contact readability." }, "broth": { "score": 0.35, "reason": "An amber surface is visible around the noodles, but it appears like a flat solid fill. Ingredients sit above it without persuasive immersion, liquid reflections or contact effects." }, "steam": { "score": 0.36, "reason": "Translucent haze changes position and intensity across the normal frames, leaving most food readable. It resembles detached blurred patches rather than irregular wisps visibly rising and spreading from the soup." }, "composition": { "score": 0.78, "reason": "The centered dish, restrained chopsticks and warm wooden background create a clear, uncluttered focal arrangement. Heavy black shadowing and the relatively small mobile dish weaken the overall balance." }, "vegan": { "score": 0.72, "reason": "Mushrooms, greens, seaweed, corn and lotus root plausibly support vegan ramen. The large crosshatched brown slices could be grilled tofu or meat-like slices; their appearance is ambiguous." } }, { "noodles": { "score": 0.42, "reason": "Thin cream strands are readable, but dense parallel waves form a flat, repetitive mat with many abrupt ends rather than supple overlapping noodle volume." }, "toppings": { "score": 0.5, "reason": "Mushroom scoring, leaf veins, lotus-root holes and garnish provide variety. However, perfectly round mushroom disks, bead-like corn, rigid seaweed cards and oversized crosshatched slices look artificial." }, "bowl": { "score": 0.76, "reason": "Nested elliptical rims, restrained highlights and a dark lower wall establish a coherent supported vessel. The nearly black body lacks material detail and clearly resolved depth." }, "broth": { "score": 0.35, "reason": "An amber surface is visible around the noodles, but it reads as a flat solid fill with little sheen or depth; ingredients appear layered above it rather than immersed." }, "steam": { "score": 0.38, "reason": "Translucent haze changes position and intensity across normal frames, supporting motion. It appears mainly as diffuse patches over the food and detached background blobs, not clearly rising, spreading wisps." }, "composition": { "score": 0.77, "reason": "The centered dish, warm wood, dark vignette and restrained chopstick placement create a readable atmospheric scene without UI clutter. Mobile framing leaves substantial empty space and makes the dish relatively small." }, "vegan": { "score": 0.65, "reason": "Mushrooms, greens, seaweed, corn and lotus root look plant-based. The large grilled brown slices are ambiguous between tofu and meat-like cutlets, so full vegan confidence is not supported visually." } } ], "diagnostics": { "vegan": 0.6966666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.5833333333333334, "bowl": 0.6633333333333333, "broth": 0.43333333333333335, "steam": 0.43333333333333335, "composition": 0.68 }, "reward": 0.5748333333333333, "sampleStdDev": 0.010866615541802004, "judgments": [ { "noodles": { "score": 0.6, "reason": "Dense wavy cream strands fill the bowl with a believable volume and subtle overlap. They read as repeated, fairly uniform wavy lines with a jagged cutoff edge near the rim, and they look flat." }, "toppings": { "score": 0.6, "reason": "The shiitake caps with star cuts, lotus root, bok choy leaf, nori and scallion rings are all recognizable. The corn kernels look like glossy plastic beads, the red shreds are scribbly, and the scored slices resemble grilled meat or tofu. Overall they feel pasted on." }, "bowl": { "score": 0.65, "reason": "The bowl has a coherent elliptical rim, a dark lacquer wall with an orange inner line, and a big cast shadow. The body is not very visible, the shadow is an oversized black ellipse, and the contact with the table looks vague." }, "broth": { "score": 0.4, "reason": "Little broth is visible, only an orange-brown band at the edge under the noodles. It has no sheen or depth, and the toppings sit on top of the noodle mat instead of in liquid." }, "steam": { "score": 0.5, "reason": "Soft translucent puffs drift and change position between frames. They appear away from the bowl, off to the left, and look like blurry blobs disconnected from the soup surface." }, "composition": { "score": 0.68, "reason": "A centered bowl on a dark wood table with chopsticks and a rest gives a readable focal dish and a warm palette. The huge black shadow, the visible vertical band in the background and the weakly connected steam detract." }, "vegan": { "score": 0.6, "reason": "Mushrooms, lotus root, greens, corn, nori and noodles are vegan. The grill-scored slices look like meat or fish, so the appearance is ambiguous." } }, { "noodles": { "score": 0.6, "reason": "Dense wavy strands with a convincing cream color and a coherent mass, but they read as repetitive thin wavy lines with a jagged cut-off edge, and little depth or overlap." }, "toppings": { "score": 0.6, "reason": "Shiitake with star cuts, bok choy, lotus root, corn, nori and scallion rings are all recognizable. The corn kernels look like glossy plastic beads, the red chili threads are scribbly, and the scored slab looks like grilled meat." }, "bowl": { "score": 0.68, "reason": "Coherent elliptical rim with a lacquer highlight, an inner wall and a copper ring, plus a dark shadow base. The perspective is consistent, but the base reads as a flat black blob and the wall thickness is ambiguous." }, "broth": { "score": 0.45, "reason": "Only a thin orange-brown ring of broth is visible around the noodles. It has little sheen or depth, and the noodles fill the surface like a mat instead of sitting in liquid." }, "steam": { "score": 0.4, "reason": "Soft blurred wisps drift and change between frames, but they appear offset from the bowl, in the upper left, the right and the bottom left. They read as blobs disconnected from the soup, and the pointer frame shows odd placement." }, "composition": { "score": 0.68, "reason": "The bowl is a clear focal point in a dark, warm wood scene, with chopsticks and a rest balancing the layout. The shadow is heavy and there is a visible vertical seam in the background." }, "vegan": { "score": 0.6, "reason": "The mushrooms, bok choy, lotus root, corn and nori are vegan. The scored slab resembles grilled meat or fish, which makes the appearance ambiguous." } }, { "noodles": { "score": 0.6, "reason": "Dense wavy cream strands with readable overlap and volume fill the bowl. They look like repeated flat wavy lines with a jagged cutoff at the edges, so they read as a pattern more than supple noodles." }, "toppings": { "score": 0.55, "reason": "Shiitake with star scoring, bok choy, lotus root, corn kernels and nori are all recognizable. The large crosshatched slabs look like grilled meat or tofu, and the corn beads and red balls look glossy and plastic. The red threads and green rings are messy." }, "bowl": { "score": 0.66, "reason": "The elliptical rim is coherent, with a copper ring and dark wall, and it sits on a shadow. The shadow is a heavy black blob, and the wall and cavity lack depth." }, "broth": { "score": 0.45, "reason": "Little broth is visible, since noodles cover most of it. Only a thin orange-brown ring at the edge shows. It reads as flat and has little sheen or immersion." }, "steam": { "score": 0.4, "reason": "Soft blurry puffs drift and change position between frames. They are offset from the bowl, drift to odd places like the left edge, and read as blobs rather than wisps rising from the soup." }, "composition": { "score": 0.68, "reason": "The dark wood background has a warm vignette, the bowl is centered as the focal point, and the chopsticks and rest add context. The huge black shadow and the empty space around the bowl weaken it." }, "vegan": { "score": 0.5, "reason": "The mushrooms, vegetables, corn and nori are vegan. The crosshatched slabs with sesame look like grilled meat or fish, which makes the appearance ambiguous." } } ], "diagnostics": { "vegan": 0.5666666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7333333333333334, "bowl": 0.85, "broth": 0.6, "steam": 0.55, "composition": 0.8333333333333334 }, "reward": 0.7091666666666667, "sampleStdDev": 0.006291528696058942, "judgments": [ { "noodles": { "score": 0.65, "reason": "Cream-colored wavy strands blanket the bowl continuously with some shaded volume, but the pattern is highly uniform and repetitive, reading more like a combed-wave texture than individual supple noodles with varied bends and crossings." }, "toppings": { "score": 0.8, "reason": "Rich variety with readable details: shiitake caps with gills and star cuts, bok choy leaf, lotus-root discs with holes, bamboo shoots, scallion rings, chili threads, tomatoes, nori, and grill-marked sesame-dusted slabs. Corn cluster of uniform glossy yellow spheres is the main plastic-looking weak spot." }, "bowl": { "score": 0.85, "reason": "Coherent top-down lacquer-style bowl with a smooth elliptical rim, dark ceramic body, warm inner rim, subtle highlight, and grounded contact shadow; chopsticks and rest sit cleanly beside it. No visible distortion or broken joins." }, "broth": { "score": 0.5, "reason": "Almost no visible liquid surface: a dark brown ring near the rim is the only candidate broth, and the noodle mat covers everything, so toppings read as pasted on top rather than sitting in soup; immersion and occlusion are weak." }, "steam": { "score": 0.5, "reason": "Translucent blurred wisps that change position and shape across frames (small wisps in desktop-1, larger drifting wisps in reduced-1/2 and pointer-left), so clearly animated rather than static. However, the blobs float in empty space to the upper-left and never convincingly emerge from the soup." }, "composition": { "score": 0.85, "reason": "Atmospheric dark wood scene with warm spotlight over a centered focal bowl, tasteful chopstick props, restrained palette and clean framing; mobile crop keeps the dish readable. Steam blobs are the only minor distraction." }, "vegan": { "score": 0.8, "reason": "Ingredients are mushrooms, bok choy, lotus root, bamboo, corn, tomatoes, scallions, chili, nori and seared sesame-topped slabs; no visible egg, fish, or shrimp. The grill-marked slabs are ambiguous (could read as chashu) but plausibly seared tofu, so an intermediate score." } }, { "noodles": { "score": 0.6, "reason": "Fine wavy yellow strands form a continuous noodle bed with consistent thickness, but they read as a flat, repetitive decorative mat with little strand overlap or volume." }, "toppings": { "score": 0.65, "reason": "Good variety: shiitake with radial gills, bok choy, lotus root discs, bamboo strips, scallion rings, chili threads and a dark nori sheet. However, the prominent cluster of glossy gold beads looks plastic, and the large crosshatched slices read as generic seared slabs." }, "bowl": { "score": 0.85, "reason": "Clean lacquer-style bowl: coherent top-down ellipse, dark rim with inner copper ring, subtle highlight, thick walls and a grounded contact shadow; no broken joins or perspective mismatch." }, "broth": { "score": 0.6, "reason": "A warm amber broth ring is visible around the noodle bed with restrained sheen, but the surface is mostly flat and hidden; toppings appear laid atop the noodle mat rather than sitting in liquid." }, "steam": { "score": 0.7, "reason": "Translucent irregular wisps shift position and shape across frames: faint at top-left in early desktop frames, drifting right with the right-pointer frame and left with the left-pointer frame, fading without obscuring food. Motion is supported, though normal-frame steam is faint and intermittent." }, "composition": { "score": 0.85, "reason": "Balanced dark scene: bowl as clear focal point with warm spotlight, chopsticks on a rest to the right, wood-grain background, no text or clutter; minor emptiness at the top edge." }, "vegan": { "score": 0.4, "reason": "Mushrooms, bok choy, lotus root, bamboo, corn-like kernels and nori are plant-appropriate, but the two large grill-marked slices strongly resemble chashu pork; appearance is meat-like though not unmistakable." } }, { "noodles": { "score": 0.7, "reason": "Fine yellow wavy strands form a continuous, plausible noodle bed around and under the toppings with varied bends; however they read mostly as a flat repeating line pattern with little volume, overlap or strand separation." }, "toppings": { "score": 0.75, "reason": "Good variety with readable shiitake gills, lotus root with holes, bamboo strips, bok choy with veins, nori, scallion rings, chili threads and seared tofu slices with grill marks; corn reads as a beaded blob and the tomatoes/corn kernels look like glossy spheres rather than food." }, "bowl": { "score": 0.85, "reason": "Clean top-down lacquer bowl: consistent elliptical rim, visible rim thickness, copper inner ring, coherent cavity and a grounding shadow; only minor flatness in material texture." }, "broth": { "score": 0.7, "reason": "Amber liquid ring surrounds the noodle bed and toppings sit within the cavity, giving believable immersion; but the surface is flat and uniform with almost no sheen, depth or texture." }, "steam": { "score": 0.45, "reason": "Translucent irregular wisps appear and drift between frames (cloud moves up-left across reduced frames), so not static; but the wisps are intermittent (absent in desktop-2/3) and in pointer frames hover over the empty table near the chopsticks and far-left edge, detached from the soup." }, "composition": { "score": 0.8, "reason": "Attractive dark wood scene with the bowl as clear focal point, chopsticks on a rest for context and warm restrained lighting; minor flaws include faint smudge artifacts over empty background, a subtle vertical seam on the right, and a very small bowl in the tall mobile frame." }, "vegan": { "score": 1.0, "reason": "All visible toppings are plant-based: seared tofu with grill marks, shiitake, lotus root, bamboo, corn, bok choy, nori, scallions, chili and tomatoes; no egg, meat or seafood shapes present." } } ], "diagnostics": { "vegan": 0.7333333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.6655555555555556, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "3708dc4132c4cf332c07d585215e5971c36735187fdb5039d8d864a08581b7d8", "gradedAt": "2026-09-30T23:20:54.006165+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5333333333333333, "toppings": 0.5011111111111111, "bowl": 0.7122222222222222, "broth": 0.5344444444444445, "steam": 0.66, "composition": 0.7555555555555555 }, "reward": 0.0, "runId": "google/gemini-3.8-flash/high", "rewardStdDev": 0.0798182542010517, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.48666666666666664, "bowl": 0.76, "broth": 0.47, "steam": 0.5966666666666667, "composition": 0.84 }, "reward": 0.5941666666666667, "sampleStdDev": 0.006825198409814448, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable highlights and overlap, but largely form repeated parallel flat loops rather than a supple, volumetric noodle tangle." }, "toppings": { "score": 0.48, "reason": "A veined leaf, scored mushroom cap and small scallion rings are recognizable. However, glossy red and green beads, rigid rectangular seaweed and densely crosshatched round pieces give much of the food a synthetic appearance." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner shading, consistent thickness and contact shadow form a coherent supported vessel. Its shallow visible wall and uniform highlight make the material and cavity somewhat schematic." }, "broth": { "score": 0.47, "reason": "Warm amber shading and sparse oil rings suggest soup, but the surface resembles a smooth solid gradient, with little liquid sheen or convincing immersion around the toppings." }, "steam": { "score": 0.57, "reason": "Normal frames show changing translucent haze above the soup while the food remains readable. The steam is broad and fog-like, lacking distinct irregular wisps; the sparse frames do not establish continuous rising and fading." }, "composition": { "score": 0.85, "reason": "A centered, clearly readable dish anchors the restrained dark wood setting, with harmonious amber and green accents and no distracting text or UI. Mobile framing preserves the whole bowl; no material compositional defect is observed." }, "vegan": { "score": 0, "reason": "A prominent egg half with white surrounding an orange yolk is unmistakably visible." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and highlights, but largely form repeated parallel flat loops with little tangled volume or believable immersion." }, "toppings": { "score": 0.48, "reason": "A veined leaf, scored mushroom and scallion rings are readable, but glossy colored spheres, rigid rectangular seaweed and heavily crosshatched disks make much of the food look synthetic." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner shadow, consistent thickness and tabletop contact form a coherent vessel. The shallow wall treatment and uniform highlight make it resemble a platter more than a deep ramen bowl." }, "broth": { "score": 0.46, "reason": "Warm amber shading and small oil rings suggest soup, but the surface looks like a smooth solid gradient, with ingredients layered above it rather than visibly immersed." }, "steam": { "score": 0.6, "reason": "Normal frames show changing translucent vapor above the soup, with spreading and fading shapes. Broad blurred clouds lack fine irregular wisps and partially wash out the rear toppings." }, "composition": { "score": 0.83, "reason": "A centered, clearly readable dish, warm restrained palette and dark wood setting create a balanced focal composition without text or clutter. Mobile framing preserves the bowl, though leaves substantial empty space." }, "vegan": { "score": 0, "reason": "A prominent egg half with white and orange yolk is unmistakably visible." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have clear highlights and overlaps, but form a repetitive stack of flat, parallel loops rather than a supple tangled volume." }, "toppings": { "score": 0.5, "reason": "A veined leaf, scored mushroom, scallion rings and egg are recognizable. However, glossy bead-like garnish, rigid rectangular seaweed and heavily gridded disks make much of the food look artificial." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner wall shading, consistent thickness and contact shadow form a coherent supported vessel. The broad uniform rim highlight and limited visible body give it a somewhat shallow, graphic appearance." }, "broth": { "score": 0.48, "reason": "Warm amber shading and scattered oil rings suggest soup, but the surface lacks convincing liquid sheen and immersion; most ingredients appear layered on top of a smooth solid field." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent clouds extending upward from the bowl. Their broad blurred masses read more as diffuse fog than fine irregular wisps, and wash out some upper toppings." }, "composition": { "score": 0.84, "reason": "A single centered dish remains clearly framed on desktop and mobile, with warm food colors against a restrained dark tabletop and no distracting text or UI. The lighting establishes a strong focal area, though the mobile layout leaves substantial empty space." }, "vegan": { "score": 0, "reason": "A clearly depicted egg half with white surrounding an orange yolk is visibly animal-derived." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.48333333333333334, "toppings": 0.4, "bowl": 0.6266666666666667, "broth": 0.5166666666666667, "steam": 0.6, "composition": 0.6333333333333333 }, "reward": 0.527, "sampleStdDev": 0.02137755832643197, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are a small cluster of thin yellow strands in the lower right. They read as noodles with some curvature, but they look like parallel flat ribbons, cover little of the bowl, and are largely hidden under scattered dots." }, "toppings": { "score": 0.4, "reason": "A bok choy leaf, shiitake, nori squares and menma are recognizable. The two waffle-textured discs are odd and read as generic disks. The egg with an orange yolk is plainly animal-derived. Red and green bead dots look like plastic spheres." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, studded edge, inner wall and highlight, and the perspective is consistent. It looks rather flat, with no visible base or foot, and the wall thickness is thin." }, "broth": { "score": 0.5, "reason": "The broth is a warm, orange-brown gradient with vignette depth. It looks fairly flat and opaque, with floating beads and rings pasted on the surface instead of sitting in the liquid." }, "steam": { "score": 0.6, "reason": "Soft translucent steam rises and drifts differently across frames, including a response to the pointer. It is fairly blurry and fog-like, and it partly covers the rim." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark vignette and is clearly the focal point, with warm colors. The scattered red and green dots clutter the food, and the toppings are crowded to one side." }, "vegan": { "score": 0.2, "reason": "A halved egg with a bright orange yolk is clearly visible, which is animal-derived. The other ingredients are plant-based." } }, { "noodles": { "score": 0.5, "reason": "Golden noodle strands are smooth and overlap with some sheen, but they form a small, thin cluster of mostly parallel curves. They look somewhat rigid and flat and occupy little of the bowl." }, "toppings": { "score": 0.4, "reason": "The bok choy, shiitake with a star cut, nori sheets and menma are recognizable. The two waffle-textured discs read as generic patterned disks. Scattered glossy red and green beads look like plastic. The egg with an orange yolk is prominent and looks non-vegan." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a highlight arc and a dotted decorative edge, with an inner wall and a shadow underneath. The rim is thick and somewhat flat, and the base and contact are barely visible." }, "broth": { "score": 0.5, "reason": "The warm brown broth has a gradient and depth at the edges. The surface is flat and murky, with floating rings and beads pasted on top rather than sitting in it." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes change between frames and drift sideways when the pointer moves. They are a bit blobby and foggy and partly veil the rim, though they do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark background with moody lighting and a clear focal dish. The scattered red and green dots add clutter, and the food is small within the bowl." }, "vegan": { "score": 0.2, "reason": "A halved egg with an orange yolk is clearly visible and animal-derived. The remaining ingredients are plant-based." } }, { "noodles": { "score": 0.5, "reason": "Golden noodle strands are smooth and overlap with some sheen, but they form a small, flat, rather uniform cluster. Their placement in the broth is only partly convincing." }, "toppings": { "score": 0.4, "reason": "Shiitake, bok choy and nori are recognizable. The two waffle-textured disks read as generic, and the nori is a rigid rectangle. A egg-like white-and-orange item sits on top, and the many red and green bead-like dots look plastic and cluttered." }, "bowl": { "score": 0.68, "reason": "The elliptical rim, inner wall, dotted glaze texture and highlight arc are coherent and consistent in perspective. The bowl has little visible thickness or base, and the support beneath it is weak." }, "broth": { "score": 0.55, "reason": "The warm orange-brown gradient has depth and a vignette. Its surface is fairly flat, and the floating beads and rings look pasted on rather than immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes drift and change between frames, and they react to the pointer by shifting direction. They are somewhat blurry and large, and they partly veil the bowl rim." }, "composition": { "score": 0.7, "reason": "The bowl is centered in a dark, moody scene with warm lighting, and the focal dish reads clearly. The scattered dots add clutter, and the lower third of the frame is empty." }, "vegan": { "score": 0.2, "reason": "A halved egg with an orange yolk is clearly visible, which is an animal-derived ingredient. The other toppings (mushroom, bok choy, nori, noodles) are plant-based." } } ], "diagnostics": { "vegan": 0.2 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6166666666666667, "toppings": 0.6166666666666667, "bowl": 0.75, "broth": 0.6166666666666667, "steam": 0.7833333333333333, "composition": 0.7933333333333333 }, "reward": 0.686, "sampleStdDev": 0.021633307652783894, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are smooth continuous yellow curves with a few bends and a small nest under the egg, but they are sparse, flat ribbon-like strands confined to one region, with large empty broth areas and little volume or believable immersion." }, "toppings": { "score": 0.6, "reason": "Recognizable stylized items: halved soft egg, shiitake with star cut, bok choy, nori sheets, menma sticks, scallion rings and chili threads. However the two waffle-grid discs read as cookies/patties rather than food, and scattered red/green dots look like candy sprinkles pasted on the surface." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical lacquer bowl from a high angle: rim, inner cavity, wall thickness at lower edge, dotted rim texture and a contact shadow beneath. Minor issues: a single long highlight arc on the rim looks slightly scratch-like and the dish reads shallow." }, "broth": { "score": 0.6, "reason": "Warm brown liquid surface with restrained specular glow from the steam light and a few bubbles, but it is flat with limited depth; scallion rings and colored dots float on top without occlusion, and left side is empty, weakening immersion." }, "steam": { "score": 0.8, "reason": "Across desktop frames the translucent wisps clearly change shape and position: rising above the bowl in frame 1, drifting upper-left by frame 3, and tilting toward the right/left in pointer frames. Wisps are irregular, fade continuously, and never obscure the food." }, "composition": { "score": 0.78, "reason": "Centered focal bowl in a dark atmospheric scene with a warm spotlight glow and soft vignette; presentation is clean and balanced. Slight noise from scattered colored dots, and the mobile frame leaves large empty margins." }, "vegan": { "score": 0.0, "reason": "A clearly visible soft-boiled egg half with white and orange yolk sits in the bowl \u2014 an unmistakable animal-derived topping." } }, { "noodles": { "score": 0.6, "reason": "Pale yellow strands form a loose nest with smooth continuous curves and consistent thickness, but the bundle is sparse, lies flatly in the lower bowl, and reads more like decorative spaghetti loops than a substantial noodle mass with volume." }, "toppings": { "score": 0.6, "reason": "Good readable items: glossy soft-egg half with orange yolk, shiitake slice with white cross-cut, bok choy leaf, nori squares, menma strips, scallion rings and red shreds. Defects: numerous scattered red/green glossy spheres read as plastic balls, and the two crosshatched brown discs look like waffle-pattern cookies rather than convincing chashu." }, "bowl": { "score": 0.75, "reason": "Coherent oval rim, inner cavity and wall band with dotted rim texture, rim highlight arc, lacquer-like brown material, and a grounding shadow beneath. Ellipse perspective is consistent; only the simplistic highlight stroke and purely decorative rim dots keep it from higher." }, "broth": { "score": 0.6, "reason": "Brown liquid surface has a gentle sheen, warm gradient and scattered oil/bubble rings, but toppings and noodles sit on top with token shadows rather than believable immersion, and the surface reads fairly flat and solid in the darker regions." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise from the bowl, and their shape and position clearly differ across desktop-1/2/3, confirming animated rise and fade; plumes also shift under pointer frames. Steam stays clear of the food without becoming opaque fog." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a restrained dark atmospheric scene with warm lighting, soft vignette, background bokeh dots and steam depth; no text or clutter. Slight deduction for the noisy scatter of glossy garnish balls around the bowl's interior." }, "vegan": { "score": 0, "reason": "A clearly visible soft-boiled egg half with bright orange yolk is an unmistakable animal-derived topping, which per rubric earns 0 despite otherwise plant-based ingredients (mushroom, nori, bok choy, menma, scallions)." } }, { "noodles": { "score": 0.7, "reason": "Golden strands form a continuous swirled bundle with consistent thickness and coherent overlaps in the lower-right of the bowl; however the wavy loop pattern is repetitive, the bundle sits flat, and most of the bowl shows no noodles at all." }, "toppings": { "score": 0.65, "reason": "Varied and mostly readable: shiitake cap with cut pattern, bok choy leaf, nori sheets, menma sticks, scallion rings, chili threads and an egg half. Defects: the two waffle-lattice discs read as cookies rather than recognizable food, and the scattered glossy red/green spheres look like decorative plastic beads pasted on the broth." }, "bowl": { "score": 0.75, "reason": "Coherent lacquer-style bowl with a distinct rim, inner cavity, dotted outer band and a consistent rim highlight across all frames; slight ellipse implies tilt while toppings are drawn circular (top-down), a mild perspective mismatch, but construction is intact and supported." }, "broth": { "score": 0.65, "reason": "Warm amber-brown liquid surface with restrained sheen, oil rings and bubble specks, and light occlusion under the egg and noodles; surface is otherwise very clean and flat, reading slightly like a gradient backdrop rather than deep soup." }, "steam": { "score": 0.75, "reason": "Translucent wisps rise from the soup and clearly evolve across time frames: plume shifts, stretches and drifts (moving right through desktop-2/3 and dispersing in pointer frames) without hiding the food. Some frames show blobby, detached fog masses off to the side rather than fully continuous wisps." }, "composition": { "score": 0.8, "reason": "Single centered bowl as clear focal point against a restrained dark atmospheric background with warm lighting, subtle line texture and no clutter or UI; presentation is balanced, though the scene is quite sparse and the food reads slightly small." }, "vegan": { "score": 0, "reason": "A soft-boiled egg half with white and orange yolk is clearly visible as a topping in every frame, an unmistakable animal-derived ingredient." } } ], "diagnostics": { "vegan": 0.0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.06666666666666667, "veganFailed": true }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "e8a71abe6091c460c073e5ecf186b77258f48ffd2e8e94b97befcec119d1ca8a", "gradedAt": "2026-09-30T23:21:57.408807+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.44222222222222224, "toppings": 0.508888888888889, "bowl": 0.6966666666666667, "broth": 0.56, "steam": 0.5877777777777777, "composition": 0.7455555555555555 }, "reward": 0.5737222222222222, "runId": "google/gemini-3.8-flash/medium", "rewardStdDev": 0.05013878885801038, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.46, "toppings": 0.4766666666666666, "bowl": 0.7566666666666667, "broth": 0.49666666666666665, "steam": 0.53, "composition": 0.7933333333333333 }, "reward": 0.5710000000000001, "sampleStdDev": 0.004582575694955844, "judgments": [ { "noodles": { "score": 0.46, "reason": "Thin curved strands are visible, but mostly read as sparse surface scribbles with abrupt ends rather than a supple, overlapping noodle volume." }, "toppings": { "score": 0.47, "reason": "Leaf veins, scallion rings and ridged rectangular pieces add detail, but glossy spherical kernels, crosshatched disks and a rigid rectangular nori sheet dominate the artificial ingredient depiction." }, "bowl": { "score": 0.75, "reason": "The elliptical rim, tapered walls and attached foot form a coherent vessel with readable depth. Very dark, minimally shaded walls and thin outline bands weaken the ceramic material and contact." }, "broth": { "score": 0.48, "reason": "Warm brown shading suggests soup, but the surface lacks convincing liquid sheen and immersion cues; many toppings appear layered onto a flat surface." }, "steam": { "score": 0.53, "reason": "Normal frames show translucent patches changing position and rising above the bowl. They resemble disconnected blurred blobs more than connected irregular wisps emerging from the soup." }, "composition": { "score": 0.79, "reason": "A single readable dish, diagonal chopsticks and restrained warm lighting create a balanced dark presentation without text or clutter. Broad haze and large empty areas slightly dilute the focal emphasis." }, "vegan": { "score": 0.66, "reason": "Greens, seaweed and corn-like kernels suggest plant ingredients, but the large browned crosshatched slices are ambiguous and could resemble meat rather than clearly identifiable tofu or mushrooms." } }, { "noodles": { "score": 0.46, "reason": "Thin curved strands are visible, but remain sparse, disconnected-looking surface squiggles with little coherent overlap or noodle volume." }, "toppings": { "score": 0.48, "reason": "Veined greens, scallion rings and textured rectangular pieces provide variety, but glossy spheres, crosshatched oval disks and a rigid rectangular nori sheet dominate the food depiction." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, tapered walls and attached foot form a coherent vessel with readable depth. Dark, minimally shaded walls and an oversized rounded foot weaken the ceramic finish." }, "broth": { "score": 0.5, "reason": "Warm brown shading suggests soup, but the surface lacks convincing liquid sheen and contact around ingredients; many toppings appear laid over a flat colored plane." }, "steam": { "score": 0.54, "reason": "Normal frames show translucent patches shifting upward and changing distribution. They resemble separated blurred blobs more than connected irregular wisps, with some haze crossing the food." }, "composition": { "score": 0.8, "reason": "The single dish remains the clear focus against a restrained dark background, with warm highlights and balancing green accents. The mobile arrangement is readable, though generous empty space and broad haze reduce presentation precision." }, "vegan": { "score": 0.68, "reason": "Greens, corn-like kernels, seaweed and rectangular plant-like toppings support vegan plausibility, but the large grilled oval slices are ambiguous and could read as meat rather than mushrooms or tofu." } }, { "noodles": { "score": 0.46, "reason": "Thin curved strands are visible, but mostly form isolated scribbles and loose loops rather than a coherent, overlapping volume of supple noodles." }, "toppings": { "score": 0.48, "reason": "Leaf veins, scallion rings and ridged rectangular pieces add detail, but glossy bead-like vegetables, large crosshatched disks and a rigid rectangular nori sheet dominate the artificial-looking assortment." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, tapered walls and centered foot form a coherent supported vessel. Warm rim highlights help define thickness, though the dark body has limited ceramic texture and cavity shading." }, "broth": { "score": 0.51, "reason": "Amber shading suggests broth, with some darker submerged strands, but the broad smooth surface lacks liquid reflections and most toppings appear layered above it without convincing immersion." }, "steam": { "score": 0.52, "reason": "Normal frames show translucent patches changing position and extending upward. They resemble separated blurred puffs more than connected irregular wisps, with some haze washing over the food." }, "composition": { "score": 0.79, "reason": "A single readable dish remains the focus against a restrained dark background, with warm colors and diagonally placed chopsticks. The framing is uncluttered across desktop and mobile, though lighting feels somewhat graphic and uneven." }, "vegan": { "score": 0.65, "reason": "Greens, seaweed, corn-like kernels and rectangular plant-like pieces support a vegan reading, but the large browned crosshatched slices are ambiguous and could resemble meat rather than clearly identifiable mushrooms or tofu." } } ], "diagnostics": { "vegan": 0.6633333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.4666666666666667, "bowl": 0.6, "broth": 0.55, "steam": 0.6, "composition": 0.6333333333333333 }, "reward": 0.525, "sampleStdDev": 0.009013878188659966, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin, short, scribbly squiggles scattered on the broth surface. They don't form a convincing mass and read as thin wire-like curls rather than supple strands with volume." }, "toppings": { "score": 0.5, "reason": "Nori sheet, bok choy leaf, mushroom-like domes, menma, corn and scallions are all recognizable. The mushrooms look like scored beige domes, and the red beads look like glossy plastic. The nori is a rigid dark rectangle, and the black tangle looks odd." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent copper-lit rim, a dark body and a foot with ring lines. The perspective is plausible and the base sits on the table. The lower body is very dark and murky, and the rim looks thin." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a warm gradient and some glow, so it reads as liquid. The glossy red beads float on it like pasted spheres. Immersion is weak and the surface is somewhat flat." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and change between frames, rising and fading. Some are blobby and mostly appear as separate puffs far from the bowl. In the pointer frames the steam gets denser and partly covers the food." }, "composition": { "score": 0.65, "reason": "The bowl is centered as a clear focal point in a moody dark scene with warm colors. The nori partly floats awkwardly above the bowl and there is a lot of empty dark space." }, "vegan": { "score": 0.8, "reason": "Toppings are plant-based: nori, bok choy, corn, menma and mushroom-like domes. There are no visible animal products, though the domes could be mistaken for egg or meat." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, small, wispy squiggles scattered on the broth surface. They read as bean sprouts or shreds rather than a supple noodle mass, and there is little volume or overlap." }, "toppings": { "score": 0.45, "reason": "Toppings include a bok choy leaf, nori, scallion rings, corn, and enoki-like bundles. The two large round, crosshatched items look like eggs or buns, and the red beads look like plastic spheres. The nori is a flat dark card and the dark squiggles are unclear." }, "bowl": { "score": 0.6, "reason": "The rim is a coherent ellipse with a copper highlight, and the dark body and foot are plausible with a subtle shadow. The body is very dark and faintly outlined, and the foot lines are a bit vague. The chopsticks look fine." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange-red with a gradient and glow, and it reads as liquid. The red beads float on it like glossy spheres. Immersion of the toppings is weak, and the surface is fairly flat." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and change between frames. They are blobby in places and partly overlap the nori, but they rise and fade without hiding the food." }, "composition": { "score": 0.65, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm color contrast and no distracting UI. The lower part of the frame is empty and the nori placement is awkward." }, "vegan": { "score": 0.6, "reason": "The ingredients are mostly plants: greens, corn, scallion, nori and enoki. The two large round items resemble eggs or meat, which makes the appearance ambiguous." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, short, squiggly beige strands scattered on the broth surface. They read as bean sprouts or scribbles rather than a supple, voluminous noodle mass. There is little overlap or volume, and some strands are flat loops." }, "toppings": { "score": 0.45, "reason": "The toppings include nori, bok choy, scallion slices, corn and bamboo shoots, which gives variety. The two mushroom-like domes with a grid pattern look like glossy eggs or buns. The red beads look like plastic spheres, and the black noodle bundle is odd. The nori is a rigid, rectangular card standing upright." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim with a copper edge, a dark body and a foot, plus a soft shadow. The lower walls are very dark and murky. The rim perspective is slightly inconsistent with the base, and the contact with the table is weak." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange-red with a gradient and some depth. It looks fairly flat, and the red beads sit on top like glossy spheres. The ingredients are only loosely integrated into the liquid." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and change between frames, rising up and to the left. They look like blurry blobs at times and are not well connected to the broth surface. Motion is visible, but the plumes appear somewhat disconnected." }, "composition": { "score": 0.6, "reason": "The bowl is a centered focal point in a dark, moody scene with pleasing warm colors. The oversized nori and the tilted chopsticks distract a little, and the lower third is empty." }, "vegan": { "score": 0.6, "reason": "The toppings are mostly plant-based: nori, bok choy, corn, scallion and bamboo. The two round gridded domes could be read as eggs, which leaves the appearance ambiguous." } } ], "diagnostics": { "vegan": 0.6666666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.4666666666666667, "toppings": 0.5833333333333334, "bowl": 0.7333333333333333, "broth": 0.6333333333333333, "steam": 0.6333333333333333, "composition": 0.81 }, "reward": 0.6251666666666666, "sampleStdDev": 0.013650396819628858, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles appear as thin flat tan squiggle lines and small loops lying on the broth surface; recognizable but artificial, lacking volume, layering or believable immersion." }, "toppings": { "score": 0.6, "reason": "Good variety: scallion rings, menma strips, leaning nori sheet, a leafy green and crosshatched grilled slices read well; but corn as glossy yellow beads and red items as uniform specular spheres look plastic, and discs are flat ellipses." }, "bowl": { "score": 0.7, "reason": "Coherent rim ellipse, tapering dark body and visible foot with consistent perspective; chopsticks rest plausibly on the rim. Exterior shading is very dark, so wall form reads faintly but no broken joins." }, "broth": { "score": 0.6, "reason": "Warm orange-brown surface with a soft radial gradient and restrained sheen; ingredients sit on top with faint shadows, but contact/immersion is shallow and the liquid reads as a flat gradient plane in places." }, "steam": { "score": 0.55, "reason": "Frames show wisps changing position and intensity over time (rising near nori, larger plumes in pointer-left frame), so motion is supported; however puffs are soft disconnected blobs, partly detached from the bowl." }, "composition": { "score": 0.78, "reason": "Balanced centered dish with warm spotlight glow against a dark atmospheric background, chopsticks adding diagonal interest, no clutter or UI; large empty black margins slightly thin out the framing." }, "vegan": { "score": 0.8, "reason": "Visible ingredients (nori, scallions, bamboo shoots, leafy green, corn-like kernels, tomato/chili-like spheres, crosshatched slices plausibly tofu) suggest vegan ramen; nothing clearly animal-derived, though the scored discs are ambiguous." } }, { "noodles": { "score": 0.4, "reason": "Noodles appear only as thin flat gold squiggle strokes (an S-curve with a loop at right, scattered wavy segments) drawn on the broth surface. They lack strand bundles, volume, thickness variation and overlap, reading as wire scribbles rather than supple noodles." }, "toppings": { "score": 0.6, "reason": "Good variety and some readable details: scallion rings, corn kernels, bamboo-shoot strips, a nori sheet with faint seam lines, seaweed strands and a leaf garnish. Weaknesses: the large crosshatched tan ellipsoids dominating the center look plastic and ambiguous, and the uniform glossy red spheres read as plastic beads rather than food." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with a light rim ellipse, tapered walls, pedestal foot and consistent angled perspective; chopsticks rest believably on the rim. Minor shortcomings: the exterior is so dark the lower silhouette nearly merges with the background, and the rim highlight is a simple uniform stroke." }, "broth": { "score": 0.65, "reason": "Smooth orange-brown liquid surface with soft highlight and edge darkening; toppings sit within the rim and the nori dips into the liquid. Defects: the surface is glassy-smooth with little texture, and most toppings (corn, red spheres) float with weak contact shadows, looking slightly pasted on." }, "steam": { "score": 0.65, "reason": "Across the time-ordered frames the blurred wisps shift and disperse (blobs near the nori at 1s drift up and left by 2.9-4.7s, and stronger plumes rise over the bowl by ~9.8s), indicating animated rising steam that fades without hiding food. Deduction: many puffs read as disconnected round bokeh blobs rather than continuous wisps; reduced frames correctly show none." }, "composition": { "score": 0.85, "reason": "Polished dark atmospheric presentation: warm glowing bowl as clear focal point, balanced with chopsticks on the right, subtle light beam and tabletop, no text or clutter. Minor issue: stray soft blobs in the upper-left corner read as floating bokeh rather than steam." }, "vegan": { "score": 0.6, "reason": "Visible ingredients are mostly plant-based (nori, scallions, corn, bamboo shoots, seaweed, leaf, tomato-like spheres). The two large crosshatched tan ellipsoids are ambiguous; they could read as grilled tofu/mushroom or as a fish-cake-like slice, so an intermediate score is used." } }, { "noodles": { "score": 0.5, "reason": "Pale strands read as thin flat squiggle strokes scattered on the broth surface rather than a coherent noodle bed; they lack thickness, volume, believable immersion and coherent overlap, though bends are noodle-like rather than a maze." }, "toppings": { "score": 0.55, "reason": "Good variety: nori sheet leaning on the rim, menma strips, scallion rings, a leafy sprig, corn, and a seeded round bun/patty with lattice marks. Defects: corn and red chili read as glossy plastic beads with strong speculars, the large spheres look plasticky, and the nori is a rigid flat card extending past the rim." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical copper rim, tapering dark body, visible foot, and believable chopsticks resting on the rim to the table. Minor issues: shallow depth and a very dark exterior that flattens form reading, plus a subtle lip alignment on the far side." }, "broth": { "score": 0.65, "reason": "Rich miso-toned liquid surface with warm center glow, gradient depth and restrained sheen; the bun and leaf show contact shadows. However, noodle squiggles and glossy beads appear pasted on the surface rather than immersed, and the surface is somewhat flat." }, "steam": { "score": 0.7, "reason": "Across desktop frames (1s\u21929.8s) translucent wisps clearly rise, drift left, spread and fade near the nori and bowl rim, so motion is supported. Wisps are somewhat blobby and disconnected, and the larger fog at the late frame partially veils the nori and left toppings. Reduced frames appropriately show near-none." }, "composition": { "score": 0.8, "reason": "Well-centered focal dish, warm broth glow against a restrained dark tabletop, diagonal chopsticks and leaning nori balance the frame, no text/UI clutter. Slightly empty left half and dark bowl exterior slightly reduce impact." }, "vegan": { "score": 1.0, "reason": "All visible ingredients read plant-based: nori, scallion, corn, menma, leafy green, mushroom-like dark strands, and round brown patties with sesame/lattice that read as buns or grilled tofu; no egg, meat, fish or dairy shapes." } } ], "diagnostics": { "vegan": 0.8 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.71, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.70000000000073, "frameIntervalP95Ms": 83.39999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5d8a475ae3d7c0da689522ef5e0e7c206b3660ff0cd2fede3ecca132985c3956", "gradedAt": "2026-09-30T23:24:27.462768+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.48444444444444446, "toppings": 0.6033333333333333, "bowl": 0.7633333333333333, "broth": 0.6177777777777778, "steam": 0.6444444444444445, "composition": 0.7777777777777778 }, "reward": 0.6366111111111111, "runId": "meta/muse-spark-1.3/high", "rewardStdDev": 0.10830080708293363, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.45333333333333337, "toppings": 0.5766666666666667, "bowl": 0.85, "broth": 0.5866666666666667, "steam": 0.6333333333333333, "composition": 0.8433333333333333 }, "reward": 0.6428333333333334, "sampleStdDev": 0.004536885862938758, "judgments": [ { "noodles": { "score": 0.45, "reason": "Cream-colored strands are readable, but mostly form sparse, nearly parallel shallow arcs with little supple bending, overlap or noodle volume." }, "toppings": { "score": 0.57, "reason": "Mushroom gills, leafy greens, scallion rings, tofu and corn provide recognizable variety. Shapes remain very smooth and schematic; corn resembles glossy balls and nori a thick rounded strip." }, "bowl": { "score": 0.85, "reason": "A continuous elliptical rim, consistent curved walls and subtle blue ceramic highlights establish convincing depth and thickness. Construction is coherent, though the dark base and diffuse shadow make contact slightly indistinct." }, "broth": { "score": 0.6, "reason": "Warm amber color and soft highlights suggest appetizing broth, with some ingredient occlusion at the front rim. The surface is overly uniform and ingredients have limited convincing immersion or contact effects." }, "steam": { "score": 0.62, "reason": "Normal frames show translucent haze changing position and shape above the soup without obscuring it. The steam reads more as broad blurred clouds than fine rising, spreading wisps." }, "composition": { "score": 0.84, "reason": "The centered dish, warm soup against a cool dark bowl, restrained background and uncluttered mobile framing create a clear focal point. The large bowl body somewhat outweighs the narrow visible food surface." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, greens, scallions, corn and seaweed; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.45, "reason": "Pale strands are readable, but mostly form sparse, nearly parallel surface lines with little supple bending, overlap, or noodle volume." }, "toppings": { "score": 0.58, "reason": "Mushroom gills, green leaves, scallion rings, tofu and corn provide recognizable variety. Shapes remain overly smooth and repetitive, and the nori resembles a thick rounded strip." }, "bowl": { "score": 0.84, "reason": "The elliptical rim joins coherently to rounded walls, with convincing thickness and restrained blue ceramic shading. The base contact is somewhat vague, but no material perspective defect is observed." }, "broth": { "score": 0.57, "reason": "Warm amber color and gentle highlights suggest soup, with some foreground occlusion. The surface is unusually smooth and banded, and ingredients appear mostly arranged on top rather than immersed." }, "steam": { "score": 0.64, "reason": "Normal frames show changing translucent vapor above the soup without obscuring the food. The vapor is broad and blurry, resembling diffuse clouds more than articulated rising wisps; continuous flow cannot be established from these samples." }, "composition": { "score": 0.85, "reason": "A centered dish, warm food against a cool dark bowl, restrained background particles, and uncluttered desktop and mobile framing create a clear focal point. The low angle gives the bowl body somewhat more prominence than the food." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly read as mushrooms, tofu, greens, scallions, corn and seaweed; no clearly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Pale strands are readable and consistently thin, but mostly form sparse, nearly parallel surface lines with little supple bending, overlap or noodle volume." }, "toppings": { "score": 0.58, "reason": "Mushroom gills, folded greens, scallion rings, tofu and corn provide recognizable variety. Shapes remain overly smooth and schematic, especially the rounded nori card and bead-like corn." }, "bowl": { "score": 0.86, "reason": "A continuous elliptical rim joins coherent curved walls and a rounded base; soft blue highlights convincingly suggest dark ceramic. No material construction defect is observed, though ground contact is indistinct." }, "broth": { "score": 0.59, "reason": "Warm amber color and restrained highlights are appetizing, with some ingredient occlusion at the front rim. The surface is unusually smooth and banded, and ingredients show limited immersion or liquid contact." }, "steam": { "score": 0.64, "reason": "Translucent vapor changes position and spread across the normal frames without obscuring the food. Broad blurred patches read more as diffuse haze than finely articulated rising wisps." }, "composition": { "score": 0.84, "reason": "The centered dish remains clear on desktop and mobile, with warm soup contrasting against cool ceramic and a restrained dark background. No distracting text or clutter appears; the large bowl body slightly outweighs the narrow food surface." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, greens, scallions, tofu, corn and seaweed; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.5, "bowl": 0.6066666666666667, "broth": 0.55, "steam": 0.5333333333333333, "composition": 0.64 }, "reward": 0.5253333333333334, "sampleStdDev": 0.003752776749732603, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin pale flat streaks on the surface, with no volume or supple strands. They read as thin lines rather than a noodle mass." }, "toppings": { "score": 0.5, "reason": "Mushrooms, bok choy, scallion rings, corn and a nori strip are recognizable. Scallion slices look like flat green discs, and everything is tiny and squashed by the steep viewing angle." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent rim, a rounded body and a soft shadow. It is very dark and flat, with little ceramic highlight. The rim is wide relative to the very shallow visible soup." }, "broth": { "score": 0.55, "reason": "Orange broth with a warm glow and a rim ring looks appetizing. The surface is thin and flat, with little texture, and the toppings sit on top rather than in it." }, "steam": { "score": 0.5, "reason": "Soft translucent grey plumes change shape across frames and rise above the bowl. They are blobby and blurry and only faintly wispy." }, "composition": { "score": 0.62, "reason": "The bowl is centered in a dark, atmospheric scene with a warm glow. The bowl is large and dark compared with the small food area, and there is no clutter." }, "vegan": { "score": 0.9, "reason": "Only mushrooms, greens, corn, seaweed and noodles are visible, with no animal products. The pale block could be tofu or daikon." } }, { "noodles": { "score": 0.4, "reason": "Noodles appear only as thin flat cream lines on the surface. They are barely readable as strands and have little volume or overlap, with no sense of immersion in the broth." }, "toppings": { "score": 0.5, "reason": "Small mushrooms with gills, bok choy leaves, scallion rings, tofu and corn are all recognizable. They are tiny and somewhat plastic-looking, and the scallion discs are flat and glossy. Nori is a dark rectangle. Variety is decent but the detail is low." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent rim, a smooth hemispherical body and soft shading with a warm glow beneath. It looks a bit flat and dark, and the base is indistinct with weak contact or foot. The rim perspective is fine." }, "broth": { "score": 0.55, "reason": "The orange broth has a readable gradient and sheen, and the ingredients sit roughly on it. The visible surface is thin and lacks texture or depth, and the toppings look slightly pasted on." }, "steam": { "score": 0.55, "reason": "Soft translucent blurry plumes shift between frames and drift up and fade. They look like faint blobs rather than wisps, and they are fairly diffuse and low contrast, but they do not hide the food." }, "composition": { "score": 0.65, "reason": "The dark atmospheric scene has a centered bowl as the clear focal point, a warm palette and no clutter. The food is small within the bowl, and the large dark bowl body dominates the frame." }, "vegan": { "score": 0.9, "reason": "Only mushrooms, greens, tofu, corn, nori and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are only thin pale flat streaks on the soup surface. They are readable as noodles but not voluminous or supple, and they show no bends or overlap." }, "toppings": { "score": 0.5, "reason": "Mushrooms with gills, green leaves, scallion rings, tofu/white block and corn kernels give decent variety. They are small and somewhat glossy and disc-like, and the nori is a dark rectangle at the rim." }, "bowl": { "score": 0.6, "reason": "The dark navy bowl has a coherent rim, a rounded body and a soft glow beneath it. The view is a low side angle and the rim/interior perspective is slightly odd. The bowl is fairly flat and featureless, and it looks lightly floating over its shadow." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and an inner rim shading. It is fairly flat, and the toppings sit on top of it more than in it." }, "steam": { "score": 0.55, "reason": "Soft translucent blurry plumes change shape and position across frames and fade as they rise. They read as blobs rather than wisps, and they are faint." }, "composition": { "score": 0.65, "reason": "The bowl is centered in a dark atmospheric scene with warm lighting, and there is no clutter. The food is tiny relative to the big dark bowl body, so the focal interest is small." }, "vegan": { "score": 0.9, "reason": "Only mushrooms, greens, tofu, corn and noodles are visible. There is no meat or egg." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.7333333333333333, "bowl": 0.8333333333333334, "broth": 0.7166666666666667, "steam": 0.7666666666666667, "composition": 0.85 }, "reward": 0.7416666666666667, "sampleStdDev": 0.008036375634160854, "judgments": [ { "noodles": { "score": 0.6, "reason": "Pale-yellow noodle strands are readable on the broth surface with gentle bends, but they render as flat, repetitive parallel arcs with no volume, highlight, or immersion depth." }, "toppings": { "score": 0.75, "reason": "Good variety: glossy snow peas with visible peas, sliced scallion rings, mushroom slices with gill detail, corn kernels, nori rectangles and pale cylinders. Mostly organic and readable, though corn is slightly ball-like, scallions are flat ellipses, and the white cylinders are ambiguous." }, "bowl": { "score": 0.8, "reason": "Coherent navy ceramic bowl with a consistent rim ellipse, visible interior wall, smooth shaded body and rim highlight. Minor defect: it floats with only a soft glow/faint reflection beneath rather than a clear contact shadow, and lacks a foot/base detail." }, "broth": { "score": 0.75, "reason": "Warm orange-red surface with restrained sheen, edge-to-center depth gradient and small contact shadows under toppings, so ingredients read as sitting in the liquid. Nori rectangles look slightly pasted flat." }, "steam": { "score": 0.8, "reason": "Across desktop-1\u21922\u21923 the wisps clearly rise, change shape, thin and disperse; they are translucent, irregular, and never obscure the dish. Slightly blobby/cloudy rather than fine wisps, but animated and non-mechanical." }, "composition": { "score": 0.85, "reason": "Balanced centered focal dish against a dark atmospheric background with subtle ember particles and warm under-glow; no text, UI or competing elements. Bowl slightly cropped at the frame edge on mobile, and the floating presentation is a stylistic choice that mostly works." }, "vegan": { "score": 0.9, "reason": "Visible toppings (snow peas, scallions, mushrooms, corn, nori, pale vegetable cylinders, noodles, plant-based broth) are all plausibly vegan; no egg, meat, fish or dairy shapes are evident." } }, { "noodles": { "score": 0.6, "reason": "Noodles read as smooth pale-yellow strands sweeping across the broth with coherent curves, but they are uniformly thin flat strokes with little volume, thickness variation, or believable plunge into the liquid \u2014 readable yet simplified." }, "toppings": { "score": 0.75, "reason": "Good variety: gilled brown mushroom slices, green edamame-like pods, scallion rings, corn kernels, a dark nori sheet, and a white cylinder. Shapes are recognizable and mostly organic, though somewhat glossy/simplified and the plain white cylinder is featureless." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim, consistent above-angle perspective, smooth deep navy ceramic body with soft highlight and believable wall thickness; a faint smudge on the body and no visible foot are the only minor shortcomings." }, "broth": { "score": 0.7, "reason": "Warm amber surface with restrained sheen and toppings sitting on it reads acceptably, but the liquid is somewhat flat and graphic with limited depth or occlusion around the ingredients." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps change position and shape across the time-ordered frames (rising plume, drift right, diffuse left), showing continuous motion without obscuring the food; slightly blobby cloud shapes keep it from higher." }, "composition": { "score": 0.85, "reason": "Clean centered focal dish in a dark atmospheric scene with a warm under-glow, subtle floor line, and sparse dust motes; no clutter or competing elements, though the bowl's hover above the glow is a stylistic choice with minimal contact grounding." }, "vegan": { "score": 0.95, "reason": "Visible ingredients (mushrooms, edamame, scallion, corn, nori, white vegetable cylinder, plant-colored broth) are all plausibly vegan; no eggs, meat, or seafood appear." } }, { "noodles": { "score": 0.6, "reason": "Noodle strands are readable as pale wavy lines weaving across the broth, but they are sparse, thin and flat with little volume or layered nesting; several run as parallel decorative strokes rather than a coherent submerged noodle mass." }, "toppings": { "score": 0.7, "reason": "Good variety: edamame pods, scallion rings, mushroom caps (one with readable gill lines), a corn cluster, negi-like white pieces and a dark nori strip. Forms are somewhat simplified \u2014 flat scallion disks with square white centers and pill-shaped white pieces \u2014 but most read as their ingredients." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic hemisphere with a consistent elliptical rim, smooth glossy body, plausible depth and soft highlights joining rim to wall; no broken geometry. Minor deduction for the footless floating look with only a glow beneath rather than firm ground contact." }, "broth": { "score": 0.7, "reason": "Orange-red liquid surface with restrained sheen, subtle edge highlights and toppings that cast slight shadows and appear seated in the broth; a bit flat overall with limited depth beyond the surface plane." }, "steam": { "score": 0.75, "reason": "Frames at 1006/2800/4598/6980/9348ms show translucent gray wisps that rise, disperse and change shape and position continuously, never obscuring the food; they are quite diffuse and at times split into soft separate blobs, losing definition. Reduced frames correctly freeze it." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point on an atmospheric dark backdrop with warm under-glow, faint ember particles and reflection; no text, UI or clutter, and the mobile crop stays balanced. Background polish slightly outshines the simplified food detail." }, "vegan": { "score": 0.95, "reason": "Visible toppings \u2014 edamame, scallion, mushrooms, corn, negi, nori \u2014 are all plant-derived with no meat, egg or fish-like forms; broth appearance is consistent with miso/vegetable style." } } ], "diagnostics": { "vegan": 0.9333333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9444444444444444, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "789c0029035db19a1eb23ff4592e25300bc4fccafd8f1ed39b478ee1ca1fca51", "gradedAt": "2026-09-30T23:26:11.868166+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.45111111111111113, "toppings": 0.5322222222222223, "bowl": 0.5144444444444445, "broth": 0.5033333333333333, "steam": 0.47, "composition": 0.6777777777777778 }, "reward": 0.5147777777777778, "runId": "meta/muse-spark-1.3/low", "rewardStdDev": 0.07697769037074843, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.38666666666666666, "toppings": 0.5066666666666667, "bowl": 0.42666666666666664, "broth": 0.46, "steam": 0.37666666666666665, "composition": 0.6666666666666667 }, "reward": 0.4585, "sampleStdDev": 0.008046738469715547, "judgments": [ { "noodles": { "score": 0.4, "reason": "Continuous golden strands are readable, but almost all form parallel shallow arcs with exposed ends, little overlap, and no convincing tangled volume." }, "toppings": { "score": 0.52, "reason": "Mushroom caps and stems, pointed leaves, scallion rings, and tofu are recognizable. Shapes remain schematic: tofu is flat squares, nori a rigid striped rectangle, and small vegetables glossy spheres." }, "bowl": { "score": 0.43, "reason": "Nested elliptical edges provide a coherent rim and some highlights, but the vessel reads as a shallow ring without substantial walls or a base, floating above its detached shadow." }, "broth": { "score": 0.46, "reason": "Warm amber shading and scattered specks suggest soup, but the surface looks like a solid flat fill, with little liquid sheen or believable immersion around toppings." }, "steam": { "score": 0.38, "reason": "Normal frames show the pale clouds moving upward and changing shape. However, bright opaque cores and detached blurred blobs resemble glowing smoke rather than translucent wisps emerging continuously from soup." }, "composition": { "score": 0.66, "reason": "The central dish, diagonal chopsticks, and restrained warm palette read clearly against a dark uncluttered background. Bright steam competes for attention; mobile framing leaves extensive empty space and clips the chopsticks." }, "vegan": { "score": 1, "reason": "Visible mushroom, tofu-like squares, greens, seaweed, corn-like kernels, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.38, "reason": "Consistent golden strands are readable, but nearly parallel shallow arcs resemble rigid lines rather than supple, overlapping noodles with volume." }, "toppings": { "score": 0.5, "reason": "Mushroom caps, green leaves, scallion rings and tofu are recognizable. Shapes remain schematic: tofu is flat squares, nori a rigid striped rectangle, and small vegetables glossy spheres." }, "bowl": { "score": 0.42, "reason": "The elliptical rim has coherent outlines and subtle highlights, but the bowl reads as a shallow ring without developed walls or a base and floats conspicuously above its detached shadow." }, "broth": { "score": 0.46, "reason": "Warm amber shading and scattered flecks suggest soup, but the surface lacks liquid sheen and depth; ingredients appear laid onto a flat surface with little immersion or contact." }, "steam": { "score": 0.35, "reason": "Normal frames show pale clouds moving upward and changing shape. They resemble disconnected luminous fog blobs with bright opaque cores rather than delicate translucent wisps emerging continuously from the soup." }, "composition": { "score": 0.66, "reason": "A clear central dish, warm restrained palette and diagonal chopsticks create a readable dark composition without UI clutter. The bright steam competes for attention, and mobile framing clips the chopsticks." }, "vegan": { "score": 1, "reason": "Visible mushrooms, tofu-like pale squares, greens, seaweed, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.38, "reason": "Continuous golden strands are visible, but mostly form parallel, nearly straight stripes with little supple bending, overlap or submerged volume." }, "toppings": { "score": 0.5, "reason": "Mushroom caps, green leaves, scallion rings and tofu are recognizable, but simplified semicircles, flat squares and glossy spheres lack organic texture; nori is a rigid rectangular panel." }, "bowl": { "score": 0.43, "reason": "The elliptical rim is coherent and highlighted, but the vessel reads as a shallow ring without a convincing outer wall or base, floating well above its detached shadow." }, "broth": { "score": 0.46, "reason": "Warm amber shading and scattered flecks suggest soup, but the surface looks like a flat opaque fill with little liquid sheen or believable ingredient immersion." }, "steam": { "score": 0.4, "reason": "Normal frames show pale clouds rising and changing shape without obscuring the food, but their bright dense centers and disconnected blurred blobs resemble glowing fog rather than translucent wisps." }, "composition": { "score": 0.68, "reason": "The uncluttered dark scene, warm palette and diagonal chopsticks give the dish a clear focus. Bright steam competes for attention, and the mobile dish is small with chopsticks cropped at the edge." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, scallions, corn-like kernels, tofu-like squares, nori and noodles plausibly form vegan ramen; no material animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.43333333333333335, "toppings": 0.47333333333333333, "bowl": 0.5, "broth": 0.48333333333333334, "steam": 0.4666666666666667, "composition": 0.6 }, "reward": 0.48333333333333334, "sampleStdDev": 0.02428133714055577, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin parallel orange-yellow lines that read as stripes across the broth. They are flat, rigid and repetitive, with little volume or overlap." }, "toppings": { "score": 0.47, "reason": "Nori, mushrooms, tofu cubes, scallion rings, corn and a pair of leaves are all recognizable. The tofu cubes are flat cards, the scallion rings are neon green and plastic-looking, and the nori is a rigid rectangle. Variety is decent but depiction is generic." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim ring and a dark blue body, but the perspective is odd. It floats above a detached shadow, the base is not visible, and the walls are thin. The chopsticks look flat and stuck on." }, "broth": { "score": 0.45, "reason": "The orange broth has a gradient and some specks, but it looks like a flat matte disc. Toppings sit on top rather than in it, with little immersion or sheen." }, "steam": { "score": 0.5, "reason": "Steam rises and drifts across the frames, and it varies in shape. It is a blown-out bright glowing blob, which looks like a light rather than wisps, and in places it is opaque." }, "composition": { "score": 0.6, "reason": "The bowl is centered and is the clear focal point against a dark, warm atmosphere. The floating bowl with a detached shadow and the overexposed steam weaken the presentation." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, nori, corn, scallions and leaves, with no visible meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin parallel golden lines with gentle curves on the broth surface. They read as noodles but look flat and uniform, with little volume or overlap, and sit on top of the broth." }, "toppings": { "score": 0.5, "reason": "Mushrooms, a nori sheet, leaves, tofu cubes, corn and scallion rings are all recognizable. The nori is a rigid dark rectangle, the tofu pieces are plain flat squares, and the scallion rings are neon-green circles. The variety is fine, but the forms are simple and symbolic." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent elliptical rim with a dark interior lip and a navy body. The body is shallow and flat, with no visible base or foot, and it appears to float above a detached shadow ellipse. Support and contact are weak." }, "broth": { "score": 0.5, "reason": "The broth is an orange gradient with a soft sheen and speckles. It reads as a flat disc, and the toppings sit on it without ripples or immersion." }, "steam": { "score": 0.5, "reason": "The steam moves and changes across frames, rising and drifting as translucent wisps. It is a glowing blob with a bright hotspot that looks more like a light flare than steam, and it is detached from the bowl in the early frames." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark, atmospheric background with warm colors and the food as the clear focus. The floating bowl with its detached shadow, the bright steam glare and the chopsticks that vanish into the dark all weaken the presentation." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are mushrooms, tofu, nori, leaves, corn and scallion. There is no meat, egg or fish." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin flat curved streaks across the broth. They read as noodles but look like rigid parallel ribbons, with no volume or overlap." }, "toppings": { "score": 0.45, "reason": "Mushrooms, nori, scallion rings, tofu cubes and leaves are recognizable. Tofu blocks and ring outlines are generic, the nori is a stiff rectangle, and the corn dots are simple." }, "bowl": { "score": 0.5, "reason": "The wide ellipse bowl has a rim, a dark interior edge and a navy wall, but it is very shallow and looks like a wok or pan. It floats above a detached shadow with no base, and the chopsticks appear to attach behind the rim." }, "broth": { "score": 0.5, "reason": "The orange broth has a warm gradient and some speckle, but it reads as a flat, opaque surface. Toppings sit on top rather than being immersed." }, "steam": { "score": 0.4, "reason": "Steam moves and changes across frames, but it forms a blown-out, bright glowing blob that looks like a light rather than wisps. It drifts far from the bowl and is sometimes disconnected." }, "composition": { "score": 0.6, "reason": "The bowl is centered in a dark, atmospheric scene with warm colors and a clear focus. The floating shadow and the overexposed steam glow detract from it." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are tofu, mushrooms, nori, scallions, corn and greens, with no meat or egg visible." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5333333333333333, "toppings": 0.6166666666666667, "bowl": 0.6166666666666667, "broth": 0.5666666666666667, "steam": 0.5666666666666667, "composition": 0.7666666666666667 }, "reward": 0.6025, "sampleStdDev": 0.04394598957811731, "judgments": [ { "noodles": { "score": 0.55, "reason": "Strands are continuous and readable as noodles, but they lie as near-parallel horizontal ribbons with only slight, uniform bends \u2014 flat and mechanical rather than supple, with no convincing strand volume or overlap in the broth." }, "toppings": { "score": 0.6, "reason": "Good variety: mushroom caps with stems, a striped nori sheet, green leaves, scallion rings, corn dots and orange spheres. The tofu pieces are flat white parallelogram cards rather than cubes, and toppings read as pasted shapes more than seated ingredients." }, "bowl": { "score": 0.55, "reason": "The rim is a coherent thin ellipse with a dark body, but the vessel reads as a shallow pan: hairline rim with no cavity depth or wall thickness, chopsticks enter awkwardly at the rim edge, and the bowl visibly floats above a detached shadow ellipse rather than resting on the surface." }, "broth": { "score": 0.5, "reason": "The liquid is a flat orange radial gradient with no sheen, texture or depth; toppings and noodles sit on top of the surface with little believable immersion, occlusion or contact." }, "steam": { "score": 0.6, "reason": "Steam is clearly animated across the desktop frames (desktop-1 faint wisp, desktop-2/3 rising and spreading plume, pointer-left dispersed wisps), but it renders as blurry, fairly opaque fog blobs with a bright hotspot rather than translucent irregular wisps." }, "composition": { "score": 0.75, "reason": "Balanced dark atmospheric scene with the dish as a clear focal point, warm rim light, chopsticks leading to the upper right, and no text or clutter; the levitating bowl slightly undermines grounding." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, nori, tofu, leaves, scallion, corn and chili-like spheres in plant broth \u2014 no animal-derived toppings appear." } }, { "noodles": { "score": 0.45, "reason": "Noodles are rendered as parallel flat wavy ribbons lying in uniform rows on the broth surface; consistent thickness but no volume, overlap or coherent strand placement \u2014 repeated flat decorative strokes dominate." }, "toppings": { "score": 0.55, "reason": "Nori sheet, mushroom caps with stems, green leaves, scallion rings and corn are recognizable; however the tofu is a set of flat white rectangular cards pasted at odd angles, and glossy orange/yellow spheres read as plastic balls rather than food." }, "bowl": { "score": 0.6, "reason": "Coherent low-angle ellipse rim with highlight, dark outer wall crescent and a ground shadow; but the vessel reads as an extremely shallow pan rather than a ramen bowl, and the chopsticks terminate abruptly at the rim without believable contact." }, "broth": { "score": 0.5, "reason": "Broth is a flat orange radial-gradient disc with no sheen, texture or depth; toppings and noodle strokes look pasted on the surface with minimal occlusion or immersion shading." }, "steam": { "score": 0.65, "reason": "Steam clearly animates across desktop frames \u2014 the plume rises, drifts and spreads from a compact cloud to a diffuse haze \u2014 but its bright hot core reads as glowing fog more than translucent wisps." }, "composition": { "score": 0.75, "reason": "Centered dish with steam above, shadow below and chopsticks as an asymmetric accent on a restrained dark starfield; readable focal point with no clutter, though the mobile crop leaves the bowl small with much empty space." }, "vegan": { "score": 0.75, "reason": "Nori, mushrooms, greens, tofu, corn and scallion are plausibly vegan; the glossy orange spheres are ambiguous (could read as fish balls), so an intermediate score applies." } }, { "noodles": { "score": 0.6, "reason": "Continuous wavy yellow-orange strands span the surface with plausible bends, but they are flat uniform lines laid in mostly parallel horizontal bands with no volume, layering, or immersion depth \u2014 stylized yet mechanically repetitive." }, "toppings": { "score": 0.7, "reason": "Good variety: pale mushroom caps with stems, a textured dark nori sheet, green herb leaves, white tofu squares, yellow corn dots, scallion rings, and sesame specks all read recognizably. The plain glossy red-orange spheres and slightly pasted-on look of the tofu squares keep it below polished." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim with a highlight arc, dark navy lower body, and chopsticks resting across the right edge. The bowl visibly floats above its detached ground shadow with a gap, and the top-view interior with minimal side wall reads slightly hybrid, though this seems a deliberate stylization." }, "broth": { "score": 0.7, "reason": "Warm orange radial-gradient surface with a soft sheen gives believable liquid depth; toppings generally sit on the surface with small contact shadows, though the broth is essentially a flat painted disc and immersion cues are minimal." }, "steam": { "score": 0.45, "reason": "Steam animates across frames (glow shifts, rises and fades from desktop-1 to pointer-right, becoming gray wisps in pointer-left), so it is not static, but it reads as an opaque bright fog/light blob floating detached well above the soup rather than translucent wisps emerging from the surface." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric scene: centered focal bowl, chopsticks anchoring the right, warm amber palette, subtle dust particles and ground shadow, no distracting text or clutter. The detached steam glow slightly competes with the dish." }, "vegan": { "score": 1.0, "reason": "All visible toppings \u2014 nori, mushrooms, tofu, corn, scallion, herbs, tomato/chili spheres, sesame \u2014 are plausibly plant-based; no animal-derived ingredients appear." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9388888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "757fa53665265fb28ed33303d605953dd91da4cef5a27700ef6c47d168ec6c0f", "gradedAt": "2026-09-30T23:25:44.778270+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5688888888888889, "toppings": 0.5211111111111111, "bowl": 0.5355555555555556, "broth": 0.5933333333333334, "steam": 0.5766666666666667, "composition": 0.7177777777777777 }, "reward": 0.5687777777777778, "runId": "meta/muse-spark-1.3/medium", "rewardStdDev": 0.0868149905983557, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.59, "toppings": 0.48, "bowl": 0.5, "broth": 0.53, "steam": 0.43, "composition": 0.7366666666666667 }, "reward": 0.5291666666666667, "sampleStdDev": 0.0030550504633038962, "judgments": [ { "noodles": { "score": 0.58, "reason": "Continuous golden strands have smooth bends and highlights, but mostly form parallel shallow waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.48, "reason": "Leaves and scallion rings are readable, with a varied plant-based assortment. However, mushrooms resemble smooth brown disks, tofu lacks texture, corn is circular beads, and nori looks like a thick rounded card." }, "bowl": { "score": 0.5, "reason": "The elliptical rim and shaded curved walls convey depth and dark lacquer, but the foot is visibly separated from the body, making the construction look unsupported." }, "broth": { "score": 0.53, "reason": "Warm amber color and pale surface highlights suggest soup, but broad flat color bands and minimal contact effects make ingredients appear arranged on top rather than immersed." }, "steam": { "score": 0.43, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the food. They read mostly as disconnected blurred blobs rather than irregular wisps emerging from the liquid." }, "composition": { "score": 0.75, "reason": "A centered dish, restrained warm-versus-cool palette, and uncluttered dark background provide a clear focal point. The mobile dish is quite small amid extensive empty space and disproportionately large steam." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly represent tofu, mushrooms, leafy vegetables, corn, scallions, and seaweed; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.61, "reason": "Continuous pale strands have soft bends and highlights, but form a sparse, mostly parallel horizontal layer with little tangled volume or convincing submersion." }, "toppings": { "score": 0.48, "reason": "Leaves and scallion rings are readable, with varied colors, but mushrooms resemble smooth brown disks, corn resembles round beads, tofu lacks texture, and nori is a thick rounded rectangular card." }, "bowl": { "score": 0.5, "reason": "The elliptical rim and shaded curved walls establish depth, but the outlined foot is visibly separated from the body, making the construction appear detached." }, "broth": { "score": 0.53, "reason": "Warm amber color and broad highlights suggest soup, but the surface is largely flat and opaque, with ingredients appearing layered above it rather than immersed." }, "steam": { "score": 0.43, "reason": "Normal frames show changing positions and shapes of translucent vapor without obscuring food, but it consists mainly of disconnected blurry blobs high above the soup rather than emerging wisps." }, "composition": { "score": 0.73, "reason": "A clear central dish, restrained dark palette, and warm illumination create an uncluttered focal scene. Excess empty space and a particularly small mobile dish weaken the proportions." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn, scallions, and seaweed; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.58, "reason": "Continuous pale strands have rounded bends and highlights, but mostly form parallel shallow waves with little tangled volume or convincing submerged overlap." }, "toppings": { "score": 0.48, "reason": "Leaves and scallion rings are readable, with useful color variety. Mushrooms resemble smooth brown disks, corn resembles round beads, tofu lacks texture, and nori looks like a thick rounded card." }, "bowl": { "score": 0.5, "reason": "The elliptical rim and shaded curved walls establish a recognizable dark bowl, but the foot is visibly detached beneath the body, weakening construction and support." }, "broth": { "score": 0.53, "reason": "Warm amber color and small highlights suggest appetizing soup, but the broad flat surface and layered central patches provide limited liquid depth; toppings largely appear placed above it." }, "steam": { "score": 0.43, "reason": "Normal frames show changing translucent vapor positions without obscuring the food. However, the vapor consists mainly of disconnected blurry blobs high above the bowl rather than irregular wisps visibly emerging from the soup." }, "composition": { "score": 0.73, "reason": "The centered dish, restrained warm-and-cool palette, and uncluttered dark background create a clear focal scene. Excessive empty space and a small mobile dish weaken presentation, while mobile vapor is disproportionately prominent." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn, scallions, and seaweed with noodles; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5, "bowl": 0.4066666666666667, "broth": 0.55, "steam": 0.5833333333333334, "composition": 0.6 }, "reward": 0.5088333333333334, "sampleStdDev": 0.0020207259421636597, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are soft, cream-colored wavy strands with some overlap and highlights, but they are small, cramped in a thin band and flat, with no clear individual strand volume or lifted noodles." }, "toppings": { "score": 0.5, "reason": "Toppings are recognizable: bok choy leaves, corn kernels, tofu, shiitake, scallion rings, nori. The tofu is a plain rounded block, the nori is a rigid dark card standing upright, and the corn is glossy spheres, so the generic forms limit the score." }, "bowl": { "score": 0.42, "reason": "The bowl has a rim, dark glossy walls and highlights, but the foot is detached and floats below the bowl with a visible gap, and the inner-wall curve looks odd. The bowl reads as two separate parts." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and some sheen, with ingredients plausibly sitting in it. The surface is fairly flat and smooth and the noodles and toppings look slightly pasted on." }, "steam": { "score": 0.55, "reason": "Soft translucent grey plumes rise and shift between frames, varying in shape and position. They look blurry and blobby, somewhat disconnected from the bowl, and drift high above it." }, "composition": { "score": 0.6, "reason": "A centered bowl on a dark atmospheric background with warm glow and sparse particles gives a clear focal point. The bowl is small in the frame, and the floating foot and rigid nori weaken the overall polish." }, "vegan": { "score": 0.93, "reason": "Visible toppings are tofu, mushrooms, corn, greens, scallions and nori, with no meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Noodles are pale wavy strands with soft highlights and some overlap, but they are small, flat and cluster into a thin band; they don't read as supple or voluminous." }, "toppings": { "score": 0.5, "reason": "Toppings are recognizable: bok choy leaves, corn kernels, shiitake caps, scallion rings and a nori sheet. However, the tofu is a plain white block, the nori is a rigid dark rounded card, and the corn looks like glossy spheres. Overall they are generic and symbol-like." }, "bowl": { "score": 0.4, "reason": "The bowl has a believable elliptical rim, but the body is a dark shell with an odd inner arc. The foot is a detached trapezoid floating below the bowl with a visible gap, which breaks the construction and support." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and a light sheen, which reads as appetizing. It is fairly flat, with little depth, and the toppings sit on top more than in it." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and change shape across frames, rising and fading. They are blurry and blob-like, sit high above the bowl and are a bit disconnected from the surface." }, "composition": { "score": 0.6, "reason": "The bowl is centered in a dark scene with a warm glow and subtle particles, so the focus is clear. The bowl is small in the frame, and the floating foot and the big empty space weaken the presentation." }, "vegan": { "score": 0.9, "reason": "Visible toppings are vegetables, mushrooms, tofu, corn and nori. No meat, fish or egg is visible." } }, { "noodles": { "score": 0.5, "reason": "Pale wavy strands are visible with some overlap and sheen, but they are thin, short and clustered in a small patch. They read as flat painted lines, not voluminous supple noodles." }, "toppings": { "score": 0.5, "reason": "Bok choy leaves, corn kernels, shiitake caps, tofu and scallion rings are all recognizable. The nori is a rigid dark rounded card and the tofu is a plain block. The corn is glossy and spherical, so the set looks like icons." }, "bowl": { "score": 0.4, "reason": "The bowl has a rim and a dark glossy body, but the foot ring floats detached below the bowl with a visible gap. That is a conspicuous construction flaw. The inner wall reads oddly as a double arc." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a gradient and some highlights and looks appetizing. It is fairly flat, and the toppings sit on top of it more than in it." }, "steam": { "score": 0.6, "reason": "Soft translucent blurred plumes change shape and position across frames and drift upward. They are somewhat blobby and large, and they sit far above the bowl and are disconnected from it." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark atmospheric background with a warm glow and subtle particles, and there is no clutter. The bowl is small in the frame and the floating base looks off." }, "vegan": { "score": 0.9, "reason": "Only vegetables, mushrooms, tofu, corn, nori and noodles are visible. No animal products appear." } } ], "diagnostics": { "vegan": 0.91 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6166666666666667, "toppings": 0.5833333333333334, "bowl": 0.7000000000000001, "broth": 0.7, "steam": 0.7166666666666667, "composition": 0.8166666666666667 }, "reward": 0.6683333333333333, "sampleStdDev": 0.06802266779047503, "judgments": [ { "noodles": { "score": 0.7, "reason": "Cream-colored wavy strands curve continuously across the broth with plausible thickness and partial submersion under toppings; slightly flat and repetitive painted-wave look with indistinct strand ends keeps it from higher." }, "toppings": { "score": 0.65, "reason": "Good variety: vein-detailed leafy greens, scallion rings, tofu block, corn kernels, brown mushroom domes and a tilted nori sheet. Corn reads as glossy uniform spheres and the nori is a rigid rounded card, which are somewhat generic forms; leaves and scallions are the strongest elements." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim with orange glaze band, deep hemispherical dark body with lacquer-like highlight streaks, and a small foot with contact shadow. Foot is small and fades into darkness, slightly weakening the support read." }, "broth": { "score": 0.75, "reason": "Warm orange surface with a creamy swirl, chili-oil specks near the rim, and edge shading for depth; toppings and noodles sit within the liquid with subtle contact shading rather than floating above it." }, "steam": { "score": 0.7, "reason": "Across the time-ordered desktop frames the plume configuration clearly changes, rising, spreading and drifting (leaning right in pointer-right), as translucent blurred wisps that never hide the food; puff clusters read slightly blobby and disconnected in places." }, "composition": { "score": 0.8, "reason": "Centered bowl as clear focal point against a dark starfield with a warm under-glow and reflected floor sheen; restrained palette, no text or clutter. Mobile framing keeps the dish readable though it shrinks considerably." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, corn, mushrooms, leafy greens, scallion rings, nori and noodles in plant-looking broth; no meat, fish, egg or dairy forms are apparent." } }, { "noodles": { "score": 0.55, "reason": "Noodles are continuous pale wavy strands swirled across the broth, but they read as flat ribbons/strokes with little volume, overlap or submersion depth; coherent yet noticeably stylized and artificial." }, "toppings": { "score": 0.5, "reason": "Leafy greens and scallion rings read well, but several items are generic geometry: the nori is a thick rigid rounded card standing upright, the mushrooms are glossy flat discs, the tofu is a plain block, and the corn is a cluster of simple circles." }, "bowl": { "score": 0.55, "reason": "Rim ellipse, dark ceramic walls and side highlight are coherent, but the foot is a detached trapezoid slab floating below the bowl with a visible gap, an unsupported-looking join that undermines construction." }, "broth": { "score": 0.6, "reason": "Orange surface with creamy swirls, specks and a warm gradient looks appetizing, though it is fairly flat; ingredients rest on the surface with limited occlusion or partial immersion." }, "steam": { "score": 0.7, "reason": "Frames at 1006/2930/4555 ms and pointer frames show translucent irregular puffs changing shape, rising and dissipating above the bowl without hiding food; slightly blobby and disconnected rather than continuous wisps." }, "composition": { "score": 0.8, "reason": "Centered dish on a dark starfield with a warm under-glow, steam adding depth, no clutter or distractions; food stays the focal point. Slightly sparse and the bowl reads small relative to the frame." }, "vegan": { "score": 0.9, "reason": "Visible toppings are leaves, tofu block, corn, mushroom caps, scallion rings and nori over plant broth; no animal-derived items, only mild ambiguity in the brown discs which read as mushrooms." } }, { "noodles": { "score": 0.6, "reason": "Cream wavy bands with faint strand lines read as noodles seen from above, but they sit as flat surface ribbons with little volume, strand separation or believable immersion; somewhat repetitive loop shapes." }, "toppings": { "score": 0.6, "reason": "Good variety: leafy greens with veins, scallion rings, brown shiitake caps, tofu block, corn and nori. Defects: corn is a cluster of glossy yellow balls, tofu is a plain rectangle, and the nori reads as a thick rigid rounded card leaning on the rim." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic/lacquer bowl: elliptical rim ring, rounded body with highlights and warm reflected glow, consistent perspective, attached foot. Minor shallow depth relative to rim width." }, "broth": { "score": 0.75, "reason": "Appetizing amber broth with gradient depth, creamy swirls, oil speckles and a chili-oil fleck; toppings sit on the surface with plausible contact and shadows, though the swirls look slightly painted." }, "steam": { "score": 0.75, "reason": "Translucent grey wisps clearly rise, spread and fade across the time-ordered frames (compact at 1s, diffuse and drifted right by 4.5s and later), without hiding the food; the plume is a bit blobby and starts slightly detached above the rim." }, "composition": { "score": 0.85, "reason": "Balanced centered dish in a dark atmospheric scene with warm glow, subtle particles and no distracting UI or clutter; the bowl remains the clear focal point on desktop and mobile." }, "vegan": { "score": 1.0, "reason": "Visible ingredients (greens, tofu, corn, mushrooms, scallion rings, nori, amber broth) are all plausibly plant-based; no meat, egg, fish or dairy shapes appear." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9588888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 158.34999999999945, "frameIntervalP95Ms": 183.29999999999927, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "b18b4fe935ae11f4ae2f6b670b20180906e6dba5f97b0a8af83fac14dffc0389", "gradedAt": "2026-09-30T23:19:37.335886+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.35000000000000003, "toppings": 0.5122222222222222, "bowl": 0.5255555555555556, "broth": 0.3788888888888889, "steam": 0.5800000000000001, "composition": 0.6855555555555556 }, "reward": 0.4966111111111111, "runId": "openai/gpt-5.5/high", "rewardStdDev": 0.07392945836954858, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.33333333333333337, "toppings": 0.5466666666666667, "bowl": 0.46, "broth": 0.37, "steam": 0.5733333333333334, "composition": 0.7233333333333333 }, "reward": 0.4906666666666667, "sampleStdDev": 0.009928914005737652, "judgments": [ { "noodles": { "score": 0.32, "reason": "Golden strands are readable, but dominated by flat angular zigzags, abrupt corners and cut ends rather than supple bends or convincing overlapping volume." }, "toppings": { "score": 0.53, "reason": "Green leafy stalks and scored mushroom caps are recognizable, with some shading and variety. Leaves resemble a rigid fan, mushrooms are flattened ovals, and scattered garnish floats outside the bowl." }, "bowl": { "score": 0.46, "reason": "The dark curved body and contact shadow suggest weight, but the rim ellipse and much taller soup ellipse use incompatible perspectives, making the contents look perched above the opening." }, "broth": { "score": 0.35, "reason": "Warm brown color and faint surface lines suggest soup, but the surface looks like a solid raised oval with little liquid sheen, depth or believable ingredient immersion." }, "steam": { "score": 0.56, "reason": "Faint translucent columns change curvature across normal frames and leave the food visible. They remain excessively parallel and elongated, with little visible spreading or irregular dispersal." }, "composition": { "score": 0.72, "reason": "A single centered dish, restrained dark background and warm-green palette create a clear focal point without text or UI. Floating garnish distracts, and the dish is relatively small in the mobile layout." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms and garnish plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.34, "reason": "Golden strands are readable, but mostly form flat, sharply angular zigzags with blunt ends rather than supple overlapping noodles." }, "toppings": { "score": 0.56, "reason": "Layered green leaves and scored mushroom caps are recognizable. Broad smooth surfaces, a blocklike stem, sticklike garnish, and seeds floating outside the bowl reduce organic plausibility." }, "bowl": { "score": 0.46, "reason": "The dark curved body has useful shading and a grounding shadow, but the tall soup ellipse extends well above the much flatter rim, creating incompatible cavity and rim perspectives." }, "broth": { "score": 0.38, "reason": "Warm brown coloration and subtle streaks suggest broth, but the surface reads as a raised solid oval with little liquid sheen or convincing ingredient immersion." }, "steam": { "score": 0.59, "reason": "Faint translucent trails change curvature across the normal frames and leave the food visible. They remain unusually straight, parallel columns with limited visible spreading and fading." }, "composition": { "score": 0.72, "reason": "A single centered dish, warm green-and-gold accents, and uncluttered dark surroundings establish a clear focus. The mobile dish is relatively small, and the tall steam columns dominate the upper space." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms, and small plantlike garnishes plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.34, "reason": "Golden strands are visible around the toppings, but mostly form flat, sharply angular zigzags with little supple curvature, overlap, or submerged volume." }, "toppings": { "score": 0.55, "reason": "Layered green leaves and scored mushroom caps are recognizable, with some shading and readable detail. The oversized central stalk, flat mushroom ovals, stick-like greens, and scattered floating seed shapes remain conspicuously schematic." }, "bowl": { "score": 0.46, "reason": "The dark curved body has convincing shading and a grounding shadow, but the rear rim ellipse conflicts with the much taller food-surface ellipse, making the contents appear perched above rather than contained within the opening." }, "broth": { "score": 0.38, "reason": "Warm brown coloring and subtle surface streaks suggest soup, but the surface reads as a solid mound with nested oval regions. Ingredients lack convincing wet contact, immersion, and liquid sheen." }, "steam": { "score": 0.57, "reason": "Normal frames show changes in the faint translucent trails above the soup, without obscuring the food. However, the trails are predominantly long parallel vertical ribbons with limited visible spreading or irregular dissipation." }, "composition": { "score": 0.73, "reason": "The single centered dish, warm green-and-amber palette, dark uncluttered backdrop, and restrained lighting establish a clear focal point. The mobile dish is relatively small within extensive empty space, and floating garnish marks weaken the otherwise orderly presentation." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms, and seed-like garnishes plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.31666666666666665, "toppings": 0.4066666666666667, "bowl": 0.48333333333333334, "broth": 0.35, "steam": 0.48333333333333334, "composition": 0.5666666666666667 }, "reward": 0.42583333333333334, "sampleStdDev": 0.01376892636821524, "judgments": [ { "noodles": { "score": 0.35, "reason": "Noodles are rigid, angular zigzag ribbons with sharp bends, like wire scribbles, scattered on the surface. They lack supple continuous strands and volume." }, "toppings": { "score": 0.4, "reason": "A bok choy is recognizable, and the mushrooms show gill lines, but they are flat, pale ovals. Green sticks look like rigid scallion bits, there is a small tofu cube, and the floating cream ovals are odd. The bok choy has a glowing, plastic look." }, "bowl": { "score": 0.5, "reason": "A dark bowl with handles and a rim is readable, but the rim and handle perspective is awkward. The broth disc looks like a lid sitting above the bowl cavity, and the ovals floating outside the bowl look detached." }, "broth": { "score": 0.35, "reason": "Broth reads as a flat brown wood-grain-like disc, not a liquid. There is little immersion, and the toppings look pasted on top. There is a warm glow but no real sheen or depth." }, "steam": { "score": 0.5, "reason": "Faint vertical wisps rise and change between frames, so there is motion. They are thin, uniform, straight streaks that look mechanical, but they do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark, warm background, and the glow focuses attention on it. The palette is cohesive, but the stray floating ovals are distracting and the food details are weak." }, "vegan": { "score": 0.85, "reason": "Only noodles, bok choy, mushrooms, scallions and a tofu-like cube are visible, with no animal products. The pale ovals are ambiguous but read as sesame or seeds." } }, { "noodles": { "score": 0.3, "reason": "Noodles are rigid, angular zigzag ribbons with flat, sharp bends that read as wire or sticks rather than supple strands. They lie flat on the surface and show no volume or overlap, and they shift shape between frames." }, "toppings": { "score": 0.42, "reason": "A bok choy with a green leaf and pale stem is recognizable, and mushroom caps with gill lines are readable. The mushrooms are flat tan ovals and the tofu cube is barely visible. Green scallion sticks and floating cream ovals look like scattered generic bits. The bok choy has a glowing, artificial look." }, "bowl": { "score": 0.45, "reason": "The dark bowl has a rim, handles and a shadow, but the perspective is inconsistent. The rim ellipse is much wider than the body and the handles look detached. The bowl reads more like a wok or pot. Cream ovals float outside it." }, "broth": { "score": 0.35, "reason": "The broth is a flat, wood-grain-like brown-orange disc with thin lines. It has little depth or sheen, and the toppings look pasted on top rather than immersed. The orange glow is appealing." }, "steam": { "score": 0.45, "reason": "Faint vertical streaks rise and change between frames, so steam is present and translucent. The streaks are rigid, straight and uniform, like mechanical plumes, and they start well above the bowl rather than from the soup." }, "composition": { "score": 0.55, "reason": "The bowl is centered on a dark, moody background with a warm glow and clean framing. The food is the focus, but the scattered floating ovals and the glowing bok choy distract." }, "vegan": { "score": 0.85, "reason": "The toppings are bok choy, mushrooms, scallion, noodles and possibly tofu. No animal ingredients are visible." } }, { "noodles": { "score": 0.3, "reason": "Noodles are rigid, flat angular zigzag ribbons with sharp bends that look like wire or sticks. They are scattered over the surface rather than forming supple, overlapping strands, and they shift shape between frames." }, "toppings": { "score": 0.4, "reason": "The bok choy is recognizable, and the shiitake-like caps have gill lines. However, they are flat, vague ellipses. The green scallion sticks are stiff, and the small beige ovals scattered outside the bowl look like random confetti. The tofu cube is barely visible." }, "bowl": { "score": 0.5, "reason": "The dark bowl has a coherent rim, handles and a curved body with a shadow. The body is very dark and flat, and the rim perspective is odd: the broth disc appears to sit above the rim. Floating ovals sit around the bowl." }, "broth": { "score": 0.35, "reason": "The broth is an orange-brown flat disc with faint wood-grain lines, so it reads more like a wooden surface than liquid. Immersion and contact with the ingredients are weak, and the toppings look pasted on top." }, "steam": { "score": 0.5, "reason": "Faint vertical wisps rise and change between frames, and they do not obscure the food. They are thin, streaky and mechanical, and they start well above the bowl." }, "composition": { "score": 0.55, "reason": "The bowl is centered on a dark, warm, atmospheric background with a clear focal point and a pleasing glow. The scattered ovals outside the bowl and the overly bright bok choy glow are distracting." }, "vegan": { "score": 0.85, "reason": "Visible items are noodles, bok choy, mushrooms, scallions and possibly tofu. No meat, egg or fish is visible." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.4, "toppings": 0.5833333333333334, "bowl": 0.6333333333333333, "broth": 0.4166666666666667, "steam": 0.6833333333333333, "composition": 0.7666666666666667 }, "reward": 0.5733333333333334, "sampleStdDev": 0.025289984842489205, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles read as flat, cream-colored zigzag polylines with sharp angular bends lying on the surface; rigid broken ribbon segments dominate rather than supple continuous strands, and they lack believable immersion in the broth." }, "toppings": { "score": 0.6, "reason": "Recognizable variety: a large stylized bok choy leaf with pale stalk, two mushroom slices with gill strokes, scallion rods and a small tofu-like cube. Defects: scallions are rigid straight sticks and many tan seed ellipses float detached in mid-air around the bowl exterior." }, "bowl": { "score": 0.6, "reason": "Coherent dark ceramic body with rim ellipse, curved highlight and grounded shadow, but the flat teal ring handles extend outward like detached Saturn rings rather than attached ears, an odd join that undermines the construction." }, "broth": { "score": 0.3, "reason": "The interior is a solid brown disc with wood-grain-like concentric arcs and no sheen, depth or liquid quality; toppings and noodles sit pasted flat on top rather than sitting in soup." }, "steam": { "score": 0.65, "reason": "Translucent gray vertical wisps above the bowl that change shape and extent across the normal desktop frames (shorter in desktop-1, taller in desktop-2/3, shifting in pointer frames), showing continuous animated motion; slightly stringy/stiff in form but irregular and never hiding the food." }, "composition": { "score": 0.8, "reason": "Clean, balanced center-weighted scene with warm rim glow, readable focal dish, dark atmospheric backdrop and no text or clutter; floating seed particles are a minor distraction but the presentation stays coherent." }, "vegan": { "score": 0.95, "reason": "Visible ingredients are plant-based: bok choy, mushrooms, scallions, sesame seeds, noodles and a tofu-like cube; no meat, fish or egg is discernible, with only the small brown cube mildly ambiguous but most plausibly fried tofu." } }, { "noodles": { "score": 0.4, "reason": "Yellow strands are rigid flat zigzag ribbons with sharp repeated angular bends lying on the surface, not supple nested strands with volume; the angular stick-like segments dominate the broth." }, "toppings": { "score": 0.55, "reason": "Bok choy is well shaped with leaf blade and pale stalk; mushroom slices and scallion strips are readable but very flat, and many cream sprout ellipses float in mid-air outside the bowl rather than sitting in the soup." }, "bowl": { "score": 0.6, "reason": "Coherent dark ceramic hemisphere with soft shadow and side handle rings, but the broth ellipse meets the silhouette with no visible rim wall thickness and the thin gray handle arcs read ambiguously against the dark body." }, "broth": { "score": 0.55, "reason": "Restrained sheen and subtle swirl texture with a darker shadow region aiding integration, but the surface is largely a flat solid orange disc with toppings pasted on top and little immersion occlusion at their bases." }, "steam": { "score": 0.65, "reason": "Translucent wavy wisps that clearly change shape and position across desktop-1/2/3 frames, fading upward without hiding food, though the parallel vertical strands are somewhat mechanical and start slightly detached above the rim." }, "composition": { "score": 0.7, "reason": "Centered dish with warm radial glow, dark atmospheric palette and no text or clutter; slightly weakened by garnish debris floating around the bowl and the gap between bowl and steam." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are bok choy, mushroom slices, scallions, sprout/garlic ellipses, a tofu-like block and noodles in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.4, "reason": "Noodles read as flat yellow zigzag ribbons with sharp angular bends and detached segments rather than supple continuous strands; rigid ribbon character dominates the surface." }, "toppings": { "score": 0.6, "reason": "Bok choy with white stalk and two gilled mushroom slices are recognizable with decent detail; scallions are stiff stick forms, and pale sesame ovals float impossibly around and outside the bowl rim." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, dark rounded body, handle arcs and a grounding shadow; rim/body perspective is consistent, though the very dark body hides wall detail and the handle ring reads slightly ambiguously." }, "broth": { "score": 0.4, "reason": "Surface reads as a solid glowing reddish plate with a strange darker elliptical ring rather than liquid; no sheen or depth, and toppings appear pasted on top without believable immersion or occlusion." }, "steam": { "score": 0.75, "reason": "Faint translucent wisps rise above the bowl and change shape and position across the timed desktop and pointer frames, fading upward without obscuring the food; strand-like plumes are slightly mechanical but clearly animate." }, "composition": { "score": 0.8, "reason": "Centered bowl on a warm-glow dark background with clear focal focus, balanced proportions and no text or clutter; stray floating seed ovals add minor visual noise." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are bok choy, mushrooms, scallions, sesame seeds, noodles and plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9444444444444444, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 83.39999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "be599a71add6dddcb6427b94673060d335cdb3576418d7ad9b4340cddf6f13f3", "gradedAt": "2026-09-30T23:20:10.666689+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.45666666666666667, "toppings": 0.5955555555555555, "bowl": 0.7544444444444445, "broth": 0.5744444444444444, "steam": 0.658888888888889, "composition": 0.7388888888888889 }, "reward": 0.6212777777777778, "runId": "openai/gpt-5.5/low", "rewardStdDev": 0.07789421839430227, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.37, "toppings": 0.5866666666666667, "bowl": 0.7933333333333333, "broth": 0.49, "steam": 0.6266666666666667, "composition": 0.75 }, "reward": 0.5973333333333334, "sampleStdDev": 0.018009256878986815, "judgments": [ { "noodles": { "score": 0.35, "reason": "Readable golden strands have consistent thickness, but mostly form straight, angular zigzags in a flat stacked mat rather than supple curls with convincing volume." }, "toppings": { "score": 0.58, "reason": "Green leaves with central veins, mushroom caps with readable gills, and corn provide recognizable variety. Shapes remain flat and repetitive, with little organic texture or dimensional overlap." }, "bowl": { "score": 0.78, "reason": "The continuous elliptical rim, curved dark walls, restrained highlights and contact shadow create coherent supported geometry. Broad angular shading and a minimally defined base limit the material finish." }, "broth": { "score": 0.47, "reason": "Warm orange gradients suggest broth, but the surface looks largely solid and graphic. Ingredients appear laid over it, with little visible immersion, liquid contact or surface sheen." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent trails rising from the soup and fading without obscuring food. The trails are predominantly narrow, angular ribbons with limited spreading or soft curling." }, "composition": { "score": 0.74, "reason": "A single unobstructed dish stands out against a restrained dark brown background, with complementary warm broth and green toppings. The low placement and large empty upper area weaken balance, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms, corn and scattered green garnish plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.4, "reason": "Distinct yellow strands have consistent thickness, but predominantly straight, angular runs form a flat crisscrossed mat rather than supple noodles with convincing volume." }, "toppings": { "score": 0.6, "reason": "Greens have readable stems and leaf veins, mushrooms show contrasting caps and gills, and corn adds variety. Repeated flat leaf shapes and uniformly circular kernels remain schematic with little organic texture." }, "bowl": { "score": 0.82, "reason": "Continuous elliptical rim, aligned curved walls, restrained dark highlights, and a grounded contact shadow create coherent depth and support. The broad gray rim and faceted wall shading slightly weaken the ceramic finish." }, "broth": { "score": 0.52, "reason": "Warm orange shading suggests soup depth, with faint submerged strand marks. The surface lacks convincing liquid sheen or ingredient contact, leaving the food looking layered over a largely flat fill." }, "steam": { "score": 0.64, "reason": "Normal frames show changing translucent trails rising above the soup without obscuring it. The trails are unusually narrow, angular, and parallel, reading more as drifting line strokes than spreading vapor." }, "composition": { "score": 0.76, "reason": "A single unobstructed dish remains the focus against a restrained dark background, with coordinated warm soup and green accents. Low placement and extensive empty upper space, particularly on mobile, reduce balance." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, corn, herb pieces, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.36, "reason": "Distinct yellow strands are visible, but nearly straight zigzags form a flat, wire-like lattice with abrupt bends and little supple volume." }, "toppings": { "score": 0.58, "reason": "Greens have readable stems and central veins, and mushrooms have caps and gill marks. Repeated smooth leaf silhouettes and uniformly circular corn remain schematic and lack organic texture." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls and grounded shadow form a coherent vessel with readable depth. Broad angular shading and a subdued rim highlight make the ceramic material somewhat flat." }, "broth": { "score": 0.48, "reason": "Warm amber gradients distinguish the soup from the bowl, but the surface has little liquid sheen or contact detail; ingredients largely appear laid over a flat colored plane." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent strands rising above the soup without obscuring it. The wisps are unusually straight and narrow, with limited curling or spreading." }, "composition": { "score": 0.75, "reason": "A single readable dish, restrained warm lighting and uncluttered dark background create a clear focus. The low placement leaves excessive empty space above, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, gilled mushrooms and corn plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.43333333333333335, "toppings": 0.5, "bowl": 0.6866666666666666, "broth": 0.5666666666666667, "steam": 0.5833333333333334, "composition": 0.65 }, "reward": 0.5581666666666667, "sampleStdDev": 0.010515861036231582, "judgments": [ { "noodles": { "score": 0.4, "reason": "The noodles are thin, straight, parallel strands packed into a rigid, comb-like raft. They look flat and mechanical, with few natural bends and little volume, and they sit on top of the broth rather than in it." }, "toppings": { "score": 0.5, "reason": "The bok choy, mushroom slices and corn are recognizable, and the mushrooms have gill lines. The shapes are flat and symbolic, the scale is small, there is no nori or tofu, and the scattered green flecks look like confetti." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a dark glossy body with a soft highlight, and a contact shadow. The perspective is consistent. It is a little plain, and the rim looks thin and a bit disconnected from the interior." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a warm radial glow and reads as liquid, but it is flat and smooth. The noodles and toppings look pasted on the surface with little immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Thin, translucent wisps rise and change shape between frames, which gives believable ambient motion. They are streaky, fairly uniform and somewhat mechanical, and they are faint." }, "composition": { "score": 0.62, "reason": "The bowl is centered and is clearly the focal point against a dark, restrained background, with warm colors. The bowl sits a little low, there is a lot of empty space, and the food detail is modest." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are noodles, bok choy, mushrooms, corn and greens, and there is no meat, egg or fish." } }, { "noodles": { "score": 0.45, "reason": "Noodles are a flat rectangular mat of mostly straight parallel strands with rigid line ends. They read as noodles but look stiff and lack organic bends or volume." }, "toppings": { "score": 0.5, "reason": "Two bok choy leaves, mushroom slices with gills, a corn cob and scallion bits are recognizable and varied. They are small and flat, with simple vector shapes, and the corn is a grid of dots." }, "bowl": { "score": 0.7, "reason": "The dark bowl has a coherent elliptical rim, inner lip highlight, curved body and a grounded shadow. The matte material is plain but the perspective is consistent, with no obvious distortion." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a warm radial glow and depth that look appetizing. It is fairly flat and smooth, with little surface texture, and the noodles sit on top more than in it." }, "steam": { "score": 0.55, "reason": "Thin translucent wisps rise and change shape between frames, so the motion is real. They are streaky, hair-like and somewhat uniform, and appear to start above the rim rather than from the soup." }, "composition": { "score": 0.65, "reason": "A centered bowl with warm glow against a dark background gives a clear focal point with no clutter. The bowl sits somewhat low, the food is small, and the top of the frame is empty." }, "vegan": { "score": 0.92, "reason": "Only noodles, mushrooms, bok choy, corn and scallions are visible, with no meat or egg, so it looks vegan." } }, { "noodles": { "score": 0.45, "reason": "The noodles are thin, straight, parallel strands stacked in a rigid, flat raft. They look like rigid ribbons with few bends and little volume, and they sit on top of the broth rather than in it." }, "toppings": { "score": 0.5, "reason": "The bok choy leaves, sliced mushrooms with gills and corn cluster are recognizable, and the scattered scallions add variety. The shapes are flat and simple, and the corn is a grid of dots. There is no nori or tofu." }, "bowl": { "score": 0.68, "reason": "The dark bowl has a coherent elliptical rim, a believable wall curve, a base shadow and a subtle highlight. The rim is slightly thick, and the body is very dark with little material detail." }, "broth": { "score": 0.6, "reason": "The warm orange broth has a radial glow and looks appetizing. It is fairly flat and gradient-like, and the noodles float above it with little immersion or occlusion." }, "steam": { "score": 0.6, "reason": "The wisps are translucent, vary from frame to frame and rise above the bowl. They are streaky and somewhat uniform, and they are faint." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, atmospheric scene with warm lighting, and there is no clutter. The bowl is smallish, and the top half of the frame is empty." }, "vegan": { "score": 0.9, "reason": "The toppings are bok choy, mushrooms, corn and scallions, all plant-based. There are no visible animal products." } } ], "diagnostics": { "vegan": 0.9166666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.7, "bowl": 0.7833333333333333, "broth": 0.6666666666666666, "steam": 0.7666666666666667, "composition": 0.8166666666666667 }, "reward": 0.7083333333333334, "sampleStdDev": 0.03556098048891974, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles read clearly as a pale-yellow strand bundle with plausible thickness, but the strands are mostly straight, flat and wire-like with minimal supple bends, forming a rigid crisscross mat on the surface rather than a supple nest." }, "toppings": { "score": 0.75, "reason": "Recognizable stylized ingredients: mushroom slices with visible gills, two bok choy leaves with pale stems, a corn cluster and scattered scallion flecks. Organic shapes and readable details; variety is decent but limited (no nori/tofu) and shapes are quite simplified." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with an inner rim ring, smooth gradient walls, believable rounded base with contact shadow. Perspective of rim and body is consistent; slightly shallow depth and a uniform mechanical rim ring are the only shortcomings." }, "broth": { "score": 0.75, "reason": "Glowing orange-red liquid surface with subtle streak highlights and darker edge occlusion near the rim; toppings sit within the broth. The strong radial glow reads slightly more like lighting than liquid texture." }, "steam": { "score": 0.7, "reason": "Translucent thin wisps rise above the bowl and change shape/position across desktop-1, desktop-2, desktop-3 and pointer frames, indicating animated steam; reduced-motion frames show far fewer wisps, consistent with motion. Wisps are sparse, slightly strand/grass-like, and some start detached above the rim." }, "composition": { "score": 0.85, "reason": "Balanced centered bowl as the focal dish against a dark atmospheric background with restrained warm glow; no clutter, text or competing elements. Food remains the focus on both desktop and mobile crops." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are all plant-derived: bok choy, mushrooms, corn, scallions, noodles in broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.6, "reason": "Yellow strands are continuous with slight bends and coherent layering/shadows, reading as a noodle bed across the broth. However they are largely uniform, mostly horizontal, and lie flat with little volume or supple curvature, giving an artificial, ribbon-like feel." }, "toppings": { "score": 0.7, "reason": "Two bok choy-style greens with pale stems, two mushroom slices with readable gill hatching, a corn cluster, and scattered herb flecks give real variety and organic-ish shapes. Corn reads as a slightly mechanical dot grid, and leaf forms are simple flat vector shapes." }, "bowl": { "score": 0.8, "reason": "Dark ceramic-style bowl with a coherent elliptical rim, lighter inner ring, tapered body, subtle sheen highlight and soft ground shadow; perspective is consistent between rim, cavity and base. Only minor shortcoming is the rather simple, uniform rim treatment." }, "broth": { "score": 0.7, "reason": "Warm orange-red liquid surface with a central glow, faint darker swirl streaks and restrained sheen; ingredients sit on the surface with small stem contacts. It reads slightly flat/smooth, with the radial glow doing much of the work rather than rich liquid texture." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the translucent wisps clearly change shape, position and drift while rising from the broth, fading without hiding the food; reduced frames show it suppressed. Wisps are a bit thin/stringy rather than volumetric." }, "composition": { "score": 0.8, "reason": "Centered focal dish against a dark atmospheric background with warm backlight glow; balanced proportions, coherent lighting, no clutter or competing elements. The upper half is quite empty and the styling minimal, but presentation is clean and readable." }, "vegan": { "score": 1, "reason": "Visible toppings are greens, mushrooms, corn, herbs, noodles and plant-based broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.5, "reason": "Noodles read as a flat horizontal bundle of thin, mostly straight golden strands stretched across the bowl, resembling straw or wires more than supple strands; slight waviness and layered overlap help, but they lack believable bends, volume, and immersion." }, "toppings": { "score": 0.65, "reason": "Recognizable stylized bok choy with white stems, mushroom slices with gill detail, a corn kernel cluster, and scattered scallion pieces give decent variety; however shapes are flat cut-paper forms that sit atop the noodles rather than feeling anchored in the soup." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, dark inner wall ring, tapered body with highlight sheen, and a grounded base with contact shadow; perspective is consistent. Minor shortcomings: the vessel is very wide and shallow, reading more like a pan than a ramen bowl." }, "broth": { "score": 0.55, "reason": "Warm orange surface with a bright radial glow and faint streaks suggests liquid, and some subtle shadow shapes under ingredients hint at occlusion, but the glowing center reads somewhat like a light source, and toppings/noodles show little believable immersion or contact." }, "steam": { "score": 0.8, "reason": "Translucent brownish wisps rise above the bowl and clearly change shape, height, and position across the time-ordered normal and pointer frames, showing continuous rising motion without obscuring the food; reduced frames show appropriately faint static steam." }, "composition": { "score": 0.8, "reason": "Centered focal dish with a restrained dark atmosphere, warm backlight glow, balanced proportions, and no text, UI, or clutter; presentation is clean and readable, though the scene is somewhat sparse." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are leafy greens, mushrooms, corn, scallions, noodles, and plant-based broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9722222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 33.399999999999636, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "943ece6d456b7ac098fb2963d4037c811928129261cf0a9e3827e53f825c84b3", "gradedAt": "2026-09-30T23:20:09.167161+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.32, "toppings": 0.34555555555555556, "bowl": 0.6266666666666667, "broth": 0.4366666666666667, "steam": 0.5077777777777778, "composition": 0.6555555555555556 }, "reward": 0.46111111111111114, "runId": "openai/gpt-5.5/medium", "rewardStdDev": 0.05740531656885421, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.36, "toppings": 0.38666666666666666, "bowl": 0.73, "broth": 0.41000000000000003, "steam": 0.47333333333333333, "composition": 0.7166666666666667 }, "reward": 0.4983333333333333, "sampleStdDev": 0.011150485789118498, "judgments": [ { "noodles": { "score": 0.35, "reason": "Sparse yellow strands are visible, but mostly resemble stiff short wires with angular bends and little overlapping noodle volume." }, "toppings": { "score": 0.38, "reason": "Green leaves, brown mushroom-like pieces and small colorful garnishes provide variety, but smooth capsule shapes dominate; leaves lack organic edges and mushrooms lack readable cap-and-gill detail." }, "bowl": { "score": 0.72, "reason": "The elliptical rim joins a coherent rounded body with soft ceramic-like highlights. The unusually broad orange lip and nearly invisible base/contact make the vessel somewhat artificial and floating-looking." }, "broth": { "score": 0.4, "reason": "Warm orange shading suggests soup, but the surface is a largely featureless matte disc, with little liquid sheen or visible immersion around the centrally clustered ingredients." }, "steam": { "score": 0.48, "reason": "Translucent blurred plumes change position and distribution across normal frames without obscuring the food. They resemble disconnected soft blobs, including a persistent horizontal oval, rather than connected irregular wisps emerging from the soup." }, "composition": { "score": 0.73, "reason": "A single centered bowl remains readable against a restrained dark amber background, with no distracting text or UI. Extensive empty space and very small food details weaken the focal presentation, especially on mobile." }, "vegan": { "score": 0.9, "reason": "Visible green vegetables, mushroom-like pieces and noodles plausibly form vegan ramen; no clearly animal-derived ingredient is observed, though several tiny garnishes are indistinct." } }, { "noodles": { "score": 0.38, "reason": "Thin yellow strands are visible, but mostly form sparse straight sticks and shallow hooks rather than supple, overlapping noodles with volume." }, "toppings": { "score": 0.4, "reason": "Green leaves, brown mushroom-like pieces and small colorful garnishes provide variety, but smooth capsule shapes dominate and lack organic edges or readable food texture." }, "bowl": { "score": 0.74, "reason": "The elliptical rim and rounded dark body have coherent perspective and soft material shading. The unusually broad orange lip and poorly defined ground contact reduce ceramic plausibility." }, "broth": { "score": 0.43, "reason": "Warm orange shading suggests soup, but the surface resembles a smooth solid disc with little liquid sheen or texture; ingredients lack convincing immersion and contact." }, "steam": { "score": 0.48, "reason": "Translucent blurred forms change position across normal frames without obscuring the food. However, isolated oval clouds and vertical smudges lack connected, irregular rising wisps." }, "composition": { "score": 0.72, "reason": "A single centered bowl, restrained warm glow and uncluttered dark background establish a clear focus. Extensive empty space and the tiny central food cluster weaken visual balance." }, "vegan": { "score": 0.9, "reason": "Visible noodle, leaf and mushroom-like forms plausibly depict plant-based ramen, with no clearly animal-derived topping. Some tiny garnishes remain indistinct." } }, { "noodles": { "score": 0.35, "reason": "Thin yellow strands are visible, but sparse, mostly straight or angular strokes provide little supple curvature, overlap, or noodle volume." }, "toppings": { "score": 0.38, "reason": "Green leaves, brown mushroom-like pieces, and small colorful garnishes provide variety, but smooth capsule shapes dominate, with minimal organic texture or readable ingredient detail." }, "bowl": { "score": 0.73, "reason": "The elliptical rim and rounded dark body join coherently, with shading suggesting ceramic depth. The broad orange rim looks somewhat flat, and base contact is weakly established." }, "broth": { "score": 0.4, "reason": "Warm orange color clearly separates the soup from the rim, but the uniform gradient looks solid rather than liquid; ingredients lack convincing immersion, surface contact, or reflections." }, "steam": { "score": 0.46, "reason": "Translucent blurred forms change position across normal frames without hiding the food. However, isolated vertical smudges and a persistent horizontal oval read more as floating blobs than irregular rising wisps." }, "composition": { "score": 0.7, "reason": "A single centered bowl, restrained warm lighting, and uncluttered dark surroundings create a clear focal scene. Extensive empty space and the very small garnish cluster reduce visual balance and food readability." }, "vegan": { "score": 0.85, "reason": "Visible green vegetables, mushroom-like pieces, yellow noodles, and small plant-like garnishes plausibly form vegan ramen. No clearly animal-derived topping is visible, though several tiny pieces are ambiguous." } } ], "diagnostics": { "vegan": 0.8833333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.26666666666666666, "bowl": 0.5166666666666667, "broth": 0.45, "steam": 0.43333333333333335, "composition": 0.55 }, "reward": 0.395, "sampleStdDev": 0.006614378277661482, "judgments": [ { "noodles": { "score": 0.3, "reason": "A few thin yellow strands are scattered on the surface as flat, wiry lines. There is no noodle volume, no overlap and no sense of immersion in the broth." }, "toppings": { "score": 0.3, "reason": "The toppings are tiny capsule-like green, brown and pink shapes that read as generic pills or beads. They do not read as recognizable leaves, mushrooms or nori, and the detail is minimal." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent wide rim, a dark body with a soft highlight, and a rounded base. The rim is overly flat and wide and looks like a lid, and the body is murky, so the materials are not convincing." }, "broth": { "score": 0.45, "reason": "The orange-red broth has a smooth radial gradient and a subtle inner ring, which gives some depth. It looks flat and solid, and the toppings sit on top of it rather than in it." }, "steam": { "score": 0.45, "reason": "Soft translucent wisps drift and change position across frames. They look like blurry blobs plus a horizontal smear, and they are detached from the soup surface." }, "composition": { "score": 0.55, "reason": "The bowl sits centered in a dark, atmospheric scene with a warm glow and no distracting UI. The food is very small and the bowl is dim, so the focal dish lacks appeal." }, "vegan": { "score": 0.8, "reason": "Only vegetable-like and mushroom-like items and noodles are visible, with no obvious meat, egg or fish. The pink bits are slightly ambiguous." } }, { "noodles": { "score": 0.3, "reason": "Only a few thin yellow wire-like strands sit on the surface. They are sparse, rigid and scribble-like, with no volume or supple overlap, and they are not immersed in the broth." }, "toppings": { "score": 0.25, "reason": "Toppings are tiny capsule-like green, brown and pink-white shapes. They look like generic plastic pills rather than leaves, mushrooms or nori, and they have little readable detail." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent wide rim, a dark body and a base shadow, with consistent perspective. The rim is very broad and looks like a lid or plate, the body is flat dark, and the material cues are weak." }, "broth": { "score": 0.45, "reason": "The orange broth has a warm gradient and reads as liquid. It looks like a flat concentric disc with little texture, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.4, "reason": "Soft translucent wisps drift and change position across frames. They are blurry, blob-like and partly disconnected from the bowl, and a horizontal smear looks mechanical." }, "composition": { "score": 0.55, "reason": "The bowl is centered in a moody dark scene with a warm glow and no clutter. The food is tiny and unappealing within the bowl, and the overall scale is small." }, "vegan": { "score": 0.75, "reason": "Only vegetable-like, mushroom-like and noodle shapes are visible, with no obvious meat, egg or fish. The pink-white items are ambiguous and could read as fish cake." } }, { "noodles": { "score": 0.3, "reason": "Only a few thin yellow wire-like strands sit on the surface. They are small, sparse and scribble-like, with no volume or sense of a noodle mass in the broth." }, "toppings": { "score": 0.25, "reason": "The toppings are tiny capsule and pill shapes in green, brown and pink-white. They look like generic glossy blobs and read as neither leaves, mushrooms nor nori. Scale is very small and variety is not readable." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent wide rim, a dark body and a soft shadow, and the perspective is consistent. The rim is very flat and broad, the body looks like a dark pot, and the join is smooth but featureless with weak ceramic detail." }, "broth": { "score": 0.45, "reason": "The orange-red broth has a soft radial gradient and a warm glow. It reads as a flat disc with concentric rings, with no surface texture, and the toppings sit on top rather than in it." }, "steam": { "score": 0.45, "reason": "Soft translucent wisps rise and change position across frames, so there is real motion. They are blurry blobs and a flat horizontal smear, and they look detached from the bowl. They do not hide the food." }, "composition": { "score": 0.55, "reason": "The bowl is centered in a dark atmospheric scene with a warm glow and no clutter. The food is tiny and the focal interest is weak, and the large empty dark areas leave the image underwhelming." }, "vegan": { "score": 0.8, "reason": "No meat, egg or fish is visible. The toppings look plant-based, though the pink-white items are ambiguous." } } ], "diagnostics": { "vegan": 0.7833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.3, "toppings": 0.38333333333333336, "bowl": 0.6333333333333333, "broth": 0.45, "steam": 0.6166666666666667, "composition": 0.7 }, "reward": 0.49, "sampleStdDev": 0.07232738070744714, "judgments": [ { "noodles": { "score": 0.3, "reason": "Noodles are a handful of thin yellow squiggly strands lying flat on the broth surface. They read as sparse wire-like scribbles scattered on top rather than a coherent, supple noodle mass with volume or believable immersion." }, "toppings": { "score": 0.45, "reason": "A few tiny simplified pieces: green scallion-like slivers, brown flat slices suggesting mushrooms, yellow bits, and small pink-striped ovals. They are sparse, small relative to the bowl, and some read as generic glossy shapes rather than clearly readable ingredients." }, "bowl": { "score": 0.7, "reason": "Coherent wide-rimmed bowl: elliptical rim, smooth dark ceramic hemisphere body with soft sheen, consistent perspective and a contact shadow below. Defects: the rim is very shallow and saucer-like, and the rim's orange glaze blends confusingly with the broth color." }, "broth": { "score": 0.5, "reason": "The broth is a smooth glossy orange ellipse with a radial gradient. It looks somewhat solid/disc-like with no texture, ripples, or depth, and the noodles and toppings sit flattened on the surface rather than appearing submerged in liquid." }, "steam": { "score": 0.55, "reason": "Steam is animated: translucent brownish wisps change shape and position across the desktop frames, rising and dispersing without hiding the food. However, the plumes are discrete blob-like forms and a lens-shaped wisp repeats in nearly every frame, giving a flickering, semi-mechanical feel rather than continuous flow." }, "composition": { "score": 0.75, "reason": "Balanced centered dish in a dark atmospheric scene with a warm glow, coherent lighting, no clutter or competing elements. Minor shortcomings: the bowl occupies a small portion of the frame with large empty dark areas, and the food reads small at a distance." }, "vegan": { "score": 0.5, "reason": "Most toppings are plausibly plant-based (scallion-like greens, mushroom-like slices, corn-like bits), but the small pink-and-white striped ovals are ambiguous and could be read as narutomaki fish cakes; appearance is ambiguous, so an intermediate score applies." } }, { "noodles": { "score": 0.35, "reason": "Only a handful of thin yellow squiggle strands scattered on the surface; they read as wire-like curves with no noodle body, volume, or immersion in the broth, and no coherent overlapping mass." }, "toppings": { "score": 0.4, "reason": "Sparse sprinkling of simple shapes: flat green ellipses (scallion-like), smooth brown blobs (mushroom-ish), and small pink/white capsule pieces. Shapes are crude and scattered rather than composed, lacking texture and scale variety." }, "bowl": { "score": 0.7, "reason": "Coherent construction: elliptical terracotta rim, smooth dark hemispherical body with soft sheen, believable contact shadow, consistent perspective across frames. Minor shortcomings: rim is unusually wide and plate-like and the broth sits nearly flush with it." }, "broth": { "score": 0.45, "reason": "Orange surface reads as a smooth flat gradient disc with mild central sheen; appetizing color but little texture or depth, and toppings/noodles sit pasted on top without visible immersion, occlusion, or meniscus." }, "steam": { "score": 0.75, "reason": "Normal frames at ~1s/3s/4.8s/8.3s/12s show wisps changing shape, rising and dispersing as translucent irregular plumes that never obscure the food. Some plumes read as detached oval blobs, and the reduced frames correctly freeze the animation." }, "composition": { "score": 0.75, "reason": "Dark atmospheric scene with a single centered focal bowl, warm backlight glow, moody palette, and no text or clutter; the dish stays readable. The glow is tasteful, though the scene is sparse and the mobile layout leaves large empty space below." }, "vegan": { "score": 0.4, "reason": "Most items are plant-plausible (greens, brown pieces, corn-like bits), but the small pink-and-white striped pieces resemble narutomaki/fish-cake slices; the appearance is ambiguous rather than clearly animal, so an intermediate score applies." } }, { "noodles": { "score": 0.25, "reason": "Only a handful of thin yellow strands scrawled across the center; they read as flat wire scribbles with no noodle mass, volume, or immersion into the broth." }, "toppings": { "score": 0.3, "reason": "Toppings are glossy pill/capsule-like shapes (striped pink-white cylinders, smooth brown and green lozenges) clustered in the middle; they look like plastic/candy forms rather than vegetables, mushrooms, or fishcake, with no texture or scale variety." }, "bowl": { "score": 0.5, "reason": "A dark hemisphere body with a wide orange rim and readable inner ring, but the deep round body conflicts with the open top-down rim ellipse, there is no foot or visible support (the bowl floats in darkness), and the ceramic surface is featureless smooth gradient." }, "broth": { "score": 0.4, "reason": "Broth is a flat orange radial-gradient disc with no surface texture or sheen variation, and the toppings sit pasted on top with no contact shadows or occlusion, so immersion is not believable." }, "steam": { "score": 0.55, "reason": "Steam is translucent and clearly changes across the time-ordered frames (single plume at 1s, mushroom-shaped puff at 3s, multiple columns at 4.8s), but it is blobby cloud-like fog with detached floating patches and never connects believably to the soup surface." }, "composition": { "score": 0.6, "reason": "Centered dish with a warm glow against a dark vignette is atmospheric and free of clutter, but the bowl appears to float unsupported in a void and the sparse tiny toppings leave the frame feeling empty rather than appetizing." }, "vegan": { "score": 0.5, "reason": "Green and brown shapes could be vegetables or mushrooms, but the pink-and-white striped glossy capsules are ambiguous and could read as fish-cake/naruto sticks; no clearly animal item is unmistakable, so an intermediate score applies." } } ], "diagnostics": { "vegan": 0.4666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.7111111111111111, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100.0, "frameIntervalP95Ms": 116.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "7d29292ba8c3e655f2b42507c623e06827ddbae63062805b38334ed6f490863e", "gradedAt": "2026-09-30T23:18:25.041224+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.55, "toppings": 0.5966666666666667, "bowl": 0.7677777777777778, "broth": 0.58, "steam": 0.7133333333333334, "composition": 0.7588888888888888 }, "reward": 0.6536111111111111, "runId": "openai/gpt-5.5/xhigh", "rewardStdDev": 0.0462037556594379, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6066666666666667, "bowl": 0.8233333333333333, "broth": 0.54, "steam": 0.65, "composition": 0.7766666666666667 }, "reward": 0.6455, "sampleStdDev": 0.0039051248379533164, "judgments": [ { "noodles": { "score": 0.5, "reason": "Golden strands have readable thickness and highlights, but mostly form a flat crisscross network with angular bends and abrupt ends rather than supple, volumetric tangles." }, "toppings": { "score": 0.62, "reason": "Mushrooms have shaded caps and radial details, and leaves and scallion rings are recognizable. However, leaves are stiff and repetitive, tofu lacks texture, and nori resembles a thick rounded card." }, "bowl": { "score": 0.83, "reason": "A coherent elliptical rim, continuous curved walls, warm lacquer-like highlights and grounded shadow establish convincing depth. No material construction defect is observed; the lower body and base are somewhat lost in darkness." }, "broth": { "score": 0.55, "reason": "Warm amber color and scattered oil-like flecks suggest soup, but the surface reads as a flat opaque ellipse, with limited sheen or immersion and ingredients largely layered on top." }, "steam": { "score": 0.64, "reason": "Normal frames show changing translucent plumes above the soup without hiding the food. Their broad, sharply bounded ribbon shapes and repetitive vertical arrangement feel more graphic than irregular dissipating wisps." }, "composition": { "score": 0.78, "reason": "The single bowl is a clear focal point with harmonious amber and green colors against a restrained dark background and no distracting text. Mobile framing clips the bowl sides and leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible noodles, mushrooms, leafy greens, scallions, tofu-like pale pieces and seaweed plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have readable highlights and overlaps, but many have angular bends and abruptly cut ends, forming a flat, wiry tangle rather than supple noodle volume." }, "toppings": { "score": 0.6, "reason": "Mushroom caps with radial gill marks, veined leaves, tofu and scallion rings provide recognizable variety. Repeated smooth shapes lack organic texture, and the nori resembles a thick rounded card." }, "bowl": { "score": 0.82, "reason": "The elliptical opening, continuous thick rim and shaded curved walls form a coherent, substantial vessel with warm lacquer-like highlights. The dark base and contact area are less clearly resolved." }, "broth": { "score": 0.55, "reason": "Warm amber coloring and scattered oil-like flecks suggest soup, but the surface looks largely flat and opaque; noodles and toppings show little convincing immersion or liquid contact." }, "steam": { "score": 0.65, "reason": "Normal frames show changing translucent plumes rising above the dish without obscuring it. Broad, sharply bounded ribbon shapes and several disconnected wisps feel more graphic than naturally dispersing steam." }, "composition": { "score": 0.77, "reason": "The warm bowl and green toppings remain the clear focus against a restrained dark background, without text or UI. Mobile framing clips the dish sides and leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy vegetables, scallions and seaweed with noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Golden highlighted strands are readable and overlap, but many have angular bends, blunt ends and a flat wire-like arrangement rather than supple, immersed volume." }, "toppings": { "score": 0.6, "reason": "Mushrooms have shaded caps and radial details, and leaves and scallion rings are recognizable. Repeated smooth shapes, minimally textured tofu tiles and a thick rounded rectangular nori card limit organic plausibility." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, continuous curved walls and warm lacquer-like highlights create coherent depth and thickness. No material perspective defect is observed; the dark base and contact are somewhat indistinct." }, "broth": { "score": 0.52, "reason": "Warm brown liquid has small oil-like highlights and curved surface accents, but reads as a mostly flat opaque oval; noodles and toppings appear laid above it with little convincing immersion." }, "steam": { "score": 0.66, "reason": "Normal frames show changing translucent plumes that extend upward from the soup without hiding the food. Their broad, sharply bounded ribbon shapes and repeated parallel structure feel more graphic than naturally dissipating wisps." }, "composition": { "score": 0.78, "reason": "The single dish is a clear focal point with warm amber highlights, contrasting greens and a restrained dark background free of text or UI. Mobile framing clips the bowl sides and leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible noodles, mushrooms, leafy vegetables, scallions, tofu-like pieces and seaweed plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.55, "toppings": 0.5333333333333333, "bowl": 0.68, "broth": 0.5666666666666667, "steam": 0.7066666666666667, "composition": 0.7 }, "reward": 0.612, "sampleStdDev": 0.014291605927956433, "judgments": [ { "noodles": { "score": 0.55, "reason": "Thin golden strands curve and overlap with some organic bends, but they are flat, uniform and read as scattered thin lines lacking volume. Some ends are cut off abruptly and they sit on top of the broth rather than in it." }, "toppings": { "score": 0.55, "reason": "Shiitake caps with gill lines, scallion rings, leaves, tofu cubes and a nori sheet are all recognizable. However, the nori is a rigid rounded dark card, the tofu cubes are plain blocks, and the scallions are repetitive, so the set reads as generic symbols." }, "bowl": { "score": 0.68, "reason": "The copper-lacquer bowl has a coherent elliptical rim, inner wall, warm highlight and a believable curved body. The base line is a little odd and the bottom fades into darkness with weak contact or support, but there is no major distortion." }, "broth": { "score": 0.6, "reason": "The warm amber glow and subtle ripple rings give the broth depth. The broth is dark and a bit flat, and the toppings and noodles look laid on top rather than immersed." }, "steam": { "score": 0.72, "reason": "Translucent wispy plumes change shape between frames and rise and fade, with a soft gradient that does not hide the food. They are somewhat similar in form and look slightly like ribbons." }, "composition": { "score": 0.72, "reason": "The bowl is a clear focal point in a dark, moody scene with a warm palette and good glow. The bowl is cropped at the bottom and the nori card feels awkwardly placed." }, "vegan": { "score": 0.95, "reason": "Mushrooms, tofu, scallions, nori, leaves and noodles are all plant-based, with no visible meat or egg." } }, { "noodles": { "score": 0.55, "reason": "Thin golden strands curve and overlap with some organic bends, but they read as scribbly thin lines that are flat and uniform. Several strands end abruptly, and there is little volume or sense of sitting in the broth." }, "toppings": { "score": 0.5, "reason": "Scallion rings, shiitake caps with gill lines, tofu cubes, leaves and a nori sheet are all recognizable. The nori is a rigid rounded card floating above the bowl. The tofu is blocky, the leaves are generic, and the scallions are a repetitive scatter." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a warm copper gradient, a believable cavity and a curved body with a rim highlight. The base fades into darkness so contact is vague, and the thin band line on the body looks slightly odd." }, "broth": { "score": 0.55, "reason": "A warm glowing amber center with faint concentric ripples gives some depth. It is fairly flat, and the dark band near the rim looks like a solid ring. The toppings sit on top more than they are immersed." }, "steam": { "score": 0.68, "reason": "Translucent, irregular wisps rise and change shape between frames without hiding the food. They are somewhat repetitive and ribbon-like, and the plumes are somewhat uniform." }, "composition": { "score": 0.68, "reason": "A centered, readable bowl sits in a dark atmospheric scene with warm lighting and good color harmony. The nori card is a distracting element, and the bowl is cropped at the bottom." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, tofu, scallions, nori, leaves and noodles are visible. There are no animal products." } }, { "noodles": { "score": 0.55, "reason": "Thin golden strands overlap with some curves and read as noodles. They are short, flat and uniform in width, and they look scribbled on top of the broth. There are some loose fragments and little sense of volume." }, "toppings": { "score": 0.55, "reason": "Scallion rings, shiitake caps with gill lines, tofu cubes, leaves and a nori sheet are all recognizable, which gives decent variety. The nori is a stiff rounded rectangle standing upright like a card. The tofu cubes are plain blocks and the scallion rings look like flat symbols." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim with a warm copper gradient, a dark cavity and a curved body. The lower line looks odd, and the base fades into darkness with little visible support or contact. Overall it is competent and consistent in perspective." }, "broth": { "score": 0.55, "reason": "The warm glow in the center of the broth gives some depth, and faint ripple rings are visible. The surface is a fairly flat dark brown with little sheen. The ingredients sit on top more than in the liquid." }, "steam": { "score": 0.72, "reason": "Translucent wisps rise in layers and change shape from frame to frame, so the motion is convincing. They are somewhat uniform, vertical and ribbon-like. Some of them obscure the nori a little." }, "composition": { "score": 0.7, "reason": "The bowl is centered and is a clear focal point against a dark, moody background with warm lighting. The proportions are balanced. The nori placement is a little awkward, and the bowl crops at the bottom." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, scallions, leaves, nori and noodles. No animal products are visible." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.65, "bowl": 0.8, "broth": 0.6333333333333333, "steam": 0.7833333333333333, "composition": 0.7999999999999999 }, "reward": 0.7033333333333334, "sampleStdDev": 0.025166114784235794, "judgments": [ { "bowl": { "reason": "Coherent wide ceramic bowl with a consistent rim ellipse, thick rim band, smooth gradient body and a visible base line; no broken joins or perspective clash, though the form is unusually shallow and the lower body fades into shadow.", "score": 0.8 }, "broth": { "reason": "Dark broth with a warm central glow, but the surface is marked by flat concentric elliptical rings resembling target discs, and noodles/toppings read as laid on top rather than partially submerged; depth and immersion are weak.", "score": 0.55 }, "composition": { "reason": "Desktop framing is balanced with a warm glow and the dish as clear focus, no text or clutter; however the mobile capture clips the bowl's right edge, and background atmosphere does most of the mood work.", "score": 0.7 }, "noodles": { "reason": "Continuous yellow strands with varied bends and crossings are recognizable, but they read as thin flat squiggles scattered across the surface with uniform thinness, abrupt ends and little volume or believable submersion.", "score": 0.6 }, "steam": { "reason": "Translucent gray wisps that change shape and position across the normal frames, rising without hiding the food; plumes are somewhat uniform in their S-curve styling and emerge from behind the bowl rather than clearly from the soup, but genuine motion is evident (and correctly static under reduced motion).", "score": 0.75 }, "toppings": { "reason": "Good variety: shiitake-style mushrooms with gill detail, green leaves, scallion rings, tofu cubes and an upright nori sheet. Defects: the nori looks like a glossy rounded card, tofu cubes are flat and plastic-like, scallions cluster as confetti, and some leaves drape oddly over the rim.", "score": 0.65 }, "vegan": { "reason": "Visible ingredients are mushrooms, tofu, scallions, leafy greens, nori and noodles in plant broth \u2014 all plausibly vegan, with no animal-derived toppings.", "score": 1.0 } }, { "noodles": { "score": 0.6, "reason": "Continuous pale-yellow strands with wavy bends are readable as noodles, but they read as flat uniform-stroke squiggles pasted over the surface rather than a coherent mass with volume or immersion; placement is scattered with gaps." }, "toppings": { "score": 0.6, "reason": "Mushroom caps with gill fans, tofu cubes, scallion rings and stylized green leaves are recognizable, but the nori is a large glossy dark slab with rounded corners and vertical stripes that reads as a rigid glassy card standing in the bowl; leaves are simple flat vector shapes." }, "bowl": { "score": 0.8, "reason": "Coherent ellptical rim with warm highlight band, smooth tapered walls, subtle foot line and grounding shadow; rim, body and base perspective agree and toppings sit correctly behind the rim. Stylistically simplified ceramic but no broken geometry." }, "broth": { "score": 0.65, "reason": "Warm amber surface with central glow, small oil specks and faint ripple ellipses gives liquid depth, but the concentric ellipse rings look mechanical and toppings rest on the surface with little visible occlusion or immersion shadow." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the bowl and change shape and position across desktop-1/2/3, thinning and fading near the top without hiding food. Plumes are somewhat similar wavy bands and the reduced frames freeze them, but normal-frame motion is evident." }, "composition": { "score": 0.85, "reason": "Dark atmospheric scene with the bowl as a clear focal point, warm rim glow balancing cool steam, floating ember specks for depth, and no clutter or UI. Mobile crops the bowl slightly at the edges but presentation remains balanced." }, "vegan": { "score": 1, "reason": "Visible toppings are mushrooms, tofu cubes, scallions, leafy greens, nori and noodles in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.6, "reason": "Pale yellow wavy strands are recognizable as noodles and vary in curvature, but they read as flat squiggles lying on the broth surface with repetitive S-curves, little strand volume, and no believable dipping or clumping into the liquid." }, "toppings": { "score": 0.7, "reason": "Good variety: shiitake caps with gill lines, scallion rings, leafy greens, tofu cubes, and an upright nori sheet. Defects: tofu cubes are plain glossy rounded blocks, the nori has card-like rounded corners with rigid vertical ribs, and the leaves are simple flat almond shapes." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with visible thickness, glossy copper/orange ceramic body with a believable taper, warm highlight streak, and soft contact shadow beneath. Slightly bulbous pot-like body and only a faint foot line, but no broken joins or perspective mismatch." }, "broth": { "score": 0.7, "reason": "Warm amber-brown liquid with a center glow, scattered light speckles (oil/sesame), and restrained ripple arcs; toppings and noodles sit in contact with the surface. Slightly flat in depth and the backlight glow reads a bit artificial." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps clearly change silhouette across the normal frames (desktop-1 through pointer-left), rising, undulating and fading without obscuring the food; plume shapes are somewhat repetitive, and wisps are static in the reduced-motion frames, which is expected behavior there." }, "composition": { "score": 0.85, "reason": "Bowl is the clear focal point, centered with warm copper tones against a tasteful dark vignette and subtle floating particles; steam fills the upper frame without clutter, text, or competing elements. Mobile crop also keeps the dish readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, scallions, leafy greens, nori, noodles and plant-looking broth; no eggs, meat, fish or dairy appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "2fa90a883cf41d32be3982100041bda64a4aebd7e4dca81277c03f7e7455e8e8", "gradedAt": "2026-09-30T23:16:59.100385+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5011111111111112, "toppings": 0.42000000000000004, "bowl": 0.7077777777777777, "broth": 0.40444444444444444, "steam": 0.5344444444444444, "composition": 0.7288888888888889 }, "reward": 0.5402777777777777, "runId": "openai/gpt-5.6-luna/high", "rewardStdDev": 0.046512344040567845, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.45333333333333337, "toppings": 0.41000000000000003, "bowl": 0.7333333333333333, "broth": 0.28, "steam": 0.43666666666666665, "composition": 0.7666666666666667 }, "reward": 0.51, "sampleStdDev": 0.005766281297335413, "judgments": [ { "noodles": { "score": 0.46, "reason": "Golden strands are continuous and readable, but mostly form a flat crisscross of stiff arcs with exposed blunt ends rather than supple, immersed noodle volume." }, "toppings": { "score": 0.43, "reason": "Green leaves, yellow kernels and shaded tofu-like pieces provide variety, but oval leaves with single straight stripes and rounded rectangular blocks remain generic, minimally textured food symbols." }, "bowl": { "score": 0.73, "reason": "The elliptical rim joins a coherent curved body, with warm highlights and darker lower shading suggesting lacquer and depth. The broad highlight is somewhat patch-like, and the base/contact is weakly defined." }, "broth": { "score": 0.28, "reason": "A dark surface is visible inside the rim, but it reads almost like the bowl interior: little liquid sheen, surface texture or immersion establishes broth, and ingredients appear layered above it." }, "steam": { "score": 0.44, "reason": "Five faint plumes change bends across normal frames and fade toward their tops without obscuring food. However, they resemble evenly spaced, constant-width waving lines rather than spreading, dissipating wisps." }, "composition": { "score": 0.77, "reason": "The single dish is clearly framed against a restrained dark background, with harmonious red, gold and green colors and no distracting UI. Tall cropped steam on desktop and the relatively small mobile dish weaken balance slightly." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, corn-like kernels and tofu-like pieces plausibly form vegan ramen; no animal-derived topping is apparent. No material compliance defect is observed." } }, { "noodles": { "score": 0.45, "reason": "Golden strands are continuous and overlap visibly, but mostly form long, flat crossing strokes with little supple curling, volume, or immersion." }, "toppings": { "score": 0.4, "reason": "Green vegetables, yellow kernel-like pieces, and shaded tofu-like blocks provide variety, but oval leaves with single straight stripes and rounded rectangles remain generic food symbols with little organic detail." }, "bowl": { "score": 0.73, "reason": "The elliptical rim and curved red walls form a coherent vessel, with restrained highlights and shading. The broad dark lip and indistinct base/contact make its material and support less convincing." }, "broth": { "score": 0.28, "reason": "A dark reddish surface occupies the cavity, but lacks liquid sheen, texture, or visible depth; ingredients appear layered over a flat field rather than immersed." }, "steam": { "score": 0.43, "reason": "Five faint trails visibly change curvature across normal frames and fade toward their tops without obscuring food. Their uniform stroke widths and parallel, persistent columns look mechanical, with little spreading or breakup." }, "composition": { "score": 0.77, "reason": "A single centered bowl, warm gold and green accents, and restrained dark surroundings create a clear focal dish without clutter. Mobile framing leaves substantial empty space, and desktop steam trails extend beyond the top edge." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, greens, and corn-like kernels alongside noodles; no animal-derived ingredient is apparent. No material visual compliance defect is observed." } }, { "noodles": { "score": 0.45, "reason": "Golden strands are continuous and readable, but mostly form long, flat crossing arcs with blunt exposed ends rather than supple coils and convincing volume." }, "toppings": { "score": 0.4, "reason": "Green vegetables, yellow kernels and tofu-like pieces offer variety, but leaves are simple striped ovals, greens are rigid sticks, and rounded blocks lack organic texture." }, "bowl": { "score": 0.74, "reason": "The elliptical rim joins a coherent curved body with warm lacquer-like shading and visible thickness. Broad, flat highlights and weakly defined ground contact limit material conviction." }, "broth": { "score": 0.28, "reason": "A dark burgundy surface is visible inside the rim, but lacks liquid sheen, texture or depth; ingredients appear layered over it with little immersion." }, "steam": { "score": 0.44, "reason": "Five faint curved trails change shape across normal frames and leave the food unobscured. They remain uniform narrow lines with little visible spreading or dissipating, resembling waving cords more than vapor." }, "composition": { "score": 0.76, "reason": "A single centered dish, warm gold-and-red palette and restrained dark background create a clear focal point without clutter. Desktop steam extends beyond the top edge, while mobile presentation leaves considerable empty space." }, "vegan": { "score": 1, "reason": "Visible noodles, green vegetables, corn-like kernels and tofu-like blocks plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.4, "bowl": 0.6333333333333333, "broth": 0.4, "steam": 0.5666666666666667, "composition": 0.6533333333333333 }, "reward": 0.517, "sampleStdDev": 0.012757350822173073, "judgments": [ { "noodles": { "score": 0.5, "reason": "Golden thin strands overlap with some curves and a glow, but they read as flat scribbled lines crossing the surface with little volume or immersion in the broth." }, "toppings": { "score": 0.4, "reason": "Green leaves, tofu cubes and corn-like dots are recognizable, but they are simple, flat, glossy shapes. The scallion sticks are rigid, the pinkish slivers are ambiguous, and there is little texture or variety." }, "bowl": { "score": 0.62, "reason": "The red lacquer bowl has a coherent elliptical rim, tapered walls and a soft highlight. The base is vague, and the bowl seems to float with only a dark shadow for contact. The rim is thick and dark." }, "broth": { "score": 0.4, "reason": "The broth is a dark maroon surface that reads as murky and flat. Ingredients sit on top rather than in it, and there is little liquid sheen or depth." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps change shape between frames and rise, so they read as animated. They look like uniform thin lines rather than soft spreading steam, and the reduced-motion frames are straight bars." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with a warm glow, and the food is the focus. The palette is cohesive and there is no clutter, but the image is somewhat plain." }, "vegan": { "score": 0.8, "reason": "The visible ingredients are tofu, greens, corn and noodles, all plant-based. The pinkish slivers are ambiguous and could be read as meat." } }, { "noodles": { "score": 0.5, "reason": "Golden strands are thin and overlap with some curves and a glow, but they read as flat scribbled lines with repeated loops and little volume. They sit on the surface rather than in the broth." }, "toppings": { "score": 0.4, "reason": "Toppings are simple: two green leaf shapes with a stripe, cream rounded rectangles for tofu, yellow blobs for corn, and thin green sticks for scallion. Some read as generic cards and blobs, with little texture or detail, and the pinkish mushroom slivers are barely readable." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a red lacquer gradient and a highlight, and a dark interior. The base is missing and the bowl looks like it floats over a soft shadow, and the rim is slightly flat." }, "broth": { "score": 0.4, "reason": "The broth is a dark maroon surface that is mostly hidden by noodles. It has little sheen, depth or immersion, and the ingredients look laid on top rather than sitting in liquid." }, "steam": { "score": 0.55, "reason": "Five thin wavy wisps rise and sway between frames, so the motion is visible and they do not hide the food. They are thin line strokes that look similar to each other and start abruptly above the rim. In reduced motion they turn into straight bars." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark, moody background with warm colors and a clear focal dish, and there is no clutter. The scene is simple, the bowl floats, and the overall polish is modest." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are noodles, tofu, corn, greens and scallion, which are plausibly vegan. The small pinkish slivers are ambiguous and could be read as meat or shrimp." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands overlap with some curvature and a glowing center, but they read as flat scribbled lines and thin wire-like loops. There is little volume, and the strands sit on the surface rather than in the broth." }, "toppings": { "score": 0.4, "reason": "The green leaves, tofu cubes, corn-like dots and scallion sticks are recognizable. They are simple flat shapes: the tofu reads as rounded cards and the dots as plain blobs. The mushroom-like slices are barely legible and there is no nori." }, "bowl": { "score": 0.68, "reason": "The red lacquer bowl has a coherent elliptical rim, a curved body with a gradient, and a soft highlight. The base and foot are missing, so it seems to float above a shadow. The highlight patch looks slightly pasted on." }, "broth": { "score": 0.4, "reason": "The broth is a very dark maroon surface that is barely visible and reads as murky. A warm glow sits under the noodles, but there is little sheen or depth, and the toppings seem to lie on top rather than sit in the liquid." }, "steam": { "score": 0.55, "reason": "Five thin wavy wisps rise and change shape across frames, so there is real motion. They are narrow line-like strands, not soft translucent wisps, and they look mechanical. The reduced-motion frames show rigid straight bars." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point against a dark atmospheric background with warm, cohesive colors. The food is small and dim compared with the bowl. There is no clutter or text." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are noodles, leaves, tofu-like cubes, corn and scallion, all plausibly vegan. The pinkish curved slices are ambiguous, possibly mushroom or meat." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.55, "toppings": 0.45, "bowl": 0.7566666666666667, "broth": 0.5333333333333333, "steam": 0.6, "composition": 0.7666666666666667 }, "reward": 0.5938333333333333, "sampleStdDev": 0.06410213204982602, "judgments": [ { "noodles": { "score": 0.6, "reason": "Yellow strands form a coherent heap with varied bends and overlapping arcs, but they are uniform flat strokes with rounded ends, reading slightly wire-like with little volume or submersion." }, "toppings": { "score": 0.5, "reason": "Readable tofu blocks, corn clusters and scallion sticks exist, but the green ovals are featureless disks and the tofu pieces are glossy rounded cards; primitive sticker-like shapes dominate the topping set." }, "bowl": { "score": 0.8, "reason": "Deep red bowl with a coherent elliptical rim, dark inner cavity, shaded tapered body, subtle ceramic highlight and a grounded contact shadow; perspective of rim and body is consistent with only minor flatness in the interior." }, "broth": { "score": 0.6, "reason": "Dark maroon surface with a restrained warm glow beneath the noodles is appetizing and not murky, but the liquid reads flat and ingredients sit atop it with weak occlusion or immersion cues." }, "steam": { "score": 0.7, "reason": "Thin translucent tan wisps rise from the bowl and visibly change shape and position across desktop-1/2/3 and pointer frames, showing continuous animation; they never hide the food but remain somewhat uniform thin lines rather than spreading, fading plumes." }, "composition": { "score": 0.8, "reason": "Centered bowl as clear focal point against a dark atmospheric background with warm glow and soft shadow; balanced proportions and no clutter, though the scene is simple and the mobile framing leaves generous empty space." }, "vegan": { "score": 0.5, "reason": "Tofu, corn, greens and noodles are plant-appropriate, but the small pink curved strips are ambiguous and could read as fish-cake slices, so appearance cannot confirm vegan compliance." } }, { "noodles": { "score": 0.6, "reason": "Warm yellow strands with rounded ends, coherent crossings and plausible thickness read as noodles, but they are flat uniform squiggle strokes with little volume, lying on top of the broth without occlusion." }, "toppings": { "score": 0.45, "reason": "Variety exists (tan tofu blocks, green slit ovals, yellow dot clusters, pink arcs) but each is a simple geometric symbol\u2014featureless blocks, disks and spheres\u2014with no texture or organic detail, and the pink arcs read ambiguously (possibly fish-cake)." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim with visible thickness, consistent interior/exterior perspective, soft ceramic-like shading and a left highlight; supported by a contact shadow. Minor quibble: shallow profile and slightly flat interior join." }, "broth": { "score": 0.55, "reason": "Dark maroon surface with a restrained warm glow around the noodles, but the liquid is essentially a flat solid-colored ellipse with no sheen, texture, or visible immersion/occlusion of ingredients." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise and fade toward the top and clearly change shape across the normal temporal frames, so motion is supported; however they are uniform squiggly lines that never spread or dissipate, and correctly freeze as straight lines under reduced motion." }, "composition": { "score": 0.8, "reason": "Centered focal bowl against a dark atmospheric backdrop, warm/cold contrast, gentle glow and no clutter or text; simple but balanced, with the dish clearly the focus." }, "vegan": { "score": 0.5, "reason": "Tofu blocks, green vegetable slices and corn-like dots are plant-appropriate, but the thin pink arcs are ambiguous and could be read as fish-cake/naruto slices, so appearance alone cannot confirm compliance." } }, { "noodles": { "score": 0.45, "reason": "Yellow-orange strands are readable as noodles with a few gentle hooks and a warm glow, but most strands are near-straight with uniform thickness, reading as rigid rods/wires scattered in a tangle rather than supple coiled strands." }, "toppings": { "score": 0.4, "reason": "Some variety: green scallion segments, two striped green ovals, beige rounded tofu-like blocks with simple shading, orange dot clusters and thin pink strips. However, most are flat generic shapes (featureless ovals, rounded rectangles, plain dots) with little organic texture, so the generic forms dominate." }, "bowl": { "score": 0.72, "reason": "Coherent stylized bowl: elliptical rim, dark interior cavity, rounded red body with soft sheen, consistent elevation, grounded by a soft contact shadow and small rim highlights. No broken joins or perspective mismatch; slightly simplified ceramic rendering keeps it from scoring higher." }, "broth": { "score": 0.45, "reason": "Broth is a flat, very dark maroon ellipse with a warm glow behind the noodles; ingredients sit on the surface but there is essentially no liquid sheen, texture, or depth, and no occlusion/immersion cues beyond the glow, making it read as a solid dark plate." }, "steam": { "score": 0.5, "reason": "Thin translucent wisps rise from the bowl and visibly change shape/waist across the time-ordered desktop frames, indicating animated motion. However, the plumes are several nearly identical uniform wavy lines that don't spread or fade convincingly, feeling mechanical rather than organic." }, "composition": { "score": 0.7, "reason": "Clean dark atmospheric scene with the bowl centered as the clear focal point, warm-on-dark palette, pleasing glow and no clutter or UI. Somewhat sparse and the simplified food limits visual richness." }, "vegan": { "score": 0.5, "reason": "Visible toppings are tofu, scallion/greens and dot-like vegetables; thin pink strips are ambiguous \u2014 they could be pickled ginger (vegan) or naruto/crab stick (animal-derived). Ambiguous appearance warrants an intermediate score." } } ], "diagnostics": { "vegan": 0.5 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.7777777777777778, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 116.69999999999891, "frameIntervalP95Ms": 116.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5ca395667aaf9672354df68a191d4abb7b6328e3f6976c2774a9d862a4f02d0b", "gradedAt": "2026-09-30T23:17:52.936441+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.29777777777777775, "toppings": 0.27999999999999997, "bowl": 0.6155555555555556, "broth": 0.4311111111111111, "steam": 0.5633333333333334, "composition": 0.6755555555555556 }, "reward": 0.44783333333333336, "runId": "openai/gpt-5.6-luna/low", "rewardStdDev": 0.06382201814421098, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.29333333333333333, "toppings": 0.25, "bowl": 0.6466666666666667, "broth": 0.3433333333333333, "steam": 0.59, "composition": 0.71 }, "reward": 0.44433333333333336, "sampleStdDev": 0.007285830998131471, "judgments": [ { "noodles": { "score": 0.28, "reason": "Yellow curved strands are readable, but form a single row of repeated short loops with little overlap, continuity or noodle volume." }, "toppings": { "score": 0.25, "reason": "Green shapes and a pale tofu-like square suggest toppings, but featureless circles, straight strips and a flat square dominate; organic texture and ingredient detail are absent." }, "bowl": { "score": 0.64, "reason": "Elliptical opening and shaded curved body form a coherent bowl. The thin outer rim outline looks slightly separated, and the broad black shadow sits behind rather than convincingly supporting the base." }, "broth": { "score": 0.33, "reason": "Orange color clearly marks the soup surface, but nested flat ellipses give it a solid disc appearance, without liquid sheen or convincing ingredient immersion." }, "steam": { "score": 0.57, "reason": "Translucent wisps change height and opacity across normal frames without obscuring food. Their uniformly thin, repeated sinusoidal shapes lack natural spreading and diffuse edges." }, "composition": { "score": 0.71, "reason": "A single readable dish contrasts well against the restrained dark background, with warm coordinated colors and no clutter. Extensive empty space and oversized mobile steam weaken the balance." }, "vegan": { "score": 0.88, "reason": "Visible green vegetables, pale tofu-like topping and noodles are compatible with vegan ramen; no animal-derived ingredient is evident, though the simplified toppings remain somewhat ambiguous." } }, { "noodles": { "score": 0.3, "reason": "Yellow strands are readable, but form a single row of repeated tiny loops with little overlap, volume, or believable immersion." }, "toppings": { "score": 0.25, "reason": "Green shapes and a pale tofu-like square suggest toppings, but featureless circles, straight lines, and a flat square dominate; organic shapes and ingredient textures are missing." }, "bowl": { "score": 0.65, "reason": "The elliptical opening and curved shaded body create a coherent bowl silhouette. The thin outer rim looks loosely attached, and the broad shadow sits behind rather than convincingly beneath the base." }, "broth": { "score": 0.35, "reason": "The warm orange surface is readable as soup, but concentric flat ellipses lack liquid sheen or depth, and ingredients appear pasted onto the surface." }, "steam": { "score": 0.6, "reason": "Normal frames show changing heights and opacity of translucent wisps above the soup without obscuring it. The uniformly thin, similarly wavy strokes remain mechanical, with limited spreading." }, "composition": { "score": 0.72, "reason": "A single uncluttered dish remains the focus against a restrained dark background, with a harmonious warm palette. The low placement and substantial empty upper area make the framing somewhat sparse." }, "vegan": { "score": 0.9, "reason": "Visible greens and a tofu-like pale square are plausibly plant-based, with no clear animal ingredients. Several abstract toppings are too featureless to identify confidently." } }, { "noodles": { "score": 0.3, "reason": "Yellow strands suggest noodles, but form a sparse row of repeated tiny loops with no convincing overlap, supple length or volume." }, "toppings": { "score": 0.25, "reason": "Green shapes and a pale square suggest vegetables and tofu, but featureless disks, straight lines and a flat block dominate; organic shapes and ingredient texture are missing." }, "bowl": { "score": 0.65, "reason": "The elliptical opening and shaded curved body make a coherent bowl. The thin outer outline looks separated from the thick inner rim, and the broad shadow gives weak grounding." }, "broth": { "score": 0.35, "reason": "Orange color reads as soup, but nested flat ellipses lack liquid sheen or depth, and ingredients appear pasted onto the surface rather than immersed." }, "steam": { "score": 0.6, "reason": "Normal frames show changing heights and opacity in translucent wisps above the soup without obscuring it. Nearly identical narrow sinusoidal strokes remain mechanical and show little spreading." }, "composition": { "score": 0.7, "reason": "A single uncluttered dish remains readable against a restrained dark background with warm colors. The low placement leaves substantial empty space, and the hard black shadow weakens lighting cohesion." }, "vegan": { "score": 0.9, "reason": "Visible greens, pale tofu-like topping and noodles plausibly indicate vegan ramen, with no apparent animal ingredients; the simplified topping shapes leave some identity uncertain." } } ], "diagnostics": { "vegan": 0.8933333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.26666666666666666, "toppings": 0.25666666666666665, "bowl": 0.48333333333333334, "broth": 0.43333333333333335, "steam": 0.5, "composition": 0.5333333333333333 }, "reward": 0.38583333333333336, "sampleStdDev": 0.020966242709015207, "judgments": [ { "noodles": { "score": 0.25, "reason": "Noodles are tiny repeated hook-shaped squiggles along the back of the broth. They are not continuous strands and have no volume or overlap, so they read as a row of small glyphs." }, "toppings": { "score": 0.22, "reason": "Toppings are a flat tan square (tofu), a dark green circle (nori), a light green circle and two thin lines (greens), and scattered yellow dots. These are generic geometric shapes with no organic detail, so nothing reads clearly as food." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent terracotta gradient body, a dark rim ellipse and a contact shadow. However, the thin outer halo ring floats detached from the rim, and the flat-cut top edge and pointed base make the geometry slightly off." }, "broth": { "score": 0.45, "reason": "The orange gradient broth with a concentric inner ring reads as warm liquid. It looks like a flat target disc, and the toppings sit on top of it rather than being immersed." }, "steam": { "score": 0.5, "reason": "Thin wavy translucent wisps rise and change between frames, so the motion is visible. The wisps are uniform, thin-line squiggles that look mechanical, and in the reduced-motion frames they form an identical comb-like row." }, "composition": { "score": 0.5, "reason": "The bowl is centered in a dark scene with a soft warm glow, so the focus is clear. The bowl is small, the food is sparse and the lower half of the frame is empty." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are tofu, greens, nori and noodles, with no meat, egg or fish. The yellow dots are ambiguous, but nothing looks animal-derived." } }, { "noodles": { "score": 0.25, "reason": "Noodles are tiny repeated hook-shaped squiggles lined up along the rim. There is no continuous volume or overlap, and they do not read as a noodle mass in broth." }, "toppings": { "score": 0.25, "reason": "The toppings are a flat green circle, a cream square, a dark teal circle, two thin green lines and small yellow dots. They are generic plastic-looking shapes with no organic detail, so nothing reads clearly as a mushroom, leaf or nori." }, "bowl": { "score": 0.45, "reason": "The bowl has a rim, a tapered body, a shadow and a gradient. However, the outer ring ellipse floats detached around the bowl, and the body's flat top edge joins the rim awkwardly, so the perspective is inconsistent." }, "broth": { "score": 0.4, "reason": "The broth is an orange gradient ellipse with concentric rings that read as a target disc. It has little texture or depth, and the toppings sit on top of it rather than in it." }, "steam": { "score": 0.5, "reason": "Thin wavy translucent wisps rise and change between frames, so the motion is plausible. They look like similar mechanical squiggles, and in the reduced-motion frames they line up as regular parallel lines." }, "composition": { "score": 0.55, "reason": "The dark scene has a warm glow and a centered bowl that is clearly the focal point, with no clutter. The bowl is small in the frame and the food itself is weak." }, "vegan": { "score": 0.8, "reason": "The visible toppings are plant-like: greens, tofu-like cube and a dark seaweed-like ball. Nothing looks animal-derived." } }, { "noodles": { "score": 0.3, "reason": "Noodles are tiny repeated hook or loop glyphs along the rim, with no continuous strands or volume. They read as scattered symbols rather than a noodle mass in broth." }, "toppings": { "score": 0.3, "reason": "The toppings are a green circle, a flat tan square, a dark green plastic-looking disc, thin green lines and yellow dots. They are generic shapes with no organic texture, and nothing clearly reads as a mushroom, leaf or nori." }, "bowl": { "score": 0.5, "reason": "The bowl has a readable rim, a terracotta gradient body and a ground shadow. The faint outer ellipse ring floats detached from the rim, and the body is flat with few highlights. The rim and body perspectives are slightly inconsistent, with a pointed base." }, "broth": { "score": 0.45, "reason": "The orange radial gradient looks warm and readable as broth. It is a flat concentric disc, though, with little surface texture, and the toppings sit on it without immersion." }, "steam": { "score": 0.5, "reason": "Thin wavy wisps rise and change between frames, so the motion is visible. The wisps are uniform and mechanical, and they look like similar squiggle lines. They do not hide the food." }, "composition": { "score": 0.55, "reason": "The bowl is centered in a dark scene with a warm glow, and the food is the focus. The bowl is small relative to the frame and the ingredients are simplistic, but there is no clutter." }, "vegan": { "score": 0.85, "reason": "The visible toppings are plant-like: greens, a tofu-like block, a seaweed-like disc and noodles. No animal ingredients are visible." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.3333333333333333, "toppings": 0.3333333333333333, "bowl": 0.7166666666666667, "broth": 0.5166666666666667, "steam": 0.6, "composition": 0.7833333333333333 }, "reward": 0.5133333333333333, "sampleStdDev": 0.03185252475602735, "judgments": [ { "noodles": { "score": 0.3, "reason": "Noodles are rendered as a ring of small, repeated comma-shaped squiggles around the broth edge plus scattered yellow dots; they are disconnected flat marks, not continuous strands, so they read as decorative pattern rather than supple noodles." }, "toppings": { "score": 0.3, "reason": "Toppings are generic geometric primitives: a tilted beige square card standing in for tofu, flat teal and green ellipse disks, plain yellow dots, and two thin green sticks. No organic texture, detail, or variety; featureless forms dominate." }, "bowl": { "score": 0.7, "reason": "Stylized bowl is coherent: smooth orange/red gradient body, elliptical rim with a darker inner wall, believable hemisphere-like depth, grounded by a contact shadow. Slight artificiality in the floating thin rim outline and simple vector shading, but no broken geometry." }, "broth": { "score": 0.45, "reason": "Broth is a flat orange ellipse with a simple radial gradient; it reads as a solid disc rather than liquid. The tofu square appears pasted on/tilted above the surface, and immersion/occlusion of ingredients is only weakly suggested." }, "steam": { "score": 0.65, "reason": "Translucent wavy wisps rise from the soup and fade without hiding the food; across the time-ordered desktop frames their positions and shapes change (e.g., desktop-1 vs desktop-2 vs pointer frames), indicating animated rising steam. Wisps are somewhat uniform/mechanical sine curves." }, "composition": { "score": 0.75, "reason": "Centered bowl as a clear focal point on a dark atmospheric background with a restrained warm glow; balanced proportions, no text, clutter, or competing elements. Minimalism is tasteful but the scene is sparse." }, "vegan": { "score": 1.0, "reason": "Visible ingredients (tofu-like square, green vegetable shapes, seaweed/nori-like dark circle, corn-like dots, noodle squiggles, plant-colored broth) plausibly form vegan ramen; no animal-derived toppings are visible." } }, { "noodles": { "score": 0.3, "reason": "Noodles are reduced to a ring of small identical yellow comma/curl shapes plus scattered dots on the broth surface. No continuous strands, thickness variation, overlap or volume; the repeated flat loops read as decorative marks rather than supple noodles." }, "toppings": { "score": 0.35, "reason": "Toppings are generic flat geometry: two overlapping green ellipses with thin green stalks (loosely scallion), a rotated beige square (tofu), a featureless dark teal disk and small yellow dots. No texture, leaf or seaweed detail, scale variation, or grounding; plastic-looking simple shapes dominate, well below the 0.50 ceiling." }, "bowl": { "score": 0.7, "reason": "The bowl is a coherent stylized form: elliptical rim with a thin lip outline, dark inner band, smoothly curved orange body with believable gradient shading and a soft contact shadow. Perspective of rim, cavity and body is consistent; only shortcoming is the very thin wire-like rim line that reads slightly detached from the body at the sides." }, "broth": { "score": 0.55, "reason": "Broth is a restrained radial orange gradient with a darker edge ring, giving some depth and a plausible liquid read. However toppings and noodle curls show no occlusion, shadows or immersion cues and appear pasted onto the flat surface rather than sitting in liquid." }, "steam": { "score": 0.5, "reason": "Steam is present and animated: comparing desktop frames the squiggle wisps shift, sway and change height over time, and do not obscure the food. But they are thin uniform sine-line plumes rather than translucent irregular wisps, with many near-identical shapes and no visible spread or fade." }, "composition": { "score": 0.8, "reason": "Bowl is centered and clearly focal against a dark background with a subtle warm glow; color contrast between orange broth, green accents and dark surroundings is attractive; no clutter, text or competing elements. Slightly sparse and small in frame, but balanced and coherent on both desktop and mobile." }, "vegan": { "score": 0.9, "reason": "Visible toppings are tofu square, green scallion-like stalks/ellipses, a dark disk plausibly nori or mushroom, and yellow vegetable dots in plant-colored broth; no meat, fish, egg or dairy shapes are apparent." } }, { "noodles": { "score": 0.4, "reason": "Noodles are rendered as a ring of repeated small pale loop/oval shapes hugging the broth edge (mobile shows a spring-like coil row); they are schematic repeated flat loops rather than continuous supple strands with volume and overlap." }, "toppings": { "score": 0.35, "reason": "Toppings reduce to generic geometry: two flat green circles, two thin green sticks, a plain tan square (tofu) and a featureless dark teal sphere, plus scattered yellow dots; no organic texture or readable ingredient detail dominates." }, "bowl": { "score": 0.75, "reason": "Coherent stylized bowl: elliptical rim, dark inner band, smooth curved body with gradient shading, visible foot and grounded drop shadow; minor defect is a thin detached-looking ellipse ringing the rim." }, "broth": { "score": 0.55, "reason": "Glowing orange gradient surface reads as liquid with warm sheen, but it is quite flat/neon; noodle loops appear pasted along the edge and toppings sit on the surface with little occlusion or immersion." }, "steam": { "score": 0.65, "reason": "Thin translucent wavy wisps rise and fade above the bowl; shapes clearly change across desktop-1/2/3 and pointer frames, showing motion, though the strokes are somewhat mechanical squiggle lines and uniform in the reduced frames." }, "composition": { "score": 0.8, "reason": "Clean dark atmospheric scene with warm glow, centered readable dish, no clutter or UI; balanced proportions on desktop and mobile, food remains the focal point." }, "vegan": { "score": 1.0, "reason": "Visible toppings (tofu square, green vegetable circles/stalks, seaweed-like dark ball, yellow vegetable bits, noodles, plant-toned broth) contain no animal-derived ingredients." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8977777777777778, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.700000000000728, "frameIntervalP95Ms": 33.400000000000546, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "52d8ef33cc6f22bed0ca4b0785ec6fb5341ccaf83c624bdabca6d4c9e8b262d9", "gradedAt": "2026-09-30T23:15:48.095300+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5333333333333333, "toppings": 0.5733333333333334, "bowl": 0.7677777777777778, "broth": 0.6066666666666667, "steam": 0.5866666666666667, "composition": 0.7988888888888889 }, "reward": 0.6321111111111112, "runId": "openai/gpt-5.6-luna/max", "rewardStdDev": 0.06633361250639189, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.57, "bowl": 0.8333333333333333, "broth": 0.52, "steam": 0.5766666666666667, "composition": 0.8466666666666667 }, "reward": 0.6323333333333333, "sampleStdDev": 0.005251983752196222, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous yellow strands have smooth bends and readable thickness, but mostly form parallel flat waves with little tangled volume or convincing submersion." }, "toppings": { "score": 0.56, "reason": "Leaves, a shaded mushroom, tofu, scallion rings and corn are recognizable and varied. Texture is sparse; corn resembles uniform disks, tofu is very geometric, and nori is a rigid rectangular card." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, inner cavity and rounded walls join coherently, with restrained highlights and a grounding shadow. No material construction defect is observed; the ceramic finish and base definition are simplified." }, "broth": { "score": 0.52, "reason": "Warm amber shading gives the soup an appetizing color, but the surface lacks liquid sheen and contact details, leaving ingredients looking arranged on a flat surface rather than immersed." }, "steam": { "score": 0.58, "reason": "Normal frames show subtle changes in the height, curvature and opacity of five unobtrusive wisps. They remain narrow, stroke-like columns with little spreading or diffuse breakup." }, "composition": { "score": 0.84, "reason": "A single readable dish dominates the uncluttered dark scene, with warm food colors contrasting against the cool bowl and restrained atmospheric lighting. Mobile framing is somewhat right-heavy, but no major presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushroom, leafy vegetables, corn, scallions and seaweed, with no discernible animal-derived ingredient." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and readable thickness, but mostly form parallel flat waves with little tangled volume or convincing submerged overlap." }, "toppings": { "score": 0.57, "reason": "Leaves, a shaded mushroom, tofu, scallion rings and corn provide recognizable variety. Shapes remain schematic: corn is uniform disks, tofu lacks texture, and nori resembles a rigid rectangular card." }, "bowl": { "score": 0.85, "reason": "A coherent elliptical rim joins rounded walls with consistent depth, restrained highlights and a grounding shadow. No material construction defect is observed; the ceramic finish is somewhat simplified." }, "broth": { "score": 0.52, "reason": "Warm amber shading makes the soup readable and appetizing, but the surface lacks liquid sheen and contact detail; ingredients largely appear arranged on top rather than immersed." }, "steam": { "score": 0.58, "reason": "Normal frames show subtle changes in wisp curvature, height and opacity, and the translucent strands do not obscure food. They remain stroke-like columns with little visible spreading or dissipating structure." }, "composition": { "score": 0.86, "reason": "The single dish is clearly framed against a restrained dark background, with warm food colors contrasting the cool bowl and no text or clutter. Mobile framing is slightly right-heavy, but the focal dish remains fully readable." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushroom, leafy vegetables, corn, scallions and seaweed; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends, but mostly form parallel flat waves with little interweaving, volume, or visible immersion." }, "toppings": { "score": 0.58, "reason": "Leaves, a shaded mushroom, tofu, scallion rings, and corn are readable and varied. Texture is sparse, corn resembles uniform disks, and nori is a rigid rectangular card." }, "bowl": { "score": 0.83, "reason": "Aligned elliptical rim, coherent curved walls, restrained highlights, and a grounding shadow convincingly establish a deep dark bowl. No material construction defect is observed, though the ceramic finish is minimally developed." }, "broth": { "score": 0.52, "reason": "Warm amber shading gives the soup an appetizing color, but the surface lacks liquid sheen and contact cues; ingredients largely appear laid over a flat colored ellipse." }, "steam": { "score": 0.57, "reason": "Normal frames show changes in wisp curvature, height, and opacity without obscuring food. The five narrow strokes remain rather regular and line-like, with little spreading or diffuse breakup." }, "composition": { "score": 0.84, "reason": "The single dish remains a clear focal point through warm food colors against a restrained dark background, with no text or clutter. Mobile framing places the bowl noticeably off-center to the right." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn, scallions, and seaweed; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.5, "bowl": 0.6533333333333333, "broth": 0.5666666666666667, "steam": 0.5, "composition": 0.7166666666666667 }, "reward": 0.5656666666666667, "sampleStdDev": 0.008082903768654767, "judgments": [ { "noodles": { "score": 0.5, "reason": "Wavy yellow strands are smooth and readable, but they are thin, flat, and uniform. They read as ribbons or squiggles with little overlap or volume, and no noodles lift out of the broth." }, "toppings": { "score": 0.5, "reason": "Tofu cube, bok choy leaves, shiitake mushroom, scallion rings, corn dots and a nori piece are all recognizable. They are flat vector symbols, and the corn is plain circles. The nori is a rigid dark rectangle and the tofu is blocky." }, "bowl": { "score": 0.66, "reason": "The bowl has a coherent elliptical rim, a deep hemispherical body, a soft highlight and a contact shadow. It looks floating because there is no foot or base, and the rim is a little thick and flat." }, "broth": { "score": 0.6, "reason": "The warm orange-red gradient gives the broth depth and an appetizing color, and there are subtle ripples. The surface is fairly flat, and the toppings sit on it more than in it." }, "steam": { "score": 0.5, "reason": "Five wispy lines rise, sway and fade across the frames, so there is ambient motion. They are thin, uniform outline strokes that read as mechanical curves rather than translucent wisps." }, "composition": { "score": 0.75, "reason": "The bowl is centered and is the clear focal point. The dark vignette with a warm glow is restrained, there is no clutter, and the palette is cohesive. The bowl is somewhat small in the frame." }, "vegan": { "score": 0.95, "reason": "Tofu, mushroom, bok choy, scallion, corn and nori are all plant-based. No meat or egg is visible." } }, { "noodles": { "score": 0.55, "reason": "Wavy yellow strands with some overlap and tapering read as noodles, but they are thin, flat, stroke-like ribbons that sit on the surface and look repetitive, with little volume." }, "toppings": { "score": 0.5, "reason": "Bok choy leaves, a tofu cube, a mushroom, scallion rings and a nori piece are recognizable and vegan-appropriate. The forms are simplistic and flat, the tofu is a rigid block, the corn is a row of plain dots, and the nori is a dark green rectangle." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a curved body, highlights and a contact shadow. The rim and inner lip look slightly odd, and the bowl seems to float above its shadow with no foot or base detail." }, "broth": { "score": 0.55, "reason": "The warm orange-red gradient gives depth and a glow, and there are faint ripples after the pointer interaction. It reads as flat and somewhat solid, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.5, "reason": "Five thin wavy wisps rise from the bowl and change in opacity and position across the frames, which suggests motion. They are thin line strokes that look mechanical and similar to each other, not soft spreading steam." }, "composition": { "score": 0.68, "reason": "The bowl is centered and is a clear focal point against a dark, warm vignette. Colors are harmonious and there is no clutter. The bowl is a bit small and the lower half of the frame is empty." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushroom, bok choy, scallions, corn and nori, and no animal products are visible." } }, { "noodles": { "score": 0.55, "reason": "Wavy golden strands read as noodles with some overlap and tapering, but they are thin flat ribbon-like lines with repeated wave shapes and little volume or immersion in the broth." }, "toppings": { "score": 0.5, "reason": "Tofu cube, bok choy leaves, a mushroom, scallion rings, corn dots and a nori piece are recognizable. They are simple, flat vector symbols; the corn dots look like plastic beads and the nori is a dark rectangle." }, "bowl": { "score": 0.68, "reason": "Coherent elliptical rim, deep rounded body, curved highlights and a ground shadow give a clear dark ceramic bowl. It looks slightly floating, the rim is thin and the inner wall is barely described." }, "broth": { "score": 0.55, "reason": "Warm orange-red gradient gives depth and an appetizing color, but the surface is fairly flat with little texture. Toppings sit on it without much contact or ripple, except faint ripples in the pointer frames." }, "steam": { "score": 0.5, "reason": "Five wispy curved lines rise and fade in varying phases across frames, so there is motion. They are thin, uniform stroke lines rather than soft translucent wisps, and read as mechanical." }, "composition": { "score": 0.72, "reason": "Centered bowl on a dark vignette with a warm glow, a clear focal dish, no clutter or text, and a harmonious palette. The bowl is somewhat small, with much empty space on mobile." }, "vegan": { "score": 0.95, "reason": "Tofu, mushroom, bok choy, scallions, corn, nori and noodles all appear plant-based. No meat or egg is visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.65, "bowl": 0.8166666666666667, "broth": 0.7333333333333333, "steam": 0.6833333333333333, "composition": 0.8333333333333334 }, "reward": 0.6983333333333334, "sampleStdDev": 0.025535922410074272, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are continuous smooth wavy strands in cream and orange tones with plausible bends, but they read as flat decorative squiggles spread evenly over the broth with no clumping, volume, or emergence into the liquid; the repeated similar waves look artificial rather than like a heap of noodles." }, "toppings": { "score": 0.65, "reason": "Good variety of readable stylized ingredients: leafy greens, a tofu cube, a shiitake mushroom with cap and stem, dark nori sheets, scallion rings and yellow corn-like bits. Shapes are clean but simplified; the corn kernels are fairly featureless circles and the tofu is a plain block, keeping it from higher marks." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a well-drawn elliptical rim, rim thickness ring, deep body, grounding shadow, and tasteful highlight arcs; perspective between rim and body is consistent. Minor: the foot is not shown, resting only on shadow." }, "broth": { "score": 0.7, "reason": "Appetizing orange-brown liquid with a warm radial gradient and subtle chili-oil wave lines; ingredients mostly read as sitting in the surface. The tofu cube and mushroom float slightly proud of the broth with minimal contact shadow, and the surface is quite smooth/flat." }, "steam": { "score": 0.65, "reason": "Translucent irregular wisps rise from the rim area and change position, height and curvature across the time-ordered desktop frames (e.g., desktop-1 vs desktop-2 vs pointer frames), so clear animation is supported. Wisps are somewhat thin and line-like rather than soft volumetric vapor, and never obscure the food." }, "composition": { "score": 0.8, "reason": "Balanced centered dish against a dark atmospheric background with a warm halo, coherent lighting and no clutter, text or competing elements; the bowl is the clear focal point. Empty dark space above and below is intentional atmosphere but leaves the frame a bit sparse." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are greens, tofu, mushroom, nori, scallion and corn-like kernels in plant broth; no meat, fish, egg or dairy forms are shown." } }, { "noodles": { "score": 0.55, "reason": "Golden wavy strands are readable and consistently placed across the broth, but they render as flat, uniform squiggle ribbons lying on the surface with little volume, overlap depth, or immersion." }, "toppings": { "score": 0.6, "reason": "Good variety: bok choy leaves, shiitake mushroom with stem and gill shading, scallion rings, corn dots, chili slices and a nori sheet. Weakest are the tofu, drawn like an open cardboard box with flaps, and the nori, a plain rigid rectangle; shapes are simple flat symbols." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim with visible thickness, consistent hemispherical body, subtle curvature highlights, plausible broth-to-rim join, and a soft contact shadow grounding it. Minor: no distinct foot ring." }, "broth": { "score": 0.75, "reason": "Warm orange-brown gradient surface with darker meniscus edge reads as liquid with depth; ingredients sit at the surface with small occlusion shadows. Slightly shallow immersion\u2014noodles and toppings look laid onto the broth rather than nestled in it." }, "steam": { "score": 0.65, "reason": "Thin translucent wisps rise from the soup and change shape/height across the normal frames (curlier by desktop-3, taller and brighter in pointer frames), showing real animation without obscuring food; slightly schematic line-like wisps rather than full volumetric steam. Static in reduced frames as expected." }, "composition": { "score": 0.85, "reason": "Centered dish as clear focal point against a dark atmospheric backdrop with a restrained warm glow and ground shadow; no text, UI or clutter; balanced in both desktop and mobile viewports with no overflow." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushroom, bok choy, scallion, corn, chili and nori in plant-looking broth; no meat, fish, egg or dairy forms appear." } }, { "noodles": { "score": 0.6, "reason": "Continuous wavy strands in two tones lie believably in the broth, but they are flat surface squiggles with uniform amplitude and a repetitive, decorative look rather than tangled strands with volume." }, "toppings": { "score": 0.7, "reason": "Good variety and readability: bok choy leaves with pale midribs, a shaded shiitake cap with stem, a tofu cube, scallion rounds, chili slivers and a standing nori sheet; all recognizable, though the tofu box and scallion discs are simple flat shapes." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim, inner cavity, tapered walls and rounded base with soft highlights and a ground shadow; perspective is consistent between rim ellipse and body with no broken joins." }, "broth": { "score": 0.75, "reason": "Warm orange broth with radial glow, restrained sheen and darker edge depth; mushroom stem and nori appear genuinely immersed, though the surface is fairly flat with little texture beyond a few swirls." }, "steam": { "score": 0.75, "reason": "Thin translucent wisps rise above the bowl and clearly change shape and position across the time-ordered desktop frames (1006/2369/3724 ms), indicating continuous rising motion; they are sparse and line-like rather than fully volumetric." }, "composition": { "score": 0.85, "reason": "Centered focal dish in a dark atmospheric vignette with warm glow, balanced margins, no clutter or competing elements; mobile crop keeps the bowl readable. Background glow is tasteful and does not distract from the food." }, "vegan": { "score": 1.0, "reason": "Visible toppings are greens, tofu, mushroom, scallion, chili and nori over plant broth; no egg, meat, seafood or dairy appears." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.699999999999818, "frameIntervalP95Ms": 16.799999999999272, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "92e95ce2ca7c3293fcfcb36c31b62f40239106c398a652ced31b146d30d13a87", "gradedAt": "2026-09-30T23:17:37.995255+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.3655555555555556, "toppings": 0.4, "bowl": 0.6766666666666666, "broth": 0.5133333333333333, "steam": 0.5655555555555556, "composition": 0.7255555555555555 }, "reward": 0.5171666666666667, "runId": "openai/gpt-5.6-luna/medium", "rewardStdDev": 0.07696120956545431, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.28, "toppings": 0.4, "bowl": 0.68, "broth": 0.4066666666666667, "steam": 0.48, "composition": 0.7333333333333333 }, "reward": 0.47800000000000004, "sampleStdDev": 0.001732050807568879, "judgments": [ { "noodles": { "score": 0.28, "reason": "Yellow strands are visible, but mostly form straight crossing rods with exposed ends extending beyond the rim; supple bends, immersed volume and coherent tangles are missing." }, "toppings": { "score": 0.4, "reason": "Green vegetable clusters, a tan mushroom-like shape and a rectangular tofu-like piece provide variety, but rounded featureless shapes dominate, with little leaf structure or ingredient texture." }, "bowl": { "score": 0.68, "reason": "The elliptical rim and rounded shaded body form a coherent vessel with readable depth. The large hard-edged oval highlight looks artificial, and the base has weak grounding." }, "broth": { "score": 0.42, "reason": "Warm orange color and scattered garnish marks suggest soup, but nested flat ellipses make the surface disc-like, with little sheen or convincing ingredient immersion." }, "steam": { "score": 0.48, "reason": "Translucent plumes emerge near the soup and visibly change curvature across normal frames. Their evenly spaced, broad ribbon shapes and abrupt kinks look mechanical rather than dispersing wisps." }, "composition": { "score": 0.74, "reason": "A single readable bowl, warm colors and restrained dark background create a clear focal scene without text or clutter. Oversized tall steam occupies much of the desktop frame, while mobile leaves substantial empty space." }, "vegan": { "score": 0.93, "reason": "Visible toppings plausibly depict vegetables, tofu and a mushroom alongside noodles; no clearly animal-derived ingredient is apparent, though simplified shapes limit certainty." } }, { "noodles": { "score": 0.28, "reason": "Pale continuous strokes are visible, but they form mostly straight crisscrossing sticks with exposed ends beyond the rim, lacking supple bends, volume, and believable placement." }, "toppings": { "score": 0.4, "reason": "Green vegetables, a tan mushroom-like cap, and a yellow tofu-like block provide variety, but rounded featureless shapes dominate; leaf structure and ingredient textures are largely absent." }, "bowl": { "score": 0.68, "reason": "The elliptical rim and rounded body form a coherent vessel with warm shading. The oversized oval highlight looks pasted on, and the base lacks convincing surface contact." }, "broth": { "score": 0.4, "reason": "Warm orange color and small garnish flecks suggest soup, but nested flat ellipses resemble solid discs, with little sheen, depth, or convincing ingredient immersion." }, "steam": { "score": 0.48, "reason": "Translucent columns originate near the soup and visibly change bends across normal frames. Their broad, similarly shaped ribbons and abrupt kinks look mechanical rather than dispersing irregular wisps." }, "composition": { "score": 0.73, "reason": "A single readable dish stands out against a restrained dark background without text or UI. Warm and green accents are balanced, though the tall steam occupies disproportionate space and the mobile dish is relatively small." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly represent vegetables, tofu, and mushroom alongside noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.28, "reason": "Pale continuous strokes suggest noodles, but nearly straight crisscrossing strands protrude far beyond the rim and lack supple bends, volume, or believable immersion." }, "toppings": { "score": 0.4, "reason": "Green vegetables, a tan mushroom-like shape, and yellow blocks provide variety, but rounded capsules and featureless geometric pieces dominate; leaf structure and mushroom texture are poorly resolved." }, "bowl": { "score": 0.68, "reason": "The elliptical rim and rounded body form a coherent vessel with warm shading and a soft grounding shadow. The oversized flat oval highlight and minimally defined base weaken the ceramic finish." }, "broth": { "score": 0.4, "reason": "The warm orange surface reads as soup, with scattered garnish flecks, but nested flat ellipses and little liquid sheen or contact detail make ingredients appear laid on top." }, "steam": { "score": 0.48, "reason": "Translucent plumes emerge near the soup and visibly change curvature across normal frames. However, uniformly thick parallel ribbons and repeated sharp bends look mechanical rather than spreading, irregular wisps." }, "composition": { "score": 0.73, "reason": "A single readable bowl stands out against a restrained dark background without text or UI clutter. Warm and green accents balance well, though the very tall steam dominates the desktop framing and mobile leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible noodle strokes, green vegetables, mushroom-like topping, and tofu-like blocks plausibly depict vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 0.9766666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.3333333333333333, "bowl": 0.5666666666666667, "broth": 0.48333333333333334, "steam": 0.5333333333333333, "composition": 0.6266666666666667 }, "reward": 0.4676666666666667, "sampleStdDev": 0.025643387711714944, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin uniform tan strokes that poke out past the rim and lie on top of the broth like wires. They look rigid and flat, with no volume or overlap, and they don't sit in the soup convincingly." }, "toppings": { "score": 0.35, "reason": "The toppings are two green rounded blobs that look like plastic capsules, a flat tofu block, and a tan dome mushroom with a red curve. They are generic, lack texture, and have no nori or readable leaf detail." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a tapered body with a gradient, a soft shadow and a highlight. The rim is thin and the body looks like a flat vector shape. The noodles extending beyond the rim make the bowl look less solid." }, "broth": { "score": 0.5, "reason": "The broth is an orange-red ellipse with a slight gradient and a few specks. It reads as a liquid surface but is flat and lacks depth. The toppings look pasted on top rather than immersed." }, "steam": { "score": 0.6, "reason": "Seven translucent wavy plumes rise and fade, and they shift between frames. They are similar in shape, ribbon-like and mechanical, but they do not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered on a dark background with a warm glow, so the focal point is clear. The colors are pleasing and there is no clutter. The steam is slightly uniform and the food is small." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are noodles, greens, tofu and a mushroom, with no meat or egg. The appearance is plausibly vegan." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin uniform cream strokes laid across the bowl. A few long strands stick out past the rim like wires. They are rigid, flat and lack volume, with no sense of sitting in the broth." }, "toppings": { "score": 0.35, "reason": "Two green blobs look like rounded capsule shapes, not readable bok choy. A tofu block is a flat rectangle, and the mushroom is a simple half-dome. A red chili curve is the only other element. The shapes are generic and lack texture, and there is no nori." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent elliptical rim, a tapered body with a gradient and a soft shadow. The highlight is a flat oval that looks pasted on. The rim is thin and the base is undefined. Noodles poking out past the rim break the contact." }, "broth": { "score": 0.5, "reason": "The broth is an orange-red gradient with a subtle sheen and a dark inner rim, so it reads as soup. It is fairly flat, with little depth. Toppings sit on top more than in it." }, "steam": { "score": 0.5, "reason": "Seven translucent wavy plumes rise and fade, and their shapes shift across frames, so there is motion. They look like uniform, similar ribbons and are somewhat mechanical. They do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark background with a warm glow, and the food is the focus. The palette is cohesive and there is no clutter. The scene is simple and somewhat flat, and the bowl sits slightly low with plenty of empty space." }, "vegan": { "score": 0.85, "reason": "The toppings are greens, tofu, a mushroom and a chili, with noodles and a red broth. There are no visible animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, smooth, uniform strands that extend straight past the bowl rim like wires. They are flat and sparse, with little overlap or volume, and they sit on the surface rather than in the broth." }, "toppings": { "score": 0.3, "reason": "The toppings are two green rounded blobs (bok choy?), a flat tofu block and a tan dome that could be a mushroom. They are generic and smooth, with no texture or detail, and there is no readable nori. They read as plastic-like rounded shapes." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent elliptical rim, a tapered body with a gradient and a soft shadow. However, the noodles cross the rim and the highlight is a flat oval blob. The base is undefined and the bowl reads as simple vector art." }, "broth": { "score": 0.45, "reason": "The broth is an orange-red surface with a slight gradient and a few oil specks. It is flat and reads as concentric discs. The toppings look pasted on top rather than immersed." }, "steam": { "score": 0.5, "reason": "Six wavy translucent plumes rise and shift between frames, so there is motion. They are similar, regular, ribbon-like and mechanical. They do fade upward, and they do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark background with a warm glow, so the focus is clear and the colors are pleasing. The scene is sparse, the steam dominates the frame, and there is no clutter." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are noodles, greens, tofu and a mushroom-like dome, with no meat or egg. The broth is plausibly plant-based." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.4166666666666667, "toppings": 0.4666666666666667, "bowl": 0.7833333333333333, "broth": 0.65, "steam": 0.6833333333333333, "composition": 0.8166666666666667 }, "reward": 0.6058333333333333, "sampleStdDev": 0.044040700872412714, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are uniform flat pale ribbons with minimal bend; several strands cross the rim and float unsupported beyond the bowl on both sides. They read as rigid straws lying on the surface rather than supple strands in broth." }, "toppings": { "score": 0.45, "reason": "Two identical mirrored glossy green blobs, a plain beige dome, a yellow block and a thin red strand are recognizable as stylized ingredients, but the green forms look like plastic spheres and shapes are featureless and repeated rather than organic." }, "bowl": { "score": 0.75, "reason": "Coherent flat-design bowl: elliptical rim with dark inner ring, smooth curved walls, warm gradient, soft highlight and a grounded contact shadow; perspective of rim and body agree. Only a slightly pasted-on oval highlight as a minor flaw." }, "broth": { "score": 0.6, "reason": "Orange-red liquid surface with a darker inner ring and scattered specks reads as broth, but it is a flat ellipse; toppings and noodles sit on top with little visible immersion or occlusion." }, "steam": { "score": 0.55, "reason": "Wavy plumes clearly morph between the timed frames (shapes and bends differ across desktop-1/2/3 and pointer frames), so motion is real. However they are many near-identical dark gray ribbon plumes, more like repeated silhouettes than translucent irregular wisps." }, "composition": { "score": 0.8, "reason": "Centered bowl on a clean dark backdrop with a warm glow behind it; the dish is the clear focal point, colors are attractive, no clutter or UI. Flat-style restraint is coherent throughout, mobile framing also stays readable." }, "vegan": { "score": 0.75, "reason": "Visible items (tofu-like block, green vegetables/scallion-like pieces, red chili strand, beige dome) are plausibly plant-based; the beige dome is ambiguous but does not clearly resemble meat, fish or egg." } }, { "noodles": { "score": 0.4, "reason": "Noodles read as thin rigid pale-yellow sticks: several strands run straight off the rim and hang unsupported in mid-air past the bowl edge, with minimal curvature. Wire-like strands dominate the scene, capping this at the rigid-strand limit." }, "toppings": { "score": 0.4, "reason": "Toppings are dominated by generic glossy forms: two clusters of shiny green capsule blobs, a featureless tan dome, and a plain yellow block. A red curl and tiny green dots add slight variety, but the plastic-looking shapes dominate, so the .50 cap applies." }, "bowl": { "score": 0.8, "reason": "Coherent flat-design bowl: a clean orange-rimmed ellipse, smooth terracotta body with a soft highlight, rounded base and a believable contact shadow. Rim, cavity and body join consistently with no visible distortion; only minor simplicity keeps it from higher." }, "broth": { "score": 0.7, "reason": "Warm red-orange surface with subtle gradient, lighter streaks and scattered specks; toppings and noodles sit on/in the surface plausibly. Somewhat flat with restrained texture, and the strands crossing the rim slightly undercut immersion, but it reads as liquid." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps rise from the bowl without hiding the food, and shapes/positions clearly change across desktop-1, -2, -3 and pointer frames, confirming animated motion. Docked because the plumes are fairly uniform repeated S-curves, reading slightly mechanical." }, "composition": { "score": 0.85, "reason": "Balanced dark scene: warm glow behind a centered focal bowl, subtle star-like specks, coherent warm palette, no text or clutter, and the mobile frame keeps the dish well placed. Steam fills the upper space tastefully; presentation is polished with minor stiffness in the food itself." }, "vegan": { "score": 0.85, "reason": "Visible toppings are green vegetable clusters, a yellow tofu-like block, herbs and broth; no meat, egg, fish or dairy shapes appear. The tan dome with a red curl is ambiguous but plausibly a dumpling/mushroom, so near-full credit with slight deduction for ambiguity." } }, { "noodles": { "score": 0.45, "reason": "Noodles are uniform flat cream-colored ribbons with shallow bends; several strands drape straight across the rim and extend horizontally beyond the bowl edges into empty space, which is physically implausible, and the strands read as rigid strips rather than supple overlapping noodles." }, "toppings": { "score": 0.55, "reason": "Some variety exists (paired green vegetable blobs, a yellow tofu block, a tan dome, red chili curl, scallion specks), but the green clumps look like glossy plastic blobs and the tan dome is featureless and ambiguous; forms are generic flat shapes rather than convincing ingredients." }, "bowl": { "score": 0.8, "reason": "Coherent flat-vector bowl: elliptical rim with outline, visible interior cavity and depth where broth sits below the rim, smooth tapered body, soft body highlight and contact shadow. Minor shortcoming: the large soft highlight ellipse on the body reads slightly detached." }, "broth": { "score": 0.65, "reason": "Red-orange surface with tonal swirls, sheen, ripples and small specks reads as liquid; toppings and noodles mostly sit on the surface. Deduction: noodles lie atop rather than immersing, and the ring patterns around toppings are somewhat flat and target-like." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the bowl without hiding food, and their wave phases/positions visibly shift between the time-ordered desktop frames, supporting continuous rising/swaying motion. Slight mechanical sameness among the repeated S-curve plumes." }, "composition": { "score": 0.8, "reason": "Balanced dark scene with warm glow behind the bowl, subtle star specks, no text or clutter, and the dish remains the focal point. Mobile frame leaves the bowl small with large empty space, a minor imbalance." }, "vegan": { "score": 0.8, "reason": "Visible ingredients (greens, tofu block, chili, noodles, specks) are plant-based; the ambiguous tan dome has no identifying animal features but its identity is unclear, warranting a small deduction rather than full credit." } } ], "diagnostics": { "vegan": 0.8 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8755555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 75.0, "frameIntervalP95Ms": 83.30000000000109, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "9678a9602116ab3d409af5a958c4b2e74d58da9d7479c3d3cbb1771fe5dd5be8", "gradedAt": "2026-09-30T23:16:34.925681+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4033333333333333, "toppings": 0.4311111111111111, "bowl": 0.6288888888888888, "broth": 0.39999999999999997, "steam": 0.5744444444444444, "composition": 0.6744444444444445 }, "reward": 0.5078333333333334, "runId": "openai/gpt-5.6-luna/xhigh", "rewardStdDev": 0.08957879833482427, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.42666666666666664, "toppings": 0.41000000000000003, "bowl": 0.72, "broth": 0.3, "steam": 0.59, "composition": 0.7733333333333333 }, "reward": 0.5276666666666666, "sampleStdDev": 0.008962886439832509, "judgments": [ { "noodles": { "score": 0.44, "reason": "Golden strands are recognizable, but form a flat crisscrossed ribbon lattice with little supple curvature or convincing piled volume." }, "toppings": { "score": 0.43, "reason": "Mushroom caps and stems are readable, but the dominant rounded rectangular tofu cards and oversized scalloped leaf lack organic texture and detail; green garnishes are rigid rods." }, "bowl": { "score": 0.72, "reason": "Elliptical rim and rounded walls establish a coherent vessel with warm shading and visible thickness. Broad stripe-like highlights and an indistinct base weaken the ceramic finish and contact." }, "broth": { "score": 0.3, "reason": "An orange surface is visible beneath the noodles, but looks flat and solid, without liquid sheen or depth. Several toppings appear pasted across the exterior rather than immersed." }, "steam": { "score": 0.61, "reason": "Translucent curved plumes change shape across normal frames and leave the food readable. Their uniformly thick, rounded strokes look more like animated smoke symbols than spreading, fading wisps." }, "composition": { "score": 0.77, "reason": "A single readable dish anchors a restrained dark scene, with harmonious warm browns and green accents and no distracting UI. The low placement leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushroom forms and tofu-like rectangles plausibly depict vegan ramen; no animal-derived topping is evident." } }, { "noodles": { "score": 0.42, "reason": "Golden strands are recognizable, but mostly form a flat crisscross of broad ribbons with little supple curvature, coherent overlap or volume." }, "toppings": { "score": 0.4, "reason": "Mushroom caps and stems are readable, but oversized scalloped leaves lack organic detail, tofu resembles rounded orange cards, and green garnishes are rigid straight sticks." }, "bowl": { "score": 0.72, "reason": "The elliptical rim and rounded body have coherent perspective and warm dimensional shading. Broad stripe-like highlights and an indistinct base weaken the ceramic finish and support." }, "broth": { "score": 0.3, "reason": "An orange soup surface is visible, but it lacks liquid sheen and depth. Several toppings appear pasted across the exterior wall rather than immersed, with rim lines visibly crossing them." }, "steam": { "score": 0.58, "reason": "Normal frames show changing translucent curls above the soup without obscuring it. The wisps remain thick, uniformly stroked brown squiggles with blunt ends rather than spreading, delicately fading vapor." }, "composition": { "score": 0.78, "reason": "A single readable dish anchors a restrained dark scene with harmonious warm browns and greens, no distracting text, and a mobile layout that preserves the focal bowl. Large upper empty space and heavy concentric ground shadows slightly weaken balance." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms and tofu-like rectangles plausibly depict vegan ramen; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.42, "reason": "Golden strands are recognizable, but form a flattened crisscross lattice with little supple curvature, coherent overlap or noodle volume." }, "toppings": { "score": 0.4, "reason": "Mushroom caps and stems are readable, but the oversized scalloped leaf lacks organic detail, tofu resembles rounded glossy cards, and greens are rigid straight sticks." }, "bowl": { "score": 0.72, "reason": "The elliptical rim and rounded body form a coherent vessel with warm shading and grounded support. Thick outlined bands and broad stripe-like highlights make the material somewhat artificial." }, "broth": { "score": 0.3, "reason": "An orange soup surface is visible, but it lacks liquid sheen and depth; several toppings appear pasted over the front wall rather than immersed, weakening contact and occlusion." }, "steam": { "score": 0.58, "reason": "Normal frames show changing translucent curved plumes above the soup without obscuring it. Their thick, uniformly rounded strokes look like animated smoke symbols rather than spreading, fading wisps." }, "composition": { "score": 0.77, "reason": "A single readable dish anchors a restrained dark scene with harmonious amber and green colors and no clutter. The low placement leaves considerable empty space above, and the lighting is fairly graphic." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, tofu-like rectangles and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3333333333333333, "toppings": 0.38333333333333336, "bowl": 0.45, "broth": 0.3333333333333333, "steam": 0.48333333333333334, "composition": 0.5166666666666667 }, "reward": 0.41000000000000003, "sampleStdDev": 0.009013878188659959, "judgments": [ { "noodles": { "score": 0.35, "reason": "Noodles are a thin yellow mass at the rim with brown scribble arcs on top. They read as wire-like lines rather than supple strands, and the bowl hides most of them." }, "toppings": { "score": 0.35, "reason": "A green leaf, two tofu blocks, small mushrooms and thin green sticks are recognizable. The tofu is a rounded rigid card and the green sticks are flat lines. The leaf is large and detached, floating in front of the rim, and there is no nori." }, "bowl": { "score": 0.45, "reason": "The bowl has a rim, a tapering body and a glossy highlight curve. The dark translucent ring across the rim looks odd, and the toppings overlap the front wall, which weakens the sense of depth. The base contact with the shadow is acceptable." }, "broth": { "score": 0.3, "reason": "Only a thin orange sliver of broth is visible and the noodles cover most of it. The dark ring obscures the surface, and the toppings look pasted in front of the bowl rather than immersed." }, "steam": { "score": 0.5, "reason": "Faint squiggly wisps rise and change shape across frames, so motion is present. They look like mechanical curly strokes and are very dim, but they do not hide the food." }, "composition": { "score": 0.55, "reason": "The bowl is centered, the dark background is restrained and the warm palette is coherent. The toppings overlap awkwardly and the food is small, but nothing distracting competes for attention." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, greens and noodles are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.35, "reason": "Noodles are a thin band of flat yellow strands with brown scribble arcs at the surface. They read as a maze or wire pattern, with no volume or supple bends, and are mostly hidden by the toppings." }, "toppings": { "score": 0.4, "reason": "A green leaf with a pale midrib is recognizable, and there are small mushrooms. The tofu pieces are rounded tan cards, and the scallion lines are rigid straight sticks. Toppings float above the rim and look pasted on, so the generic card forms limit the score." }, "bowl": { "score": 0.45, "reason": "The bowl is a smooth brown gradient with a rim ellipse and a plausible base shadow. The dark inner rim band looks like a disc cutting across the toppings, and the highlight curves are odd. The toppings overlap the rim inconsistently, and the bowl looks flat." }, "broth": { "score": 0.35, "reason": "Only a thin orange sliver of broth is visible, mostly covered by noodles and the rim band. It has little sheen or depth, and the toppings do not sit in it." }, "steam": { "score": 0.5, "reason": "Dark, squiggly wisps change shape across frames and drift upward, so there is motion. They are low-contrast, look like scribbled strokes and are somewhat mechanical, but they do not hide the food." }, "composition": { "score": 0.5, "reason": "The bowl is centered in a dark, moody scene with warm colors and no clutter. The toppings overflow awkwardly, and the food lacks convincing detail." }, "vegan": { "score": 0.9, "reason": "The toppings are tofu, leaf greens, mushrooms and scallion, with noodles. No animal ingredients are visible." } }, { "noodles": { "score": 0.3, "reason": "Noodles are a thin band of yellow strands with flat, repeated arcs of brown scribble lines. They are mostly hidden by the rim and read as a maze-like mass rather than supple strands." }, "toppings": { "score": 0.4, "reason": "A green leaf is readable with a vein, and there are small mushrooms. The tofu pieces are rounded rigid cards, and the green scallion lines are straight sticks. Toppings float above the bowl and overlap the rim unnaturally." }, "bowl": { "score": 0.45, "reason": "The bowl has a believable brown glazed body with a highlight curve and a shadowed base. The dark rim ring is oddly translucent and is overlapped by the leaf and tofu, so the rim does not read as solid. The bowl also seems to float over layered shadow ellipses." }, "broth": { "score": 0.35, "reason": "Only a thin orange sliver of broth is visible, mostly covered by noodles and the rim. It has little depth or sheen, and toppings are pasted above it instead of sitting in it." }, "steam": { "score": 0.45, "reason": "Squiggly translucent wisps change shape between frames and drift upward. They are faint, look like stroked curly lines, and are somewhat mechanical." }, "composition": { "score": 0.5, "reason": "The bowl is centered in a dark, moody scene with warm colors, and there is no clutter. The toppings overlapping the rim look awkward, and the bowl is small and low-contrast." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are tofu, mushrooms, a leaf, scallion and noodles, and no animal products appear." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.45, "toppings": 0.5, "bowl": 0.7166666666666667, "broth": 0.5666666666666667, "steam": 0.65, "composition": 0.7333333333333334 }, "reward": 0.5858333333333334, "sampleStdDev": 0.012829003598617276, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are barely visible: a flat golden patch under the broth with thin cross-hatch lines suggesting strands. No supple individual strands or volume; reads as an indistinct noodle mass, only partially occluded by toppings." }, "toppings": { "score": 0.5, "reason": "A green leafy vegetable and small mushrooms are recognizable, but the tofu/naruto stand-ins are plain rounded rectangles with glossy highlights, reading as rigid cards rather than food. One rectangle floats partially above the rim, and greens read as a single flat blob with a pale center vein." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, rounded body, warm ceramic gradient, inner rim thickness and a soft ground shadow; contact with the surface is believable. Shortcomings: an odd bright arc stroke on the body reads as an unexplained line, and the interior/rim junction with the soup surface is slightly ambiguous." }, "broth": { "score": 0.55, "reason": "The broth has warm orange tone and subtle sheen, but it reads as a flat amber ellipse; noodles and toppings sit on top of the surface rather than showing immersion, occlusion or depth, and the left side is largely hidden behind the leaf." }, "steam": { "score": 0.7, "reason": "Translucent gray squiggle wisps rise from the soup and change shape and position across desktop frames 1-3, indicating continuous motion without opaque fog. They are thin, sparse curly lines, slightly disconnected from the surface and stylized, but clearly animated." }, "composition": { "score": 0.6, "reason": "Dark atmospheric scene with warm rim lighting, soft vignette and shadow, and a single readable focal bowl in both desktop and mobile crops. However the leaf and green sticks spill over the rim awkwardly, the overall scene is very sparse, and the flat dark background offers little depth." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are leafy greens, mushrooms, tofu-like blocks, noodles and broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.45, "reason": "Noodles appear as a flat yellow bed with thin brown crosshatch scribble lines over it rather than supple continuous strands; no strand separation, bends, or volume \u2014 reads as a stylized noodle sheet with maze-like texture lines." }, "toppings": { "score": 0.45, "reason": "Mushrooms (brown caps with stems) and a green leafy blob read acceptably, but two large orange rounded-rectangle tofu blocks are featureless rigid cards dominating the dish, and the long green sticks crossing the bowl are ambiguous. Generic card-like forms dominate, capped at .50." }, "bowl": { "score": 0.75, "reason": "Coherent stylized bowl: elliptical rim, smooth rounded walls, subtle left highlight, bottom curvature arc, and grounded shadow ellipse. Perspective is consistent; only the rim band is thick and graphic, a stylistic choice rather than broken geometry." }, "broth": { "score": 0.6, "reason": "Orange elliptical broth surface with gradient and darker edge reads as liquid, and mushrooms/noodles sit in it. However, the right tofu card overlaps the rim line and looks pasted above the broth, and the surface is fairly flat with little depth." }, "steam": { "score": 0.6, "reason": "Thin gray curl wisps above the bowl clearly change shape and position across the time-ordered desktop frames, so steam is animated and non-opaque. But the squiggles are repeated S-curl strokes that look mechanical and do not visibly rise, spread, or fade." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a dark atmospheric scene with warm rim lighting, soft radial glow, restrained palette, and no clutter or text. Presentation is balanced, though food detailing is too simple for top marks." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are leafy greens, tofu, mushrooms, noodles and plant broth; no meat, fish, egg, or dairy forms appear." } }, { "noodles": { "score": 0.45, "reason": "Noodles are a flat yellow patch of squiggly strokes with repeated cross-hatch loops lying on the surface plane; no strand volume, dips, or coherent overlap \u2014 reads as an indistinct scribbled mass." }, "toppings": { "score": 0.55, "reason": "A large scalloped green leaf and two shiitake-style mushrooms read as food, but the two tofu pieces are glossy rigid rounded cards, and the scallion strips are simple straight sticks; card-like forms are prominent." }, "bowl": { "score": 0.7, "reason": "Coherent hemispherical ceramic bowl with elliptical rim, consistent wall curve, gradient shading, highlight streak and a grounding shadow; no visible foot/base and the rim band is a bit heavy, but no broken geometry." }, "broth": { "score": 0.55, "reason": "Amber surface has warm color and soft sheen, but the leaf and tofu cards spill over and past the rim, appearing pasted above the liquid rather than sitting in it; immersion/occlusion is weak." }, "steam": { "score": 0.65, "reason": "Thin curly wisps clearly change shape and position across the time-ordered desktop frames, rising above the bowl without obscuring food; however they are faint dark S-strokes that read more as stylized curls than translucent vapor." }, "composition": { "score": 0.8, "reason": "Clean dark atmospheric stage with warm glow, centered focal bowl, cohesive amber palette and no clutter or UI; minor imbalance from toppings crowding the upper-left and sparse right side." }, "vegan": { "score": 1.0, "reason": "Visible toppings are greens, tofu-like blocks, mushrooms, scallion strips, noodles and broth; no animal-derived ingredients evident." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9666666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 33.400000000001455, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "dfa4b42560c5c5633a5275b68dcd34b9eb22193cb65410459f5511676c51a1cb", "gradedAt": "2026-09-30T23:08:12.687083+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4288888888888889, "toppings": 0.5344444444444445, "bowl": 0.7533333333333334, "broth": 0.5655555555555556, "steam": 0.6022222222222222, "composition": 0.7122222222222222 }, "reward": 0.5881666666666667, "runId": "openai/gpt-5.6-sol/high", "rewardStdDev": 0.05320583719020977, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.42, "toppings": 0.5533333333333333, "bowl": 0.78, "broth": 0.48, "steam": 0.5733333333333333, "composition": 0.7166666666666667 }, "reward": 0.5840000000000001, "sampleStdDev": 0.0036055512754639926, "judgments": [ { "noodles": { "score": 0.42, "reason": "Thin golden strands are visible beneath the toppings, but form a nearly parallel flat ribbon bundle with little supple bending, overlap, or volume." }, "toppings": { "score": 0.56, "reason": "Green leaves with midribs, scored brown mushroom caps, tofu-like pieces, and scallion rings provide readable variety. Shapes remain simplified and smooth, and the upright nori resembles a rigid striped card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls, and foot form a coherent supported bowl, with restrained highlights suggesting dark ceramic. The cavity and foot junction are somewhat shallow and indistinct." }, "broth": { "score": 0.48, "reason": "Warm brown shading and scattered oil-like dots suggest broth, but the surface lacks convincing liquid sheen or depth; ingredients appear laid on top with minimal immersion or contact detail." }, "steam": { "score": 0.58, "reason": "Faint translucent trails change curvature across the normal frames and leave the food visible. They remain similarly spaced vertical filaments, with weak connection to the soup and little visible spreading or dissipation." }, "composition": { "score": 0.73, "reason": "A single uncluttered dish and restrained warm-green palette suit the dark atmosphere. The large dark bowl body outweighs the small food cluster, and desktop framing crowds the foot against the bottom edge." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, scallions, and seaweed alongside noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.42, "reason": "Thin golden strands are visible, but form a nearly parallel flat band with little curling, overlap, or convincing noodle volume." }, "toppings": { "score": 0.55, "reason": "Green leaves have readable midribs, mushroom caps have scoring, and tofu has small speckles. Shapes remain simplified and smooth; the upright nori resembles a rigid striped card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls, and low foot form a coherent vessel with restrained highlights. Very dark shading weakens material definition and makes the foot junction difficult to read." }, "broth": { "score": 0.48, "reason": "Warm brown color and scattered oil-like dots suggest soup, but the surface lacks liquid sheen and depth; ingredients largely appear laid onto a flat surface." }, "steam": { "score": 0.57, "reason": "Faint translucent trails change curvature across normal frames and leave the food visible. They remain evenly spaced, narrow vertical columns with weak connection to the soup and little evident spreading." }, "composition": { "score": 0.72, "reason": "A single uncluttered dish stands against a restrained dark background with warm illumination. The desktop bowl sits very low, nearly touching the frame bottom, while its large dark body outweighs the small food area." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, scallion rings, and seaweed; no visibly animal-derived ingredients are present." } }, { "noodles": { "score": 0.42, "reason": "Thin golden strands are visible beneath the toppings, but form a nearly parallel flat ribbon bundle with little supple bending, overlap, or volume." }, "toppings": { "score": 0.55, "reason": "Green leaves with central veins, scored mushroom caps, tofu-like pieces, and scallion rings provide readable variety. Shapes remain smooth and simplified, while the upright nori resembles a rigid striped card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls, and centered foot form a coherent supported vessel with restrained highlights. The very dark body loses material detail, and the foot connection is indistinct." }, "broth": { "score": 0.48, "reason": "Warm brown color and scattered oil-like flecks suggest soup, but the surface lacks liquid sheen and depth; ingredients appear laid over it with little visible immersion or contact." }, "steam": { "score": 0.57, "reason": "Faint translucent trails change curvature across the normal frames without obscuring food. They remain long, similarly spaced vertical lines, with little visible spreading and a gap above the soup." }, "composition": { "score": 0.7, "reason": "A single uncluttered bowl and restrained warm-green palette establish a clear focus against the dark background. Desktop framing places the base almost against the bottom edge, and the oversized bowl body outweighs the small food cluster." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy greens, scallions, and seaweed with noodles; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.5166666666666667, "bowl": 0.68, "broth": 0.55, "steam": 0.5, "composition": 0.62 }, "reward": 0.5371666666666667, "sampleStdDev": 0.007216878364870361, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles appear as a small pale flat band of thin parallel lines in the center, mostly hidden under toppings. They lack volume, bends and overlap, and read as a rigid ribbon patch." }, "toppings": { "score": 0.5, "reason": "Green leaves with veins, shiitake-like mushrooms, tofu cubes, scallion rings and a nori sheet are recognizable. They are small and flat, the nori is a rigid dark rectangle, and the red threads are odd, so they read as symbols." }, "bowl": { "score": 0.68, "reason": "The dark bowl has a coherent elliptical rim, curved walls, a highlight and a foot ring with a soft shadow. The material is matte and slightly flat, and the foot looks a bit detached, but there is no major distortion." }, "broth": { "score": 0.55, "reason": "The broth is a warm brown with chili-oil specks and a dark vignette at the edge. It reads as liquid but is fairly flat, with weak sheen and limited immersion or contact with the ingredients." }, "steam": { "score": 0.5, "reason": "Faint thin wisps rise and shift slightly between frames, which is subtle and translucent. They are very faint and somewhat mechanical vertical lines, but they do not hide the food." }, "composition": { "score": 0.62, "reason": "The bowl is centered in a dark, moody scene with a warm glow and is clearly the focus. The food is small relative to the bowl, and the nori floats awkwardly behind the rim, but there is no clutter." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, greens, scallions, seaweed and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are a small pale patch of thin parallel strands at the center, mostly hidden under toppings. They read as a flat block with no supple bends or volume, and only a few red chili threads add detail." }, "toppings": { "score": 0.55, "reason": "Bok choy leaves with veins, shiitake-like mushrooms, tofu cubes, scallion rings and a nori sheet are all recognizable. Shapes are simple and flat, the tofu looks like tan cards, and the nori is a rigid dark rectangle standing upright. Toppings are small relative to the bowl." }, "bowl": { "score": 0.68, "reason": "A dark ceramic bowl with a coherent elliptical rim, a curved body, a subtle highlight and a foot base. Perspective is consistent and the material reads well. The body is very large and dark, and the base looks a little flat." }, "broth": { "score": 0.55, "reason": "The broth is a warm brown surface with chili oil specks and a gradient, so it reads as a liquid. It looks fairly flat and matte, with little depth, and the toppings sit on top more than in it." }, "steam": { "score": 0.5, "reason": "Faint, thin wisps rise and shift position across frames, so they read as translucent and non-obstructive. They are very subtle, fairly vertical and similar to each other, and they hardly spread." }, "composition": { "score": 0.62, "reason": "Dark, moody, centered single bowl with a warm glow behind it keeps the focus on the dish and has no clutter. The food is small within an oversized bowl, and the nori sticking up looks odd." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens, scallions, nori and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are a small pale patch of thin parallel strands in the middle of the bowl, mostly hidden by toppings. They look flat and rigid, with no readable bends or volume, and only a few red chili threads add detail." }, "toppings": { "score": 0.5, "reason": "Leaves with veins, shiitake-like mushrooms, tofu cubes, scallion rings and a nori sheet are all recognizable. Tofu looks like flat cards, the mushrooms are simple ovals, and the nori is a rigid dark block standing upright. Variety is fine but the execution is simple." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body, a foot and subtle highlights. The dark lacquer material reads well. The body is very large relative to the soup surface, and the foot ellipse looks slightly detached." }, "broth": { "score": 0.55, "reason": "The brown broth has a warm gradient and small chili oil specks. It reads as a flat surface, with little depth or sheen. Toppings sit on it without much immersion or ripple." }, "steam": { "score": 0.5, "reason": "Faint thin wisps rise and change shape between frames, so some motion is visible. They are very subtle, vertical and rather uniform, and they rise from the air above the bowl rather than from the soup." }, "composition": { "score": 0.62, "reason": "The bowl is centered in a dark, moody scene with a warm glow and the food is the focus. The bowl dominates the frame and the food is small. The nori jutting out at the side is slightly awkward. There is no distracting clutter." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens, scallions, nori and noodles are visible. There is no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.92 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.4666666666666667, "toppings": 0.5333333333333333, "bowl": 0.8, "broth": 0.6666666666666666, "steam": 0.7333333333333333, "composition": 0.8 }, "reward": 0.6433333333333333, "sampleStdDev": 0.035118845842842444, "judgments": [ { "noodles": { "score": 0.6, "reason": "A coherent pale-yellow bundle of wavy strands spans the bowl; strands are continuous with plausible bends, but they read as a flat, repetitive horizontal band with little volume or broth immersion." }, "toppings": { "score": 0.55, "reason": "Recognizable set: glossy green leaves, brown mushroom-cap ellipses, scallion rings, red chili threads and a dark green cylinder. Tofu squares look like flat dotted cards and the leaves are slightly plastic-glossy, keeping quality stylized." }, "bowl": { "score": 0.8, "reason": "Clean elliptical rim with a ceramic sheen arc, smooth dark body with convincing gradient shading, coherent interior/cavity and a grounding base shadow; no visible distortion or detached parts." }, "broth": { "score": 0.7, "reason": "Dark reddish-brown liquid surface with warm central glow, scattered chili-oil droplets and toppings resting at the surface; slightly flat with minimal occlusion shadows around the ingredients." }, "steam": { "score": 0.75, "reason": "Translucent irregular wisps rise and fade without hiding food; their shapes and positions clearly change across desktop-1/2/3 normal frames, indicating animated motion; wisps are somewhat uniform vertical strands." }, "composition": { "score": 0.8, "reason": "Centered focal bowl, warm amber backlight and dark atmospheric background with steam occupying the upper space; no text or clutter, though the green cylinder protruding behind the rim reads slightly odd." }, "vegan": { "score": 1.0, "reason": "Visible toppings are greens, mushrooms, tofu, scallions, chili threads and oil in plant broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.4, "reason": "The noodles read as a single flat pale-yellow band with faint stripes lying on the surface, not supple continuous strands; there is no coherent strand volume or believable immersion, so it stays a crude ribbon-like mass." }, "toppings": { "score": 0.5, "reason": "Some variety exists (green leaf blobs, scallion rings, red chili threads, a dark green stalk), but the tofu squares look like flat rigid beige cards pasted on the surface and the brown ovals are featureless smooth glossy ellipses rather than readable mushrooms, giving a plastic/stylized look." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: consistent elliptical rim with highlight arc, deep smooth walls, subtle side highlight, foot ellipse and contact shadow at the base; no broken joins or perspective mismatch, though the body is very dark and slightly plain." }, "broth": { "score": 0.6, "reason": "Brown liquid surface with warm central glow and chili-oil dots reads as soup, but the surface is largely a flat gradient and the toppings appear pasted atop it with minimal occlusion or immersion shadows." }, "steam": { "score": 0.7, "reason": "Translucent faint wisps rise above the bowl and their shapes/positions clearly differ across desktop-1/2/3 and pointer frames, showing real animation; they do not hide the food, though the wisps are fairly straight vertical streaks rather than irregular spreading plumes." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric scene with the bowl as clear focal point, warm background glow, rising steam, no text or clutter, and consistent lighting; the upright green stalk at back-right adds interest without distraction." }, "vegan": { "score": 0.9, "reason": "Visible toppings are greens, scallion rings, chili threads, tofu-like squares, a leek stalk and brown oval shapes that plausibly read as mushrooms; no clearly animal-derived ingredients, with only mild ambiguity about the brown ovals." } }, { "noodles": { "score": 0.4, "reason": "Noodles read as a flat pale-yellow band behind the toppings with only a few darker strand lines; no continuous supple strands, ends, bends or volume are visible \u2014 an indistinct flat noodle mass." }, "toppings": { "score": 0.55, "reason": "Recognizable variety: green leaf clusters, brown mushroom-like ellipses with line detail, speckled tofu squares, scallion rings and red chili threads. But forms are glossy and synthetic: smooth plastic sheen on leaves/mushrooms, flat rigid tofu cards, and a thick rigid dark-green cylinder planted vertically like a card dominates the silhouette." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic donburi: elliptical rim with thin highlight arcs, consistent tapering walls, visible interior depth, grounded foot with contact shadow. Construction is clean; only minor shortness is the plainness of the body shading." }, "broth": { "score": 0.7, "reason": "Appetizing reddish-brown surface well fitted to the rim, with darker meniscus edge, scattered chili-oil specks and restrained sheen; toppings mostly sit within the liquid with subtle contact. Slightly flat/airbrushed liquid texture with a pasted-on feel for some items." }, "steam": { "score": 0.75, "reason": "Thin translucent wisps rise above the bowl and clearly change shape and position across the time-ordered desktop frames (desktop-1/2/3 and pointer frames), showing continuous rise/sway without hiding the food. Wisps are somewhat repetitive curly strokes but are animated, not static." }, "composition": { "score": 0.8, "reason": "Centered focal dish in a dark atmospheric scene with a warm glow behind, no text/UI/clutter or competing elements; colors (dark bowl, warm broth, green accents) are balanced. Mobile framing also crops cleanly." }, "vegan": { "score": 1.0, "reason": "Visible ingredients \u2014 noodles, leafy greens, mushrooms, tofu squares, scallion, chili \u2014 are all plant-derived; no eggs, meat, fish or dairy appear." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9622222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 275.0499999999993, "frameIntervalP95Ms": 283.40000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "638c2979e1bc92dd9d1a5ee8ed86c5b81dcb50419b75d2b6478f1e36df1fcea1", "gradedAt": "2026-09-30T23:09:08.271018+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4355555555555556, "toppings": 0.55, "bowl": 0.6344444444444445, "broth": 0.5111111111111111, "steam": 0.40555555555555556, "composition": 0.7066666666666667 }, "reward": 0.5341111111111112, "runId": "openai/gpt-5.6-sol/low", "rewardStdDev": 0.09880751958189181, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.4066666666666667, "toppings": 0.5, "bowl": 0.6533333333333333, "broth": 0.4, "steam": 0.3, "composition": 0.72 }, "reward": 0.494, "sampleStdDev": 0.003464101615137758, "judgments": [ { "noodles": { "score": 0.42, "reason": "Continuous golden strands are readable, but mostly form evenly spaced horizontal waves with blunt exposed ends, little overlap, and minimal volume or immersion." }, "toppings": { "score": 0.5, "reason": "Veined leaves, capped mushrooms, scallion rings, and speckled tofu are recognizable. Simple flat blocks and disks dominate, with sparse organic texture and overly geometric stems." }, "bowl": { "score": 0.66, "reason": "The elliptical rim and attached tapered body establish a coherent cavity, with restrained gold edging and dark shading. The angular lower silhouette and weak base/contact detail reduce ceramic plausibility." }, "broth": { "score": 0.4, "reason": "Warm brown coloring suggests broth, but the surface is largely a flat opaque fill without liquid highlights or depth; ingredients appear laid on top rather than partially immersed." }, "steam": { "score": 0.3, "reason": "Faint translucent patches change above the dish across normal frames without obscuring it. They resemble disconnected blurred clouds more than wisps visibly emerging, rising, and spreading from the soup." }, "composition": { "score": 0.72, "reason": "A single readable dish, warm restrained palette, and uncluttered dark setting provide a clear focus. Large upper negative space and the small mobile presentation weaken the balance." }, "vegan": { "score": 0.85, "reason": "Tofu, mushrooms, leaves, and scallions plausibly support vegan ramen. Red-rimmed pale slices with pink centers suggest radish, though their simplified disk depiction leaves some ingredient ambiguity; no unmistakable animal topping is visible." } }, { "noodles": { "score": 0.4, "reason": "Continuous golden strands are readable, but nearly parallel, repeated shallow waves form a flat striped sheet rather than a supple tangled volume." }, "toppings": { "score": 0.5, "reason": "Veined leaves and capped mushrooms are recognizable, with some shading and tofu speckles. Mostly geometric blocks, simple disks, and stiff stems make the ingredients look schematic." }, "bowl": { "score": 0.65, "reason": "Elliptical rim, connected walls, and dark shading establish a coherent vessel. The sharply polygonal lower body and minimal material highlights limit ceramic plausibility." }, "broth": { "score": 0.4, "reason": "Warm brown coloration reads as soup, but the surface is largely flat and matte, with little liquid texture or sheen; ingredients appear laid over it rather than immersed." }, "steam": { "score": 0.3, "reason": "Faint translucent patches above the soup change slightly across normal frames without obscuring food, but read as diffuse floating blobs rather than connected rising, spreading wisps." }, "composition": { "score": 0.72, "reason": "A single uncluttered dish stands out against a restrained dark warm background. Desktop framing is balanced, but the mobile dish is small amid excessive empty space." }, "vegan": { "score": 0.8, "reason": "Tofu-like blocks, mushrooms, greens, and noodles support a plant-based reading. Pink-centered pale disks could be radish slices, but their simplified depiction leaves some ambiguity." } }, { "noodles": { "score": 0.4, "reason": "Continuous golden strands are readable, but nearly parallel, repeated shallow curves form a flat striped sheet rather than a supple tangled volume." }, "toppings": { "score": 0.5, "reason": "Veined leaves, capped mushrooms and speckled tofu are recognizable and varied, but shapes remain simplified: tofu resembles flat cards, mushroom stems are rigid bars, and round slices lack organic texture." }, "bowl": { "score": 0.65, "reason": "The elliptical rim and attached walls establish a coherent cavity with restrained gold highlights. The angular lower silhouette and very dark, minimally modeled body weaken ceramic depth and support." }, "broth": { "score": 0.4, "reason": "Warm brown color reads as soup, but the surface is largely a flat opaque fill without liquid highlights or texture; ingredients appear laid above it rather than visibly immersed." }, "steam": { "score": 0.3, "reason": "Faint translucent patches change position across normal frames without obscuring the food, but look like disconnected blurred blobs rather than wisps visibly emerging, spreading and fading from the soup." }, "composition": { "score": 0.72, "reason": "A single uncluttered dish is clearly framed against a restrained dark warm background. Desktop balance is pleasant, though the mobile dish is small amid extensive empty space and lighting offers limited depth." }, "vegan": { "score": 0.8, "reason": "Tofu, mushrooms, leaves and scallion-like rings support a plant-based appearance. Pink-centered white slices with red edges could be radish, but their generic depiction leaves some ambiguity." } } ], "diagnostics": { "vegan": 0.8166666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.5, "bowl": 0.5166666666666667, "broth": 0.48333333333333334, "steam": 0.3, "composition": 0.6 }, "reward": 0.46166666666666667, "sampleStdDev": 0.00763762615825974, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin, parallel, rigid-looking lines that form a flat stacked block with little overlap or volume. They sit on top of the broth like a ribbon sheet instead of being immersed, and they end abruptly with straight cut edges." }, "toppings": { "score": 0.5, "reason": "The toppings are recognizable: tofu cubes, mushrooms, leaves and scallion rings. The pink-centered discs look like narutomaki, and the tofu is blocky and flat. Everything is flat vector art with little texture, and the scattered toppings look pasted on." }, "bowl": { "score": 0.5, "reason": "The bowl has a readable rim and a dark body with a copper edge. The faceted polygonal base looks angular and doesn't match the smooth elliptical rim, and the walls lack highlights, so the join is awkward. Support is weak, though there is a shadow." }, "broth": { "score": 0.45, "reason": "The broth is a warm orange-red gradient with depth at the edges, but the surface is flat with no texture or sheen. Ingredients sit above it without ripples or occlusion, so they look pasted on." }, "steam": { "score": 0.3, "reason": "The steam is very faint, with barely visible dim smudges that shift a little between frames. It is too subtle to read as wisps and is mostly lost in the dark background." }, "composition": { "score": 0.6, "reason": "The bowl is centered, the dark mood is clean and the warm palette works. The food is small in the frame and the top half of the scene is empty, but nothing distracts from the bowl." }, "vegan": { "score": 0.6, "reason": "The tofu, mushrooms, leaves and scallions are vegan. The pink-centered swirl discs resemble narutomaki fish cake, or possibly egg slices, which makes the compliance ambiguous." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, near-parallel flat lines, rigid and stacked like a ribbon sheet with no volume. They sit in a rectangular block and don't look immersed in the broth." }, "toppings": { "score": 0.5, "reason": "Leaves, mushrooms, tofu cubes and scallion rings are all recognizable, with some vein and texture detail. The tofu reads as flat blocks, and the pink-centered circles look like narutomaki or fish cake. The pieces feel pasted on and generic." }, "bowl": { "score": 0.5, "reason": "The bowl has a rim, dark body and copper trim, but the faceted polygonal base is angular and oddly shaped. The perspective is inconsistent between the smooth elliptical rim and the flat angular lower body, and there are no lacquer highlights." }, "broth": { "score": 0.5, "reason": "The broth is a warm orange-brown gradient with a believable color and depth at the edges. It is fairly flat, and the noodles and toppings look placed on top rather than sitting in it, with no contact ripples." }, "steam": { "score": 0.3, "reason": "Very faint smoky glows drift above the bowl and change slightly across frames. They are barely visible and read as a vague haze rather than wisps." }, "composition": { "score": 0.6, "reason": "The bowl is centered and is a clear focal point against a dark, moody background with warm lighting. The palette is cohesive, though the scene is sparse and the bowl is small within the frame." }, "vegan": { "score": 0.6, "reason": "Tofu, mushrooms, leaves and scallions are vegan. The pink-centered circular slices resemble narutomaki or fish cake, which is ambiguous and possibly animal-derived." } }, { "noodles": { "score": 0.4, "reason": "Noodles are a flat stack of parallel, nearly straight lines with only slight waves. They read as rigid ribbons or combed strands and lack volume or overlap. They end abruptly in a rectangular block and do not sit convincingly in the broth." }, "toppings": { "score": 0.5, "reason": "Toppings are recognizable: tofu cubes, leaves with veins, mushrooms with stems and caps, scallion rings. The pink-centered discs look like generic plastic disks, and the tofu is blocky and flat. Items look pasted on top rather than organic, and there is no nori." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent elliptical rim with a copper edge and a dark faceted body with a shadow. The lower body is angular and polygonal, and the rim-to-body join is slightly odd. The dark lacquer is mostly flat with few highlights." }, "broth": { "score": 0.5, "reason": "The broth is a warm orange-brown gradient that is appetizing and reads as liquid. It is fairly flat, with little sheen or texture. The noodles and toppings look placed on the surface rather than immersed, with no contact ripples." }, "steam": { "score": 0.3, "reason": "Steam is very faint, only a dim brownish haze above the bowl. It shifts slightly between frames but is barely visible and does not read as wisps." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark, moody background with warm colors, and the food is the focus. The scene is clean, without clutter or text. The bowl is somewhat small in the frame and the lighting is simple." }, "vegan": { "score": 0.75, "reason": "Tofu, mushrooms, leaves and scallions are plant-based. The pink-centered white discs resemble narutomaki or fish cake, or perhaps radish slices, which is ambiguous." } } ], "diagnostics": { "vegan": 0.65 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5, "toppings": 0.65, "bowl": 0.7333333333333333, "broth": 0.65, "steam": 0.6166666666666667, "composition": 0.8 }, "reward": 0.6466666666666667, "sampleStdDev": 0.01376892636821524, "judgments": [ { "bowl": { "reason": "Coherent elliptical rim with dark inner band, gold rim line, faceted dark body with gradient shading, base and contact shadow; perspective consistent, though small tab-like notches at the rim's left/right ends read as minor join artifacts.", "score": 0.7 }, "broth": { "reason": "Dark reddish broth with a warm radial glow and restrained sheen; ingredients sit on the surface plausibly, but the surface is fairly flat with little texture/depth and noodle ends stop abruptly at the band edges.", "score": 0.6 }, "composition": { "reason": "Balanced centered focal bowl, warm dish glow against a clean dark vignette background, no clutter or distracting UI; presentation is coherent and atmospheric though somewhat plain.", "score": 0.8 }, "noodles": { "reason": "Noodles are continuous strands with plausible thickness and slight waviness, but they lie as a flat, uniform horizontal mat of near-parallel lines spanning the bowl with abrupt cut ends, lacking volume, overlap, or believable nesting in the broth \u2014 rigid flat-band territory.", "score": 0.5 }, "steam": { "reason": "Faint translucent wisps clearly shift shape and position across the time-ordered frames (short blobs early, a taller plume later), so motion is supported; however the steam reads as disconnected blurred blobs rather than continuous rising wisps and is very subtle.", "score": 0.6 }, "toppings": { "reason": "Recognizable variety \u2014 speckled tofu cubes with bevels, shiitake-like mushrooms with stems and angled caps, leafy greens with midveins, scallion rings, chili flecks and naruto-style discs \u2014 but all are flat cut-out shapes with simple geometry; leaves and tofu are somewhat rigid.", "score": 0.65 }, "vegan": { "reason": "The white discs with pink spiral centers and red rims are unmistakably narutomaki fish cakes, an animal-derived ingredient, which rules out full vegan credit despite otherwise plant-based toppings (tofu, mushrooms, greens, scallions).", "score": 0 } }, { "bowl": { "reason": "Coherent rim ellipse with a gold highlight ring and dark inner band, tapered faceted body meeting a shallow polygonal foot, plausible contact shadow. Low-poly faceting is a consistent stylistic choice, though the walls are stiff and the base reads slightly angular.", "score": 0.75 }, "broth": { "reason": "Warm radial gradient gives a readable liquid sheen and depth without concentric rings, but there are no immersion cues: noodles and toppings sit flat on top of the surface with no occlusion, ripples, or meniscus integration.", "score": 0.65 }, "composition": { "reason": "Centered focal dish on a dark warm-backlit background with a grounding shadow; no text, UI, or clutter; mobile view scales the scene cleanly. Upper steam smudges are slightly murky and the mobile frame leaves much empty space.", "score": 0.8 }, "noodles": { "reason": "Noodles form a uniform bundle of parallel wavy strands confined to a flat rectangular patch with abruptly cut ends; no overlap, drape, or volume, reading as a mechanical mat of repeated lines rather than supple tangled strands.", "score": 0.5 }, "steam": { "reason": "Wisps above the bowl change position and shape across the normal frames, so the steam is genuinely animated, irregular, and does not hide the food; however the plumes render as faint dark smudgy blobs that read more like smoke than translucent rising steam.", "score": 0.6 }, "toppings": { "reason": "Good variety of recognizable items: shiitake caps with stems, speckled tofu cubes, veined green leaves, scallion rings, and chili flecks. Defects: tofu blocks are flat card-like wedges and the two sliced rounds are flat concentric disks, keeping the set stylized rather than convincing.", "score": 0.65 }, "vegan": { "reason": "Tofu, mushrooms, greens, scallions, and chili are clearly plant-based, but the two white disks with pink centers and red rims strongly suggest naruto fish cake; the pattern is ambiguous enough (concentric, no spiral) that a tomato/radish reading is possible, so an intermediate score applies.", "score": 0.5 } }, { "bowl": { "reason": "Coherent dark ceramic bowl with gold-trimmed rim, tapering faceted walls and a grounded base with contact shadow; rim/wall/base perspective joins read plausibly, only a small rim seam artifact on the right edge.", "score": 0.75 }, "broth": { "reason": "Warm amber-brown liquid surface with a restrained radial sheen and dark edge falloff; ingredients sit on the surface convincingly, though the surface is very smooth and flat with little texture, swirl or immersion depth.", "score": 0.7 }, "composition": { "reason": "Balanced centered bowl against a dark atmospheric backdrop with a warm glow; dish is the clear focal point with no text, UI or clutter; mobile view scales cleanly without overflow.", "score": 0.8 }, "noodles": { "reason": "Noodles read as a tidy rectangular patch of parallel yellow strands lying flat across the bowl; strands are continuous but uniform and rigid, lacking tangle, drape into broth, or overlapping volume \u2014 a mechanical bundle rather than supple noodles.", "score": 0.5 }, "steam": { "reason": "Faint translucent wisps above the bowl change shape and position across the time-ordered desktop and pointer frames, so steam is genuinely animated; it stays subtle and never hides the food, but the wisps are soft diffuse blobs rather than defined rising plumes.", "score": 0.65 }, "toppings": { "reason": "Recognizable variety: shiitake-style mushrooms with caps and stems, leafy greens with midribs, scallion rings, chili flecks and speckled tofu cubes; shapes are coherent in the flat vector style but the tofu is a plain flat square and slices with pink centers read as narutomaki-style fishcake.", "score": 0.65 }, "vegan": { "reason": "The white discs with red rims and pink centers are the standard visual shorthand for narutomaki fishcake slices, a fish-derived topping; mushrooms, tofu, greens and scallions are otherwise plant-appropriate.", "score": 0.1 } } ], "diagnostics": { "vegan": 0.2 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.5555555555555556, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 658.3499999999985, "frameIntervalP95Ms": 650, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "ec26fea19f704b400afad16eb37832b66e8d9c29369da09a55eb5dd397ddff31", "gradedAt": "2026-09-30T23:06:19.850538+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5666666666666667, "toppings": 0.6744444444444444, "bowl": 0.7522222222222222, "broth": 0.6344444444444445, "steam": 0.6977777777777777, "composition": 0.7866666666666667 }, "reward": 0.6791666666666667, "runId": "openai/gpt-5.6-sol/max", "rewardStdDev": 0.07904165568441321, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.63, "bowl": 0.8333333333333333, "broth": 0.56, "steam": 0.71, "composition": 0.8466666666666667 }, "reward": 0.6713333333333333, "sampleStdDev": 0.003752776749732598, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable overlaps, but mostly form flat, similarly thick sweeping curves with exposed ends and little convincing submerged volume." }, "toppings": { "score": 0.64, "reason": "Mushrooms have distinct caps, stems and gill marks; tofu has shaded faces and small pores. Greens are stiff and simplified, corn resembles flat disks, and nori is an overly straight card." }, "bowl": { "score": 0.84, "reason": "Coherent elliptical rim, curved walls, subtle foot and contact shadow establish a supported vessel. Fine surface texture and restrained rim highlights suggest dark ceramic, though the front inner rim is unusually broad and dark." }, "broth": { "score": 0.57, "reason": "Warm amber shading makes the soup readable, with scattered small surface details. Limited liquid sheen and weak immersion around tofu and vegetables leave ingredients looking placed on a flat surface." }, "steam": { "score": 0.7, "reason": "Normal frames show changing heights, bends and opacity in translucent wisps emerging above the soup without obscuring it. The narrow, similarly shaped vertical plumes lack convincing spreading and irregular breakup." }, "composition": { "score": 0.85, "reason": "Warm food colors contrast effectively with the dark ceramic and restrained background; the single dish remains the clear focus without text or clutter. Mobile framing is readable, though desktop placement is somewhat bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, seaweed and noodles plausibly depict vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and some overlap, but predominantly form flat, similarly spaced arcs with little tangled volume or convincing immersion." }, "toppings": { "score": 0.63, "reason": "Mushrooms have readable caps, stems and gill marks; tofu has shaded faces and small pores. Greens are stiff and simplified, corn resembles flat disks, and nori is an overly regular card." }, "bowl": { "score": 0.83, "reason": "Coherent elliptical rim, curved walls and grounded base establish convincing depth, with fine dark surface texture and restrained rim highlights. The broad black inner band is somewhat visually heavy." }, "broth": { "score": 0.55, "reason": "Warm brown-orange shading makes the soup readable, but the surface lacks liquid sheen and contact effects; several toppings appear placed on top rather than partially immersed." }, "steam": { "score": 0.72, "reason": "Normal frames show changing heights, curves and opacity in translucent wisps emerging above the soup. They do not obscure the food, but remain narrow, similarly shaped columns with limited spreading." }, "composition": { "score": 0.85, "reason": "The single bowl is clearly emphasized by warm food colors against a restrained dark setting, with no text or clutter. Desktop spacing is balanced; the mobile bowl sits rather tightly against the side edges." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have readable bends and some overlap, but mostly form flat, evenly thick sweeping lines with little tangled volume or believable submersion." }, "toppings": { "score": 0.62, "reason": "Mushroom caps and stems, veined greens, scallion rings and dotted tofu cubes are recognizable. Shapes remain simplified and stiff, with disk-like corn and an overly straight, card-like seaweed edge." }, "bowl": { "score": 0.83, "reason": "Coherent elliptical rim, curved walls and subtle foot establish convincing depth and support. Fine dark ceramic texture and restrained rim highlights add material character; the broad black inner lip is somewhat heavy." }, "broth": { "score": 0.56, "reason": "Warm brown gradients and small surface accents suggest soup, but the surface has little liquid sheen and toppings appear perched on it rather than partially immersed." }, "steam": { "score": 0.71, "reason": "Normal frames show changes in wisp curvature, height and opacity, with translucent trails originating near the soup and leaving food readable. The narrow, similarly shaped sinusoidal columns feel repetitive and show little spreading." }, "composition": { "score": 0.84, "reason": "The single bowl is clearly emphasized by warm food colors against a restrained dark setting, with ample room for steam and no distracting UI or text. Mobile framing is tight at the bowl's sides, and the lower vessel is quite dark." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.6333333333333333, "bowl": 0.6066666666666667, "broth": 0.6, "steam": 0.6, "composition": 0.68 }, "reward": 0.6043333333333334, "sampleStdDev": 0.012055427546683426, "judgments": [ { "noodles": { "score": 0.55, "reason": "Golden noodle strands curve and overlap with some taper, but they read as thin flat stroke lines, with repeated similar curves and little volume or immersion in the broth." }, "toppings": { "score": 0.62, "reason": "Tofu cubes, shiitake-like mushrooms, bok choy, scallion rings and corn kernels are all recognizable and varied, with some shading. They look like flat vector illustration, the nori is mostly hidden behind the rim, and the tofu cubes are slightly rigid." }, "bowl": { "score": 0.6, "reason": "The dark bowl has a coherent rim, wall and curved base with subtle highlights and a contact shadow. The body looks translucent and flat, the rim is thick, and the lower wall has odd semi-transparent lines that weaken the ceramic material." }, "broth": { "score": 0.6, "reason": "The warm amber-brown broth has a gradient, chili oil specks and a glow, which is appetizing. It is fairly flat and the toppings sit on it with little contact or ripple." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise and sway differently across frames without hiding the food. They are fairly thin and uniform in style, and somewhat faint." }, "composition": { "score": 0.68, "reason": "The centered bowl is a clear focal point on a dark moody background with a warm glow and no clutter. The bowl sits a bit high, with large empty areas above and below." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, bok choy, scallion, corn and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.55, "reason": "Noodles are smooth, curved golden strands with some overlap, but they are thin, fairly flat and uniform. They read as stylized lines lying on the surface, with little volume or immersion." }, "toppings": { "score": 0.66, "reason": "Tofu cubes, shiitake-style mushrooms, bok choy, scallion rings, corn kernels and a nori corner are all readable, with gradients and small details like mushroom gills and tofu pores. They are vector-clean and somewhat generic. The nori is a stiff triangle behind the rim." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a curved body and a base shadow, and the rim and body perspectives match. The dark translucent body is flat and murky, the highlights are faint, and the base contact is weak." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange-brown with a glow gradient and small chili oil droplets, so it reads as appetizing. It looks fairly flat, and the ingredients sit on it with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise and wave, and they change between frames. They are fairly thin and similar to one another, and they fade out well above the bowl." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with warm food colors and no clutter. Its placement is a little low and the bowl body is dim, but the focal point is clear." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, corn, scallions and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin golden curved strands with some overlap and bends, but they read as flat, uniform ribbons on the surface with little volume or depth, and they look somewhat scribbled." }, "toppings": { "score": 0.62, "reason": "Tofu cubes, shiitake mushrooms with gills and stems, bok choy and scallion slices are all recognizable with some shading. They look vector-flat and a bit pasted on, and the nori is barely visible behind the rim." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, curved walls and a base shadow, with subtle highlights. It is very dark and semi-transparent looking, the rim is thick, and the inner wall is unclear, so the ceramic material is weakly conveyed." }, "broth": { "score": 0.6, "reason": "Warm amber-brown broth with a gradient and chili oil specks is appetizing. It looks fairly flat, and the ingredients sit on top with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Thin translucent wavy wisps rise and change between frames, without hiding the food. They are fairly uniform, thin line-like plumes and look somewhat mechanical." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm food colors and no clutter. The bowl is dark and blends into the background, and there is a lot of empty space." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, scallions and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.76, "bowl": 0.8166666666666667, "broth": 0.7433333333333333, "steam": 0.7833333333333333, "composition": 0.8333333333333334 }, "reward": 0.7618333333333334, "sampleStdDev": 0.03335540935640478, "judgments": [ { "bowl": { "reason": "Coherent dark ceramic bowl: elliptical rim matches the body curve, double-edge rim with subtle highlight arc, plausible foot/contact with the dark table. Simple material with faint noise texture; no broken joins or perspective clashes observed.", "score": 0.8 }, "broth": { "reason": "Rich amber-brown gradient with restrained sheen, sesame and chili speckles, and believable depth; noodles pass behind tofu and mushrooms, and toppings cast small contact shadows so they read as sitting in the soup rather than pasted on. Slight flat-illustration pasted feel keeps it short of excellent.", "score": 0.7 }, "composition": { "reason": "Balanced centered dish with warm rim glow against a dark atmospheric backdrop, nori peeking behind the rim for interest, no text or clutter. The very dark lower bowl body nearly merges with background, slightly reducing readability.", "score": 0.8 }, "noodles": { "reason": "Wavy yellow strands lie plausibly across the broth surface and are partially occluded by toppings, but they are flat uniform ribbons with a repetitive S-wave pattern and little volume or thickness variation; strand shapes look stamped rather than naturally varied.", "score": 0.6 }, "steam": { "reason": "Translucent irregular wisps rise and fade above the bowl and visibly change shape across desktop-1/2/3 and pointer frames, so the steam is animated, not static. Wisps are somewhat thin and line-like but stay translucent and never hide the food.", "score": 0.8 }, "toppings": { "reason": "Good variety and readability: two shiitake with caps and cream stems, textured tofu cubes, bok choy with pale stems, scallion rounds, corn kernel clusters, chili bits and sesame. Shapes are a bit schematic (blocky tofu, simple bok choy blobs) but organic enough for the style.", "score": 0.7 }, "vegan": { "reason": "Visible ingredients are tofu, mushrooms, bok choy, scallion, corn, chili, sesame and nori in plant broth; no animal-derived toppings apparent.", "score": 1 } }, { "bowl": { "reason": "Coherent dark ceramic bowl: elliptical rim with a believable inner lip, smoothly bulging body, subtle sheen highlight, and grounded contact with a soft reflection. Perspective between rim ellipse and body is consistent; no broken joins observed.", "score": 0.8 }, "broth": { "reason": "Rich brown miso-toned surface with a warm radial glow, chili-oil droplets, sesame specks, and soft contact shadows where mushroom stems, bok choy and tofu meet the liquid; integration is believable though the surface is fairly flat in tone.", "score": 0.75 }, "composition": { "reason": "Centered focal dish against a restrained dark vignette, warm broth glow providing strong contrast, clean reflective surface, no text or clutter; renders well at both desktop and mobile proportions.", "score": 0.85 }, "noodles": { "reason": "Amber strands read as supple wavy noodles with natural bends, coherent crossings, partial submersion and sesame seed detail; somewhat uniform ribbon-like strokes in the flat style but no maze scribbles or broken strands.", "score": 0.75 }, "steam": { "reason": "Translucent gray wisps rise, curl and fade above the soup without obscuring food; wisp shapes and positions clearly differ across desktop-1/2/3 and pointer frames, confirming continuous animated motion, while reduced frames are properly static.", "score": 0.8 }, "toppings": { "reason": "Good variety and readability: shiitake caps with gill lines and stems, leafy bok choy with vein detail, speckled tofu cubes, corn kernels, scallion rings, chili pieces and a nori sheet at the rim; organic shapes throughout, only the nori's placement slightly abrupt.", "score": 0.8 }, "vegan": { "reason": "Visible ingredients are mushrooms, bok choy, tofu, corn, scallion, nori, noodles and chili oil in plant broth; no animal-derived toppings appear.", "score": 1 } }, { "bowl": { "reason": "Coherent construction: elliptical rim matches the tapered body, a dark inner ring reads as wall thickness, subtle ceramic highlight streaks, and a soft contact shadow/reflection at the base. No perspective mismatch or detached parts observed.", "score": 0.85 }, "broth": { "reason": "Warm orange-brown surface with radial depth, darker oil/chili patches, and floating sesame and chili bits; mushroom stems and noodles visibly sit in the liquid. Slightly uniform flatness in the mid-surface keeps it from excellent.", "score": 0.78 }, "composition": { "reason": "Centered focal dish on an atmospheric dark backdrop with a restrained warm rim glow, no text, UI or clutter; mobile crop keeps the bowl centered with no overflow (diagnostics confirm). Empty dark upper area is largely justified by steam.", "score": 0.85 }, "noodles": { "reason": "Supple continuous tan strands with organic curves and some coherent overlap across the broth. However they are flat uniform ribbon strokes with repeated similar curvature and a few abrupt ends, lacking volume and texture detail.", "score": 0.65 }, "steam": { "reason": "Translucent gray wisps that visibly change shape, position and opacity across the desktop frames (1009ms to 10120ms), indicating continuous rising and fading without hiding the food. Wisps are somewhat similar repeated S-curves; static in reduced frames as expected.", "score": 0.75 }, "toppings": { "reason": "Recognizable shiitake mushrooms with caps, gill area and stems; shaded tofu cubes with sesame specks; upright bok choy-like greens; okra/chili slices with seed dots; chili pieces. The yellow disc cluster is a flat generic shape and the dark nori sheet behind the rim reads angular and rigid.", "score": 0.78 }, "vegan": { "reason": "Visible ingredients are tofu, mushrooms, leafy greens, okra/chili slices, corn-like discs, chili, sesame, noodles and plant broth; no animal-derived toppings appear.", "score": 1 } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.39999999999964, "frameIntervalP95Ms": 183.3000000000011, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "f3bb25a96c46a47f04e6a4e6a79deb4ef467feff117d426b81c8fd8c10a881ae", "gradedAt": "2026-09-30T23:08:56.749427+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4855555555555556, "toppings": 0.5933333333333333, "bowl": 0.6977777777777778, "broth": 0.5277777777777778, "steam": 0.6544444444444445, "composition": 0.7388888888888889 }, "reward": 0.6098333333333333, "runId": "openai/gpt-5.6-sol/medium", "rewardStdDev": 0.08996959362905772, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.44, "toppings": 0.58, "bowl": 0.7366666666666667, "broth": 0.43333333333333335, "steam": 0.6466666666666667, "composition": 0.7766666666666667 }, "reward": 0.5983333333333334, "sampleStdDev": 0.001258305739211822, "judgments": [ { "noodles": { "score": 0.45, "reason": "Continuous golden strands are readable, but closely repeated parallel waves form a flat ribbon-like mat with little natural overlap or submerged volume." }, "toppings": { "score": 0.58, "reason": "Green leaves have central veins, scallion rings are readable, and tofu has small speckles. Mushroom caps lack gills and organic detail, while leaves and tofu remain stiff, simplified shapes." }, "bowl": { "score": 0.73, "reason": "Elliptical rim, curved walls and attached foot form a coherent vessel with restrained dark highlights. Heavy stacked rim outlines and a very flat foot reduce the ceramic plausibility." }, "broth": { "score": 0.43, "reason": "Warm brown shading distinguishes the soup surface, but it lacks liquid sheen and contact details; the noodle mat and toppings appear laid above it rather than immersed." }, "steam": { "score": 0.65, "reason": "Translucent blurred wisps visibly change shape across normal frames without obscuring the food. They remain mostly tall parallel columns with limited spreading, and the short orange curves look mechanical and opaque." }, "composition": { "score": 0.77, "reason": "A single readable dish anchors the uncluttered dark scene, with coordinated gold, brown and green colors and effective mobile framing. Desktop placement is bottom-heavy with substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu-like cubes, leafy greens, scallions, mushrooms and noodles plausibly depict vegan ramen; no material animal-derived ingredient is visible." } }, { "noodles": { "score": 0.44, "reason": "Continuous golden strands are readable, but form a flat stack of nearly identical parallel waves with aligned ends, little overlap, and little convincing noodle volume." }, "toppings": { "score": 0.58, "reason": "Veined green leaves, scallion rings, speckled tofu, and brown mushrooms provide recognizable variety. Leaves are overly smooth and repeated, while mushroom caps lack gills and tofu has simplified slab-like texture." }, "bowl": { "score": 0.74, "reason": "Elliptical rim, curved walls, and shaded dark finish create coherent depth. The broad flat oval foot has an abrupt join, and heavy layered rim outlines weaken the ceramic impression." }, "broth": { "score": 0.43, "reason": "Warm brown surface is visible around the food, but has little liquid sheen or surface detail. Ingredients appear layered atop it rather than convincingly immersed." }, "steam": { "score": 0.65, "reason": "Soft translucent plumes visibly change shape across normal frames and do not obscure the food. They remain fairly parallel and columnar, with an awkward separation from the soup and repeated sharp orange curls below." }, "composition": { "score": 0.78, "reason": "A single readable dish, restrained warm-green palette, and uncluttered dark setting maintain focus on the ramen across desktop and mobile. The low placement and large empty upper area make the balance somewhat bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, leafy greens, scallions, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.43, "reason": "Continuous golden strands are readable, but form a flat, tightly parallel corrugated sheet with matching bends and little natural overlap or submerged volume." }, "toppings": { "score": 0.58, "reason": "Veined green leaves, scallion rings, speckled tofu and mushroom caps provide recognizable variety. Leaves are stiff and similarly shaped, while mushrooms lack gills and tofu has minimal texture." }, "bowl": { "score": 0.74, "reason": "Nested elliptical rim, curved walls and attached foot form a coherent vessel with restrained dark highlights. Heavy stacked rim outlines and a flattened, minimally shaded foot weaken the ceramic depth." }, "broth": { "score": 0.44, "reason": "Warm brown shading establishes a soup surface inside the rim, but it has little liquid sheen or texture; ingredients appear layered on top rather than convincingly immersed." }, "steam": { "score": 0.64, "reason": "Translucent blurred wisps change shape and height across normal frames without obscuring the food. They remain rather parallel and column-like, and the repeated sharp orange curls look diagrammatic rather than vaporous." }, "composition": { "score": 0.78, "reason": "A single readable dish anchors the dark, uncluttered scene, with coordinated amber and green accents and no distracting UI. The low placement and large empty upper area make the balance somewhat bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, leafy vegetables, scallions and noodles plausibly constitute vegan ramen; no visibly animal-derived ingredients are present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.43333333333333335, "toppings": 0.5, "bowl": 0.6066666666666667, "broth": 0.5, "steam": 0.55, "composition": 0.6066666666666667 }, "reward": 0.5261666666666667, "sampleStdDev": 0.0068251984098144, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are stacked parallel wavy ribbons forming a rigid layered slab, like lasagna or a block. Repeated flat lines with no supple strands or natural overlap. They do not sit convincingly in the broth." }, "toppings": { "score": 0.5, "reason": "Leaves, mushroom caps, tofu cubes and scallion rings are recognizable and varied. They are flat, stylized and pasted on top. The mushrooms look like detached caps, and the red squiggles are odd." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent rim, dark ceramic body with highlight, and a base ellipse. The rim is thick and the foot looks slightly detached and flat. It is competent but plain." }, "broth": { "score": 0.5, "reason": "The warm orange-brown gradient reads as broth. The noodle slab sits on top rather than immersed, with little surface texture or contact, so it looks pasted." }, "steam": { "score": 0.5, "reason": "Faint translucent wisps rise and change shape across frames. They are thin vertical streaks that look somewhat mechanical and very dim, but they are not opaque fog." }, "composition": { "score": 0.6, "reason": "Centered bowl with a dark moody background and warm glow gives a clear focal dish. The food is small and flat, and the lower area is empty. Overall it is pleasant but simple." }, "vegan": { "score": 0.9, "reason": "Only noodles, tofu, mushrooms, leaves and scallions are visible, with no animal products." } }, { "noodles": { "score": 0.45, "reason": "Noodles are stacked as parallel, rigid wavy ribbons forming a slab or lasagna-like block. Strands are flat and repetitive with no natural bends or overlap, and they barely sit in the broth." }, "toppings": { "score": 0.5, "reason": "Green leaves, mushroom caps with stems, tofu cubes and scallion rings are recognizable and varied. They look like flat vector symbols, and the mushrooms look odd and partly detached. The red squiggles read as odd filaments. No nori." }, "bowl": { "score": 0.62, "reason": "The dark bowl has a coherent rim, tapered body and a foot, with a subtle highlight. The base ellipse looks detached and oddly wide at the bottom, and the rim is slightly heavy. The material is fairly plain." }, "broth": { "score": 0.5, "reason": "The warm orange-brown broth has a gradient with some depth. It is mostly hidden by the noodle slab, and the ingredients sit on top of it rather than in it. The surface has little sheen or texture." }, "steam": { "score": 0.55, "reason": "Soft translucent wisps rise and change shape across frames, fading upward without hiding the food. They are thin, vertical and fairly uniform, and look somewhat mechanical." }, "composition": { "score": 0.62, "reason": "A centered bowl on a dark, moody background with warm glow keeps the food as the focus and uses a restrained palette. The scene is very dark, the noodle block looks artificial, and there is no clutter." }, "vegan": { "score": 0.9, "reason": "Only noodles, tofu, mushrooms, leaves and scallions are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.4, "reason": "Noodles form a rigid stack of parallel wavy ribbons like a slab or lasagna sheet. There are no individual supple strands and no natural overlap, and the mass sits above the broth." }, "toppings": { "score": 0.5, "reason": "Leaves, mushrooms, tofu cubes and scallion rings are recognizable and vegan. The leaves are decent, but the mushrooms are flat half-domes and the tofu is blocky. The red squiggle lines are odd, and the toppings look pasted on." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent dark lacquer body, a rim ellipse and a base with a soft highlight. The rim is slightly heavy and the foot ellipse looks detached below the body, so the contact is weak." }, "broth": { "score": 0.5, "reason": "The warm orange-brown broth has a radial glow and reads as liquid. The noodle slab and toppings sit on top with little immersion or occlusion, and the surface shows little texture." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise and change shape across frames without hiding the food. They are thin, vertical and somewhat uniform, and very dim." }, "composition": { "score": 0.6, "reason": "The bowl is centered in a dark, moody scene with warm lighting, and the food is the focus. The noodle slab looks artificial, and the lower part of the bowl is cropped a bit dark." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, scallions and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5833333333333334, "toppings": 0.7, "bowl": 0.75, "broth": 0.65, "steam": 0.7666666666666667, "composition": 0.8333333333333334 }, "reward": 0.7050000000000001, "sampleStdDev": 0.028831406486676942, "judgments": [ { "noodles": { "score": 0.6, "reason": "Continuous golden strands with mild waviness and layered shading form a readable noodle mass, but the bundle reads as flat horizontal ribbons laid in a uniform parallel band, with repeated strand patterns and little sense of volume or coiling in the broth." }, "toppings": { "score": 0.7, "reason": "Recognizable variety: brown mushroom caps with stems, green herb leaves with midribs, scallion rings, speckled tofu cubes, sesame seeds and thin red chili threads. Shapes are clean and legible, though quite simple/flat with basic shading rather than rich texture." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl: elliptical rim with a thick dark lip, tapering walls, visible foot ellipse and consistent 3/4 perspective, plus a subtle rim highlight. Simple shading and low contrast on the body keep it below polished levels; no broken geometry observed." }, "broth": { "score": 0.7, "reason": "Amber glossy liquid surface with restrained sheen and a rim highlight arc; ingredients rest on it and broth shows around the noodle bundle. Immersion is only implied since the noodle band sits largely on top of the surface rather than emerging from it." }, "steam": { "score": 0.8, "reason": "Translucent grey wisps rise from the bowl and change shape between desktop-1, 2, 3 and pointer frames, indicating animated rising motion, thinning and fading with height without obscuring the food. Wisps are a bit thin and uniform in style; reduced frames appropriately freeze." }, "composition": { "score": 0.8, "reason": "Centered dish as clear focal point against a warm dark atmospheric backdrop with a soft glow, balanced framing on both desktop and mobile, no text or clutter. Dark-on-dark bowl contrast is slightly muted but the lit broth and steam carry focus." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu cubes, scallions, leafy herbs, sesame and red chili threads in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.6, "reason": "Noodles form a continuous coherent mass with wavy strands, sesame and pepper specks, but the strands read as a uniform repeating hatched mat rather than supple individual strands with volume; flat and pattern-like, though not broken or maze-like." }, "toppings": { "score": 0.75, "reason": "Good variety: recognizable shiitake-like mushroom caps with stems, green leaves with midribs, tofu cubes with speckle texture, scallion rings, chili threads and sesame. Shapes are simple and flat but organic and readable; no plastic blocks or rigid cards dominate." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with elliptical rim band, smooth tapered body, subtle highlight and dark foot/shadow ellipse joining cleanly. Perspective is consistent; the very dark material limits readable detail and the base contact is faint." }, "broth": { "score": 0.65, "reason": "Warm amber broth surface with gradient depth visible around the noodles and at the rim; toppings largely sit in the liquid. The noodle block rests on the surface like a raft with little immersion/occlusion, and the sheen is restrained but the surface is fairly plain." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise above the bowl and clearly change shape and position across the timed desktop frames (1203ms to 30720ms), showing continuous rising/fading motion without hiding the food; wisps are slightly uniform vertical plumes but irregular enough. Reduced frames appropriately hold static steam." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric background with a warm glow behind the bowl, subtle vignette and star dots; no text, UI or clutter, dish stays the focal point, and the mobile layout scales cleanly." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, scallions, chili threads, sesame and plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.55, "reason": "Noodles read as a continuous wavy band with sesame detail, but they are flat stacked horizontal ribbons lying like a mat on the surface, with uniform spacing and no volume or dip into the broth." }, "toppings": { "score": 0.65, "reason": "Good variety: bok-choy-like leaves, mushroom caps with stems, tofu cubes, scallion rings, sesame and chili threads. Shapes are simplified; tofu is flat and angular, some mushrooms sit half off the rim, and the red squiggles float disconnected above the tofu." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl: elliptical rim with highlight ring, tapered shaded body, and a footed base with shadow. Perspective is consistent; minor nit is the slightly flat pasted-looking foot ellipse and a tiny rim-edge tick artifact." }, "broth": { "score": 0.6, "reason": "Dark amber broth surface with subtle sheen and warm glow is visible around the noodle mass, but the noodle mat and toppings read as pasted on top rather than immersed; edges are murky." }, "steam": { "score": 0.7, "reason": "Translucent grey wisps rise above the bowl and clearly change shape/position across desktop frames at 1203/5040/8455/19162/30720 ms, fading toward the top without hiding food. They are somewhat mechanical parallel squiggles; reduced-motion frames are appropriately static." }, "composition": { "score": 0.85, "reason": "Centered dish with dark atmospheric background, warm rim glow, steam as vertical counterpoint, and no clutter or text; scales cleanly on mobile. Green accents balance the warm palette." }, "vegan": { "score": 1.0, "reason": "Visible toppings are leaves, mushrooms, tofu cubes, scallions, sesame and chili in a plant-based broth; no animal-derived ingredients appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9722222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 383.3000000000029, "frameIntervalP95Ms": 383.3000000000029, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "36fb2929bc7b78b0e7322b3b60713ba754e40df04381ac93621f33a9520ae2ad", "gradedAt": "2026-09-30T23:06:52.469214+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.45333333333333337, "toppings": 0.5633333333333334, "bowl": 0.7744444444444445, "broth": 0.5622222222222223, "steam": 0.7011111111111111, "composition": 0.7777777777777778 }, "reward": 0.6255555555555556, "runId": "openai/gpt-5.6-sol/xhigh", "rewardStdDev": 0.06296699605819377, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.44333333333333336, "toppings": 0.59, "bowl": 0.86, "broth": 0.52, "steam": 0.72, "composition": 0.82 }, "reward": 0.6501666666666667, "sampleStdDev": 0.002516611478423644, "judgments": [ { "noodles": { "score": 0.43, "reason": "Thin golden strands overlap clearly, but mostly form sparse, stiff shallow arcs with exposed ends rather than supple tangled noodles with volume." }, "toppings": { "score": 0.59, "reason": "Bok choy has readable veins and shaded stalks, and mushroom caps have scored details. Tofu is minimally textured, while the gridded, regular nori resembles a rigid card." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, rounded walls, attached foot, restrained glaze highlights and contact shadow convincingly establish a dark ceramic bowl. The very dark base loses some definition; no material construction defect is observed." }, "broth": { "score": 0.52, "reason": "Warm brown shading and small oil-like rings suggest soup, but the surface looks largely matte and flat, with little meniscus or convincing immersion around toppings." }, "steam": { "score": 0.72, "reason": "Normal frames show changing bends in translucent wisps that emerge above the soup and fade without obscuring it. The evenly spaced, similarly narrow columns remain somewhat mechanical rather than spreading organically." }, "composition": { "score": 0.82, "reason": "A single uncluttered bowl remains the clear focus against a restrained dark background, with warm soup contrasting green vegetables and ample space for steam. Mobile framing preserves the dish; the lower bowl is slightly too dark." }, "vegan": { "score": 1, "reason": "Visible tofu, mushroom caps, leafy greens, seaweed and noodles plausibly form vegan ramen; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.46, "reason": "Consistent golden strands overlap readably, but mostly form sparse, nearly straight sticks with shallow bends rather than supple noodles with convincing volume." }, "toppings": { "score": 0.59, "reason": "Leaf veins, pale vegetable stalks, scored mushroom caps and scattered scallions provide recognizable variety. Tofu is minimally textured, mushrooms are flattened ovals, and nori resembles a rigid gridded card." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, curved walls, attached foot and soft ceramic highlights establish convincing depth and support. No material construction defect is observed; the dark base is slightly difficult to distinguish." }, "broth": { "score": 0.51, "reason": "Warm brown shading and small oil-like rings suggest soup, but the surface lacks convincing liquid sheen and ingredients largely appear laid on top with little immersion." }, "steam": { "score": 0.72, "reason": "Normal frames show changing bends in translucent wisps emerging behind the soup and fading upward without obscuring food. The evenly spaced, narrow columns remain somewhat mechanical, with limited visible spreading." }, "composition": { "score": 0.82, "reason": "A single readable dish, restrained amber-green palette and uncluttered dark setting provide a clear focal point across desktop and mobile. The low placement and very dark lower bowl leave slightly excessive empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu-like blocks, mushroom caps, leafy greens, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.44, "reason": "Golden strands are individually readable and overlap beneath toppings, but mostly form sparse, nearly straight strokes with little supple bending or piled volume." }, "toppings": { "score": 0.59, "reason": "Leaf veins, pale vegetable stalk, scored mushroom caps and scallion flecks provide recognizable variety. Tofu is minimally textured, mushrooms are disk-like, and the gridded nori resembles a rigid rounded card." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, rounded walls, attached foot and soft curved highlights convincingly establish a dark ceramic vessel. No material construction defect is observed, though the dark base loses some definition." }, "broth": { "score": 0.53, "reason": "Warm brown shading and small oil rings suggest soup, but the surface lacks convincing liquid sheen and depth; ingredients largely appear laid on top rather than immersed." }, "steam": { "score": 0.72, "reason": "Normal frames show changing bends in translucent wisps rising from the soup and fading above it without obscuring food. The similarly spaced, narrow columns remain somewhat mechanical rather than spreading organically." }, "composition": { "score": 0.82, "reason": "A single clearly framed dish, restrained warm-green palette and ample dark negative space create a cohesive atmosphere across desktop and mobile, without clutter or text. The bowl sits somewhat low and its lower silhouette is very dark." }, "vegan": { "score": 1, "reason": "Visible tofu-like blocks, mushrooms, leafy greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4166666666666667, "toppings": 0.5, "bowl": 0.68, "broth": 0.5166666666666667, "steam": 0.6, "composition": 0.68 }, "reward": 0.554, "sampleStdDev": 0.008660254037844393, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin, flat, pale yellow strokes with gentle curves. They read as noodles but look like sparse, wire-like lines lying on the surface, with little volume, overlap or immersion in broth." }, "toppings": { "score": 0.5, "reason": "Broccoli/bok choy, tofu cubes, shiitake caps and a nori sheet are all recognizable. The tofu is blocky and flat, the nori is a rigid grid-textured rectangle, and the mushrooms are simple ovals. The bok choy is the best-drawn element." }, "bowl": { "score": 0.68, "reason": "Coherent dark ceramic bowl with a clear rim, a curved body, a soft highlight and a small foot on a dark surface. Perspective is consistent and the form is solid. The glaze texture is somewhat plain." }, "broth": { "score": 0.5, "reason": "Warm brown broth with a radial gradient and small oil droplets gives some depth. It looks fairly flat and opaque, and the toppings sit on top more than in it." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise and change shape between frames, with subtle warm light behind them. They are fairly thin and uniform, like lines, but they do not hide the food." }, "composition": { "score": 0.68, "reason": "Centered bowl with a dark moody background and a clear focal point. Colors are harmonious and the scene is uncluttered. The nori poking out is slightly awkward, and the lower bowl is very dark." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, nori and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.45, "reason": "Noodles are thin, smooth, pale strands that read as noodles, but they are sparse, flat and mostly parallel arcs. They lack volume and overlap, and look like lines lying on the surface rather than a supple mass in broth." }, "toppings": { "score": 0.5, "reason": "Broccoli/bok choy, tofu cubes, shiitake with radial gills, a nori sheet and scallion bits are all recognizable. The tofu is blocky, the nori is a rigid upright grid-textured rectangle, and the broccoli is partly cut off by the rim. The variety is decent but the forms are stiff." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, curved walls, a glossy highlight, a foot and a shadowed base. The perspective is consistent and the dark glaze looks plausible. The inner wall thickness is minimal and the base contact is faint in the dark." }, "broth": { "score": 0.55, "reason": "The warm brown broth has a radial gradient and some specks of oil, which gives it depth and an appetizing color. The surface is fairly flat, and the toppings sit on it with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Several thin, translucent wisps rise and sway, and their shapes change noticeably between frames. They are line-like and similar to each other, and they start a little above the rim, not from the soup itself. They do not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with a warm glow and no distracting clutter. The food is the focus. The toppings are small, and the nori and steam are slightly off-balance." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, nori, scallions and noodles are visible. There are no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, flat tan curves lying on the surface. They look like a few separate wire-like strands, with no volume, overlap or immersion in the broth. They read as noodles but are sparse and rigid." }, "toppings": { "score": 0.5, "reason": "The bok choy, the shiitake caps with radial cracks and the tofu cubes are recognizable, and the scallion bits add variety. The nori is a tall rigid grid-textured slab that looks like a card. The tofu is blocky, and the bok choy is cut off by the rim and looks pasted on." }, "bowl": { "score": 0.68, "reason": "The dark ceramic bowl has a coherent elliptical rim, curved walls, a base and a soft highlight streak. The perspective is consistent. The base is mostly lost in the dark, and the material is somewhat plain." }, "broth": { "score": 0.5, "reason": "The broth is a warm brown with a radial gradient and some specks, so it is readable. It is flat and matte with little depth or sheen. Toppings sit on top of it with little immersion or contact." }, "steam": { "score": 0.6, "reason": "Several thin translucent wisps rise, and their shapes change between frames, so there is continuous motion. They are fairly thin and similar to one another, and they look like lines rather than volumetric steam. They do not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is a centered focal point in a dark, moody scene with warm and green color harmony and no distracting clutter. The bowl is slightly low and the toppings are crowded to the left. The nori slab sticks out awkwardly." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, bok choy, nori, scallions and noodles. There are no animal products." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5, "toppings": 0.6, "bowl": 0.7833333333333333, "broth": 0.65, "steam": 0.7833333333333333, "composition": 0.8333333333333334 }, "reward": 0.6725, "sampleStdDev": 0.029474565306378993, "judgments": [ { "noodles": { "score": 0.55, "reason": "Thin yellow strands with gentle bends are continuous and readable, but they are sparse, uniform flat lines that float across the broth surface like wires rather than forming a coherent noodle mass with volume or clumping; no immersion or overlap depth." }, "toppings": { "score": 0.65, "reason": "Good vegan variety: stylized broccoli with speckled florets, two scored shiitake-like mushroom slices, tofu cubes with simple shading, scattered scallion rings, and a standing nori sheet. Weaknesses: broccoli florets are blobby with an odd triangular stalk, the nori's grid lines read like graph paper, and tofu cubes are plain prisms." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a clean elliptical rim, smooth curved body, visible foot ring, subtle sheen highlight and a grounded contact shadow; consistent perspective. Minor flatness of the interior band and a slightly stiff rim highlight keep it below excellent." }, "broth": { "score": 0.6, "reason": "Warm brown surface with a soft central glow, gradient shading and a few oil-speck dots; however, toppings sit largely pasted on top with minimal occlusion or immersion, and the surface reads as a fairly flat ellipse rather than a liquid with depth." }, "steam": { "score": 0.85, "reason": "Across desktop-1/2/3 and pointer frames the translucent wisps visibly change shape, position and fading, indicating continuous rising motion; wisps are irregular, do not obscure food, and the reduced-motion frames are appropriately static. Slight plume regularity is the only shortcoming." }, "composition": { "score": 0.85, "reason": "Centered bowl with readable focal dish, restrained dark atmospheric background, soft vignette and warm/cool contrast; no clutter, text or competing elements, and the mobile framing scales cleanly. Background glow is tasteful and does not substitute for food detail." }, "vegan": { "score": 1, "reason": "Visible toppings are broccoli, tofu, mushrooms, nori, scallions and noodles in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.45, "reason": "Noodles are thin flat beige strands laid in mostly straight, near-parallel runs across the surface; they read as rigid wires/ribbons pasted on the broth with uniform width, little bend or volume, and abrupt ends rather than supple immersed strands." }, "toppings": { "score": 0.55, "reason": "Broccoli, two tofu blocks, scored shiitake-like mushroom caps and scallion rings are recognizable, but the tofu is a plain featureless block and the nori is a thick rigid grid-textured rounded card standing vertically, which reads as a plastic card rather than seaweed." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a clean elliptical rim, consistent above-angle perspective, smooth tapered body, subtle left-side highlight, small foot and grounded contact shadow; only minor stylization and the nori's seam at the rim." }, "broth": { "score": 0.7, "reason": "Rich brown broth surface with warm central glow, darker edges and scattered oil speckles gives believable liquid depth; however noodles sit visibly on top with no immersion cues and toppings contact is only lightly suggested." }, "steam": { "score": 0.7, "reason": "Multiple translucent wavy wisps rise above the bowl and clearly change shape and position between desktop-1/2/3 and pointer frames, confirming animated motion; they are thin line-like squiggles rather than full volumetric wisps and slightly sparse." }, "composition": { "score": 0.85, "reason": "Balanced centered focal dish, attractive warm broth glow against dark atmospheric backdrop, rising steam adds depth, no text or clutter; only the tall rigid nori card slightly competes at the rim." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are broccoli, tofu, mushrooms, nori, scallions and plant broth; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin uniform cream strokes arcing flatly across the broth with no volume, layering or sheen; several strands end abruptly mid-surface, reading as flat wires laid on top rather than a supple noodle mass." }, "toppings": { "score": 0.6, "reason": "Shiitake slices with radial cut marks, tofu cubes, scallion rings and a stylized green vegetable are recognizable, but the nori is a rigid flat rectangle with a mechanical grid pattern standing behind the rim like a card, and the tofu blocks are plain featureless cubes." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim with inner cavity, smooth tapering walls, subtle left-side highlight, speckled ceramic texture and a grounded foot with contact shadow; no distortion, though the material is very dark and simplified." }, "broth": { "score": 0.65, "reason": "Warm reddish-brown surface with a central glow and scattered oil/chili specks reads as liquid, but ingredients sit with minimal contact shadows and a few green flecks stray onto the bowl exterior, weakening immersion." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise and fade without obscuring food; plume shapes clearly differ across the time-ordered desktop frames, indicating continuous animation, while the reduced-motion frames are consistently static as expected." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a restrained dark atmosphere with a soft warm halo, balanced proportions, coherent lighting and no text, UI or clutter; presentation is clean though quite dim overall." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, nori, scallions and noodles in plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 433.3000000000029, "frameIntervalP95Ms": 433.3000000000029, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "bb65f71048b4ccdd48b0500b822b4bda9a285277236f223fd98ab0f5a2ae9a15", "gradedAt": "2026-09-30T23:12:41.403619+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.56, "toppings": 0.5144444444444445, "bowl": 0.7455555555555555, "broth": 0.5055555555555555, "steam": 0.7144444444444444, "composition": 0.8033333333333333 }, "reward": 0.6277777777777778, "runId": "openai/gpt-5.6-terra/high", "rewardStdDev": 0.05570989672035478, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.42666666666666664, "bowl": 0.78, "broth": 0.4666666666666667, "steam": 0.68, "composition": 0.82 }, "reward": 0.5893333333333334, "sampleStdDev": 0.0022546248764114206, "judgments": [ { "noodles": { "score": 0.48, "reason": "Golden continuous strands are readable, but densely repeated parallel waves form a flat, rectangular noodle mat with little convincing overlap or volume." }, "toppings": { "score": 0.42, "reason": "Varied green pieces, pale tofu-like pieces, brown mushroom-like ovals and nori are present, but polygonal greens, featureless rectangles, simple disks and straight-edged seaweed dominate." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, rounded walls and shaded underside form a coherent supported vessel. Warm gradients suggest lacquer, though rim highlights are flat and the lower decorative strokes look loosely applied." }, "broth": { "score": 0.46, "reason": "A warm amber surface and scattered oil-like dots suggest soup, but little liquid sheen or depth is visible, and ingredients mostly appear laid on top rather than immersed." }, "steam": { "score": 0.68, "reason": "Translucent plumes visibly change shape and height across normal frames and fade upward without hiding the dish. Their narrow, nearly parallel ribbon shapes and blunt ends remain mechanical." }, "composition": { "score": 0.82, "reason": "Warm orange and gold establish a clear focal bowl against a restrained dark green background, with generous space for steam and no distracting UI. Mobile framing preserves the complete dish; lighting is somewhat flat." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, greens, mushrooms and seaweed with noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Golden strands are readable and continuous, but repeated parallel waves form a flat rectangular mat with little convincing overlap or immersed volume." }, "toppings": { "score": 0.43, "reason": "A varied set of green vegetables, tofu-like pieces, mushroom-like slices and nori is suggested, but polygonal green lumps, plain rectangles, brown disks and a rigid striped seaweed card dominate." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, rounded walls and shaded underside form coherent bowl geometry with a grounding shadow. Material highlights are limited, and the lower decorative strokes look loosely applied." }, "broth": { "score": 0.48, "reason": "Warm amber color and small reddish oil spots suggest soup, but the surface is largely flat and lacks convincing sheen, depth or ingredient immersion." }, "steam": { "score": 0.68, "reason": "Translucent plumes change height and curvature across normal frames and fade upward without obscuring the food. They remain mostly narrow, similarly shaped ribbons with conspicuous blunt ends and little spreading." }, "composition": { "score": 0.82, "reason": "The single warm-colored bowl is clearly framed against a restrained dark green background, with ample space for steam and no distracting text or UI. Mobile framing preserves the dish, though substantial empty space makes it feel small." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, greens, mushrooms and seaweed alongside noodles; no animal-derived ingredient is apparent. No material vegan-compliance defect is observed." } }, { "noodles": { "score": 0.48, "reason": "Continuous golden strands are readable, but densely repeated parallel waves form a flat, rectangular mat with little convincing overlap or submerged volume." }, "toppings": { "score": 0.43, "reason": "Green garnish, pale tofu-like pieces, brown mushroom-like ovals and nori provide variety, but polygonal green lumps, flat rectangles and simple disks dominate; organic detail is sparse." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, rounded walls and shaded underside form a coherent supported vessel. Warm shading suggests lacquer, though the rim is uniformly graphic and the lower decoration looks loosely scribbled." }, "broth": { "score": 0.46, "reason": "Warm amber color reads as soup, but the exposed surface is largely flat and lacks liquid sheen or depth; ingredients have little visible immersion or contact." }, "steam": { "score": 0.68, "reason": "Translucent plumes change position and curvature across normal frames and fade upward without obscuring the food. Their mostly upright, broad ribbon shapes and blunt ends remain mechanical rather than softly dispersing." }, "composition": { "score": 0.82, "reason": "A single warm-colored bowl is clearly framed against a restrained dark green background, with subtle grounding shadow and no distracting text or UI. Mobile framing preserves the dish, though generous empty space and an oversized steam column weaken balance slightly." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly represent tofu, green vegetables, mushrooms and seaweed alongside noodles; no animal-derived ingredient is apparent. No material visual compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5833333333333334, "toppings": 0.5, "bowl": 0.7066666666666667, "broth": 0.43333333333333335, "steam": 0.68, "composition": 0.74 }, "reward": 0.6023333333333334, "sampleStdDev": 0.011846237095944555, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles form wavy yellow strands with overlap and a dense mass, but they are fairly uniform and flat-looking, like repeated zigzag bands, and sit in a neat rectangular patch with hard cut edges at left and right." }, "toppings": { "score": 0.5, "reason": "Toppings are recognizable (bok choy-like leaves, tofu cubes, mushroom slices, scallion sticks, nori sheet), but they are small, simple and flat; the tofu reads as plain cards and the nori as a rigid rectangle." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, curved walls, gradient lacquer shading and a decorative base pattern, and it sits on a soft shadow. The rim perspective is plausible; the noodle patch cut-off slightly undermines the interior, and the base looks a bit rounded and unsupported." }, "broth": { "score": 0.5, "reason": "The orange broth is visible only at the rim edge and is mostly covered by noodles. It has a pleasant warm gradient, but little surface sheen or depth, and the noodle patch looks pasted on top rather than immersed." }, "steam": { "score": 0.68, "reason": "Translucent wispy strands rise and change shape between frames, spreading and fading. They look somewhat like thin stringy lines and are similar in style, but they do not hide the food." }, "composition": { "score": 0.75, "reason": "The bowl is centered as a clear focal point on a dark moody background with subtle particles and warm colors. There is no distracting text or clutter, and the layout is balanced with a little empty space." }, "vegan": { "score": 0.9, "reason": "Visible toppings are tofu, mushrooms, greens, scallions and nori with noodles; no meat, egg or fish is visible. Small red dots are ambiguous chili oil." } }, { "noodles": { "score": 0.55, "reason": "Noodles are wavy yellow strands with overlap and some volume, but they form a flat, repetitive, clipped rectangular mat with hard edges at left and right. They look like a pattern rather than supple strands sitting in broth." }, "toppings": { "score": 0.5, "reason": "Toppings include bok choy-like leaves, scallion sticks, tofu cubes, mushroom slices and a nori sheet. They are small, simple and scattered. The tofu reads as flat cards and the nori as a stiff rectangle. They are recognizable but generic." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, tapered walls, a rounded base and a warm gradient with a subtle noodle motif. The rim and body perspective agree. The base lacks a foot or contact detail and the shading is fairly smooth." }, "broth": { "score": 0.4, "reason": "Little broth is visible, as noodles cover nearly the whole surface. The orange glow at the edges reads as a flat gradient, and the ingredients look pasted on top rather than immersed." }, "steam": { "score": 0.68, "reason": "Translucent thin wisps rise in varying shapes and change between frames, fading upward without hiding the food. They look somewhat thin and stringy, and the plumes are fairly similar to each other." }, "composition": { "score": 0.75, "reason": "A centered bowl on a dark atmospheric background with a warm glow and subtle particles. The food is the clear focus, with no clutter or text. The colors are attractive, though the bowl sits a bit small and the scene is plain." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are noodles, greens, tofu, mushrooms and nori, all plant-based. No meat or egg is visible. The small red dots are ambiguous but read as chili." } }, { "noodles": { "score": 0.6, "reason": "Wavy yellow strands overlap with a believable crinkled look, but they are regular, repetitive and flat, and they end in a hard rectangular edge at both sides rather than filling the bowl naturally." }, "toppings": { "score": 0.5, "reason": "Nori, bok choy leaves, scallion sticks, tofu cubes and mushroom slices are all identifiable. They are small and sparse, with simple flat shapes. The tofu reads as flat cards and the nori as a rigid rectangle, so the toppings feel generic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a smooth curved body with a lacquer gradient, and a dark base shadow. The decorative noodle pattern is a nice touch. Minor flaws are a slightly flat base and the rim and inner wall being a bit thick. No major perspective break." }, "broth": { "score": 0.4, "reason": "The broth is barely visible. Noodles cover nearly the whole surface, leaving only an orange gradient ring around them. There is little sense of liquid depth or sheen, and the ingredients look laid on top rather than immersed." }, "steam": { "score": 0.68, "reason": "Thin translucent wisps rise from the bowl and change shape across frames. They spread and fade without hiding the food. They are somewhat thin and uniform, like curved lines, but they read as steam." }, "composition": { "score": 0.72, "reason": "The bowl is centered on a dark, moody background with a warm glow and faint specks. The food is the clear focus, and there is no distracting text. The warm orange palette contrasts well with the dark surroundings. The bowl is somewhat small in the frame." }, "vegan": { "score": 0.9, "reason": "Only noodles, greens, tofu, mushrooms and nori are visible. There are no egg, meat or fish. The small red dots are ambiguous but look like chili oil." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6166666666666667, "toppings": 0.6166666666666667, "bowl": 0.75, "broth": 0.6166666666666667, "steam": 0.7833333333333333, "composition": 0.85 }, "reward": 0.6916666666666668, "sampleStdDev": 0.03971250852481287, "judgments": [ { "noodles": { "score": 0.6, "reason": "Stylized wavy strands read as noodles in flat-illustration style, but the noodle bed is an obvious rectangular pattern patch with repeating hatching and fringed cut edges; strands lack volume, coherent overlap and believable dips into broth, reading more as a texture fill than supple individual noodles." }, "toppings": { "score": 0.65, "reason": "Good variety and readability: a striped nori sheet, tilted tofu cubes, mushroom slices with stem line, green chive sticks and chili-oil specks are all recognizable. Defects: the green polygonal blobs are ambiguous featureless chunks and the tofu rectangles are plain flat slabs; shapes are geometric rather than organic, though acceptable within the flat style." }, "bowl": { "score": 0.7, "reason": "Coherent ellipse rim, dark rim band, visible inner wall, smooth gradient body, plausible depth and a grounded contact shadow; perspective is consistent. Defect: the scratchy stray stroke lines across the lower body look like accidental scribbles rather than intentional glaze texture." }, "broth": { "score": 0.5, "reason": "Broth is only implied by a warm radial glow behind the noodle patch; there is no distinct liquid surface with sheen or texture, and toppings appear pasted onto the flat noodle plane rather than sitting immersed, with chili specks floating ambiguously." }, "steam": { "score": 0.75, "reason": "Time-ordered frames show clearly animated steam: wisp shapes and positions change substantially across desktop-1/2/3 and pointer frames, translucent gray irregular wisps rise and fade without obscuring the food. Some plumes are thin, straight vertical strands and one frame shows an odd loop shape, slightly mechanical." }, "composition": { "score": 0.85, "reason": "Well-balanced dark atmospheric scene: bowl centered as clear focal point, warm bowl glow against cool dark background, subtle star specks and soft ground shadow, no text, UI or clutter; mobile framing also centers the dish cleanly. Background glow supports rather than substitutes for the dish." }, "vegan": { "score": 1.0, "reason": "Visible toppings are nori, tofu cubes, greens/chives, mushrooms and chili oil over noodles and plant broth \u2014 no meat, fish, egg or dairy anywhere." } }, { "noodles": { "score": 0.5, "reason": "Noodles render as a uniform field of repeated wavy squiggle motifs; mostly flat pattern with no strand continuity, volume or coherent overlap, though the right-side strands with visible broth gaps read acceptably." }, "toppings": { "score": 0.55, "reason": "Good variety (tofu rectangles, scallion rods, brown mushroom slices with pale rims, chili-oil dots, a striped nori sheet), but the numerous angular faceted green blobs look crystalline/plastic rather than organic and the tofu pieces are flat featureless cards." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with thick lip, smooth red gradient body, warm inner rim glow, consistent perspective and grounded contact shadow; the scratchy base decoration is a minor stylistic quirk." }, "broth": { "score": 0.65, "reason": "Golden surface with scattered chili-oil dots and back-glow depth; toppings sit on the surface plausibly, but the broth is nearly fully occluded by the noodle field and reads somewhat solid where visible." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the translucent gray wisps change shape, rise and fade, indicating animated motion; wisps are irregular and never hide the food, though a few read as thin rope-like curves." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with warm red/orange against dark atmospheric backdrop, subtle speckles, soft under-glow and shadow; balanced and uncluttered on desktop and mobile." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, scallions, mushroom slices, nori and chili oil in plant broth; no meat, fish, egg or dairy forms are apparent." } }, { "noodles": { "score": 0.75, "reason": "Wavy yellow strands with wavy line texture read as noodles with plausible bends; slightly uniform and patterned, but coherent with no broken scribbles." }, "toppings": { "score": 0.65, "reason": "Tofu rectangles, mushroom slices, scallion segments and a ridged nori sheet are recognizable, but broccoli forms are faceted polygonal blobs and tofu is flat card-like, giving an artificial feel." }, "bowl": { "score": 0.75, "reason": "Coherent red bowl with rim, inner cavity, wall shading, base and ground shadow; perspective is consistent, though the scratch decoration on the lower body is odd and rim is quite flat." }, "broth": { "score": 0.7, "reason": "Warm glowing broth surface with subtle drips of chili oil; toppings sit within the surface reasonably, though the surface reads somewhat flat and pattern-like at the edges." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rising above the bowl that change shape, position and length across the time-ordered frames, indicating continuous motion without obscuring the food." }, "composition": { "score": 0.85, "reason": "Well-centered focal dish, dark atmospheric background with faint sparks and grounding shadow, coherent warm lighting, no distracting UI or clutter." }, "vegan": { "score": 0.95, "reason": "Visible toppings are tofu, mushrooms, scallions, greens and seaweed with plant broth; no animal-derived ingredients apparent." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9611111111111111, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 2199.9000000000233, "frameIntervalP95Ms": 2199.9000000000233, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "f71c926b40aa63b7355b89b59313a8fb795e29475876812225a50941201a00a0", "gradedAt": "2026-09-30T23:13:35.197883+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.33555555555555555, "toppings": 0.3522222222222222, "bowl": 0.5633333333333334, "broth": 0.3888888888888889, "steam": 0.5700000000000001, "composition": 0.6755555555555556 }, "reward": 0.4597777777777778, "runId": "openai/gpt-5.6-terra/low", "rewardStdDev": 0.09078061671067442, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.34, "toppings": 0.30666666666666664, "bowl": 0.47333333333333333, "broth": 0.4, "steam": 0.5766666666666667, "composition": 0.6433333333333333 }, "reward": 0.4301666666666667, "sampleStdDev": 0.007251436639269009, "judgments": [ { "noodles": { "score": 0.34, "reason": "Golden strands are readable, but tightly repeated concentric spirals form a flat maze-like pattern rather than supple, overlapping noodles." }, "toppings": { "score": 0.3, "reason": "Green leaf-like shapes and colorful vegetable-like pieces provide variety, but featureless disks, smooth wedges and rigid ovals dominate; organic texture and ingredient details are largely absent." }, "bowl": { "score": 0.48, "reason": "Warm shading gives the rounded body volume, but the raised rear rim and inset soup ellipse join awkwardly, and the dark foot appears attached to the front wall rather than supporting the underside." }, "broth": { "score": 0.4, "reason": "The orange-brown surface is clearly delineated, but reads as a solid matte ellipse with little liquid sheen or depth; toppings lack convincing immersion and contact." }, "steam": { "score": 0.58, "reason": "Faint translucent columns change position and visibility across normal frames without obscuring the food. Their mostly straight, similarly shaped forms show limited irregular curling or spreading." }, "composition": { "score": 0.64, "reason": "A single centered dish, warm colors and uncluttered dark surroundings create a clear focal point. The oversized bowl body outweighs the food, and the mobile view crowds the bowl against both edges." }, "vegan": { "score": 0.87, "reason": "Visible noodles, green vegetable-like shapes and red and beige pieces plausibly suggest plant-based ramen, with no unmistakable animal ingredients. The generic beige shapes leave some ingredient ambiguity." } }, { "noodles": { "score": 0.32, "reason": "Golden strands are readable, but tightly repeated concentric spirals form a flat wire-like pattern rather than supple, overlapping noodle volume." }, "toppings": { "score": 0.3, "reason": "Green leaf-like shapes and colorful pieces provide variety, but featureless tan blocks, red disks, and rigid green ovals dominate; organic texture and ingredient detail are largely absent." }, "bowl": { "score": 0.46, "reason": "Warm shading gives the rounded body volume, but the raised rear rim and broad black opening join awkwardly, while the dark base appears attached to the front wall rather than supporting the bowl." }, "broth": { "score": 0.4, "reason": "The orange surface is clearly bounded and warm-colored, but reads as a solid shallow oval with little liquid sheen or depth; ingredients show minimal immersion or contact." }, "steam": { "score": 0.58, "reason": "Faint translucent plumes change position and visibility across normal frames without obscuring the food, but remain narrow, nearly parallel columns with little irregular spreading." }, "composition": { "score": 0.65, "reason": "A single centered dish and restrained warm-on-dark palette create a clear focus without UI clutter. The oversized bowl body outweighs the small food area, and the mobile framing clips the bowl edges." }, "vegan": { "score": 0.88, "reason": "Visible greens, red vegetable-like pieces, noodles, and tan plant-compatible chunks plausibly form vegan ramen. No clear animal ingredient is visible, though the simplified tan toppings are ambiguous." } }, { "noodles": { "score": 0.36, "reason": "Thin golden strands are visible, but densely repeated concentric loops resemble a flat spiral pattern rather than supple, overlapping noodles with volume." }, "toppings": { "score": 0.32, "reason": "Green leaf-like shapes and colorful pieces offer variety, but featureless red disks, smooth tan blocks, and rigid green ovals dominate; organic texture and ingredient details are largely absent." }, "bowl": { "score": 0.48, "reason": "Warm shading gives the rounded body some material depth, but the raised rear rim and black inner band join awkwardly, while the apparent foot looks attached to the front wall rather than supporting the bowl." }, "broth": { "score": 0.4, "reason": "The reddish surface is readable as soup, but its smooth, solid-looking gradient lacks liquid sheen and depth; toppings show little convincing immersion or contact." }, "steam": { "score": 0.57, "reason": "Faint translucent columns change position and visibility across normal frames and leave the food readable. They remain unusually straight, similarly shaped streaks with limited irregular spreading." }, "composition": { "score": 0.64, "reason": "A single warm-colored dish forms a clear focal point against a restrained dark background with no distracting UI. The oversized bowl body outweighs the food, and the mobile view crops its sides and weakens table contact." }, "vegan": { "score": 0.86, "reason": "Visible noodles, green vegetables, red vegetable-like pieces, and tan plant-compatible chunks plausibly form vegan ramen. No unmistakable animal ingredient is visible, although several toppings are too generic to identify confidently." } } ], "diagnostics": { "vegan": 0.87 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.3333333333333333, "bowl": 0.4166666666666667, "broth": 0.4166666666666667, "steam": 0.4166666666666667, "composition": 0.5666666666666667 }, "reward": 0.3875, "sampleStdDev": 0.013228756555322943, "judgments": [ { "noodles": { "score": 0.3, "reason": "Noodles are flat concentric rings in a small patch, reading as a maze or fingerprint pattern rather than supple strands. There is no overlap or volume, and they look rigid and scribbled." }, "toppings": { "score": 0.35, "reason": "There is a green leaf, a tofu-like block, scallion-like ellipses and a red blob cluster. The shapes are simple flat vector forms with little texture or detail, and the red cluster is ambiguous. They are small but recognizable." }, "bowl": { "score": 0.45, "reason": "The bowl has a glossy lacquer gradient and a base, but the rim perspective is inconsistent with the body. The handle-like ears stick out at the rim and the floating dark foot piece looks detached. The top ellipse is incompatible with the side view." }, "broth": { "score": 0.45, "reason": "The broth is an orange-red surface with a slight gradient and a dark band around it. It reads as flat and painted, and the ingredients sit on top rather than being immersed." }, "steam": { "score": 0.4, "reason": "Faint thin vertical wisps drift and change between frames, and they do not obscure the food. They are very faint, narrow and streak-like rather than billowing, and look somewhat mechanical." }, "composition": { "score": 0.6, "reason": "The bowl is centered and is the clear focal point against a dark, moody background with warm lighting. The proportions are acceptable, but the bowl is stiff and the food area is small relative to the bowl body." }, "vegan": { "score": 0.85, "reason": "Visible toppings are tofu, leaf, scallion and mushroom-like shapes, with no meat, egg or fish. The red cluster is slightly ambiguous." } }, { "noodles": { "score": 0.3, "reason": "Noodles are flat concentric ring scribbles in a small patch, reading as a maze or fingerprint pattern rather than supple strands. There is no volume or overlap." }, "toppings": { "score": 0.35, "reason": "There is a green leaf, a tofu-like rounded block, scallion-like ellipses and red blobs. They are simple flat shapes, small and generic. The mushroom is barely readable and there is no nori." }, "bowl": { "score": 0.4, "reason": "The bowl has a nice gradient and highlight, but the rim is an odd dark band with mismatched ear-like handles. The perspective is inconsistent, with a wide top ellipse on a deep side view, and the foot is a detached dark slab." }, "broth": { "score": 0.4, "reason": "The broth is a smooth orange-red disc with a soft gradient and little texture or depth. The toppings sit on top of it rather than in it, and it looks like a flat disc." }, "steam": { "score": 0.4, "reason": "Faint thin vertical wisps drift and change between frames, so there is some motion. They are narrow, streak-like and mechanical rather than billowy." }, "composition": { "score": 0.55, "reason": "The bowl is centered and readable in a dark, moody scene with warm color and no clutter. The food is small relative to the bowl, and the bowl is awkwardly shaped." }, "vegan": { "score": 0.9, "reason": "Only vegetables, tofu-like and mushroom-like items and noodles are visible. No animal products." } }, { "noodles": { "score": 0.3, "reason": "Noodles are flat concentric ring scribbles in a small patch, reading as a maze or fingerprint pattern with no supple continuous strands or volume." }, "toppings": { "score": 0.3, "reason": "The toppings are a green leaf shape, a tofu-like blob, three flat green blades and red circles. They are recognizable as symbols but flat and generic, with little texture and few readable details." }, "bowl": { "score": 0.4, "reason": "The bowl has a warm lacquer gradient and a foot, but the rim perspective is inconsistent. The flat top ellipse and dark band sit oddly against the body, and the side handles look detached." }, "broth": { "score": 0.4, "reason": "The broth is a flat orange disc with a modest sheen. The toppings look pasted on top rather than immersed, with little depth or occlusion." }, "steam": { "score": 0.45, "reason": "Faint thin wisps shift position between frames, but they are narrow, vertical and mechanical, and they look like streaks rather than irregular spreading steam." }, "composition": { "score": 0.55, "reason": "The bowl is a centered focal point on a dark, atmospheric background with warm colors and no clutter. The food is small relative to the bowl and the bowl itself has flaws." }, "vegan": { "score": 0.85, "reason": "The visible toppings are tofu, greens and mushroom-like items. No meat or egg is visible." } } ], "diagnostics": { "vegan": 0.8666666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.3666666666666667, "toppings": 0.4166666666666667, "bowl": 0.8, "broth": 0.35, "steam": 0.7166666666666667, "composition": 0.8166666666666667 }, "reward": 0.5616666666666668, "sampleStdDev": 0.03907791362564452, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are rendered as flat concentric ring/spiral grooves on the surface rather than continuous strands with volume; repeated flat loops dominate, which the rubric caps at .50 despite the animation of the spiral." }, "toppings": { "score": 0.45, "reason": "Recognizable flat vector shapes: a two-tone leaf, green scallion-like stalks, red/orange disks, and ambiguous tan/beige blobs; shapes are organic but very simple, featureless disks dominate, and items look pasted on with no texture." }, "bowl": { "score": 0.8, "reason": "Coherent rounded ceramic form with a readable rim ellipse, dark interior band suggesting thickness, warm gradient shading, small rim ears, and a foot resting on the ground; only minor stylization like the emissive rim glow and inverted-looking shadow crescent." }, "broth": { "score": 0.4, "reason": "Broth reads as a flat patterned disc with concentric target-like rings; toppings sit on top with no occlusion, meniscus, or believable immersion, and depth/sheen are minimal." }, "steam": { "score": 0.8, "reason": "Across the time-ordered frames the translucent S-curved wisps shift position and visibility (single faint plume at 1s, three distinct plumes by 2.5s), indicating rising, drifting steam that stays subtle and never obscures the dish." }, "composition": { "score": 0.85, "reason": "Clean centered focal bowl on a dark atmospheric ground with warm rim glow, restrained palette, no clutter or text; mobile crop keeps the dish readable, only slight edge crowding on narrow viewports." }, "vegan": { "score": 0.75, "reason": "Visible toppings (leaf, scallion stalks, red disks, cream stick) are plausibly plant-based, but the large tan blob is ambiguous and could be read as an egg or dumpling, warranting an intermediate score." } }, { "noodles": { "score": 0.4, "reason": "Noodles are rendered as flat, uniform golden concentric wavy rings filling the bowl interior \u2014 a maze/target-loop pattern rather than supple continuous strands with volume; no believable strand thickness, immersion or overlap." }, "toppings": { "score": 0.4, "reason": "A green leaf reads as a leaf, but the remaining toppings are generic flat forms: two rounded green capsules, a featureless beige blob, a tan lump and clustered glossy red/orange dots \u2014 lacking organic detail, texture and clear identity as food ingredients." }, "bowl": { "score": 0.8, "reason": "Coherent stylized ceramic bowl: clean elliptical rim with dark rim-thickness ring, smooth terracotta gradient walls with highlight, visible foot and grounded contact shadow; perspective between rim ellipse and body is consistent with only minor softness at the rim joins." }, "broth": { "score": 0.35, "reason": "Broth reads as a solid orange ellipse decorated with concentric golden ripple rings \u2014 a flat patterned disc rather than liquid; toppings sit on top of the pattern with weak occlusion or believable immersion." }, "steam": { "score": 0.75, "reason": "Translucent curved wisps change position and shape across the time-ordered desktop frames (different plume layouts at ~1.0s, 2.5s, 4.0s and pointer frames), indicating animated rising steam that fades without hiding food; wisps are somewhat simple repeated curves. Reduced frames correctly hold static." }, "composition": { "score": 0.8, "reason": "Balanced centered focal dish on a dark atmospheric background with warm rim glow and simple table curve; no clutter or competing elements. Only minor defect: a tiny stray orange dot artifact in the top-left corner." }, "vegan": { "score": 0.75, "reason": "Visible toppings (leaf, green capsules, red dots, beige and tan blobs) are plausibly plant-based; no unmistakable animal ingredients, though the plain beige blob is ambiguous in identity." } }, { "noodles": { "score": 0.3, "reason": "Noodles are rendered as a flat golden concentric spiral/target pattern on the broth surface rather than supple continuous strands; it reads as a decorative maze disc, the exact failure mode the rubric penalizes." }, "toppings": { "score": 0.4, "reason": "Toppings are simple flat vector shapes: one green leaf, three plain green ellipses, a featureless cream blob, and flat red/tan disks with a small cylinder. Recognizable as stylized vegetables but dominated by generic featureless disks and blobs with no texture or readable detail." }, "bowl": { "score": 0.8, "reason": "The bowl is well constructed: coherent elliptical rim, dark interior band, smooth glossy terracotta body with gradient shading and specular highlight, visible foot/contact on a dark table. Slightly toy-like gloss, but no broken geometry or perspective mismatch observed." }, "broth": { "score": 0.3, "reason": "The broth is a solid flat orange-red disc with no liquid sheen, depth, or meniscus; the noodle spiral sits on top like a decal and toppings appear pasted onto the surface with minimal immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise from the bowl and change shape/position across desktop-1 through desktop-3, indicating animation; however the wisps are drawn as narrow curvy lines rather than irregular volumetric plumes, slightly mechanical in form." }, "composition": { "score": 0.8, "reason": "Balanced centered composition on a dark atmospheric ground with warm rim glow, steam leading the eye to the dish, no text, UI or clutter. Clean and readable, though the dish's interior detail is sparse." }, "vegan": { "score": 1.0, "reason": "Visible toppings are leafy greens, scallion-like ellipses, a tofu-like pale mass, red vegetable disks and a bamboo-like cylinder; no meat, fish, egg or dairy indicators appear." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8566666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.400000000000546, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8614089f757a3d0721203227d6a5d84cb10f5b0e0e2797dd3ffbc3dbc30e2c86", "gradedAt": "2026-09-30T23:11:13.369893+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5766666666666667, "toppings": 0.6511111111111111, "bowl": 0.7977777777777778, "broth": 0.5688888888888889, "steam": 0.648888888888889, "composition": 0.8144444444444444 }, "reward": 0.6733333333333333, "runId": "openai/gpt-5.6-terra/max", "rewardStdDev": 0.07516722098834772, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.58, "toppings": 0.6, "bowl": 0.8566666666666667, "broth": 0.5133333333333333, "steam": 0.6466666666666667, "composition": 0.88 }, "reward": 0.6736666666666666, "sampleStdDev": 0.0002886751345948452, "judgments": [ { "noodles": { "score": 0.58, "reason": "Visible golden strands have rounded bends and readable overlap, but are sparse, broad and ribbon-like rather than forming a supple noodle volume." }, "toppings": { "score": 0.61, "reason": "Mushroom caps and stems, shaded tofu cubes and veined green leaves are recognizable. Texture is minimal; scallions look like rigid strips, corn like uniform disks, and nori like a thick rectangular card." }, "bowl": { "score": 0.85, "reason": "A coherent elliptical rim, shaded curved walls, consistent cavity and grounded contact shadow convincingly establish a lacquer bowl. The lower contour is slightly angular, but no material construction defect is observed." }, "broth": { "score": 0.51, "reason": "Warm brown shading establishes a soup surface within the rim, but there is little liquid sheen or surface texture, and ingredients mostly appear placed atop it with weak immersion cues." }, "steam": { "score": 0.64, "reason": "Normal frames show changes in the position and opacity of translucent curved wisps emerging behind the soup. They leave the food readable, but remain similarly narrow parallel strokes with little spreading or diffuse fading." }, "composition": { "score": 0.88, "reason": "The single dish is clearly framed on desktop and mobile, with warm orange and green accents against a restrained dark background. Soft lighting supports the focal bowl without text or clutter; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen, with no discernible animal-derived topping." } }, { "noodles": { "score": 0.58, "reason": "Several continuous golden strands bend and overlap behind toppings, but they are sparse, broad and ribbon-like, with little convincing tangled volume." }, "toppings": { "score": 0.6, "reason": "Mushroom caps and stems, shaded tofu cubes and veined green leaves are readable. Details remain simplified; scallions are rigid strips, corn resembles uniform disks, and nori is an unusually regular rectangular card." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, thick lip, curved shaded walls and grounded contact shadow convincingly establish a lacquer-like bowl. No material construction defect is observed; the base and highlights are somewhat simplified." }, "broth": { "score": 0.52, "reason": "Warm brown shading establishes a soup surface inside the rim, but little liquid sheen or surface texture is visible, and most ingredients appear laid on top rather than partially immersed." }, "steam": { "score": 0.64, "reason": "Normal frames show changes in wisp positions and opacity, with translucent trails originating near the soup and leaving food readable. The evenly spaced, similarly narrow waving columns feel mechanical rather than spreading naturally." }, "composition": { "score": 0.88, "reason": "The single warm-colored dish remains a clear focal point against a restrained dark background, with ample breathing room, coherent shading and no distracting text or UI. Mobile framing preserves the complete dish; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn-like kernels and seaweed plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.58, "reason": "Golden strands have continuous curved bends and readable overlaps, but are sparse, broad and ribbon-like, with little convincing tangled volume or immersion." }, "toppings": { "score": 0.59, "reason": "Mushroom caps and stems, green leaves, tofu cubes and corn are recognizable and varied. Texture is minimal; scallions resemble rigid bars and nori is a thick, regular rectangular card." }, "bowl": { "score": 0.86, "reason": "The elliptical rim, inner cavity and curved walls form coherent depth and perspective. Warm highlights, dark side shading and a contact shadow convincingly support the ceramic bowl; the bottom contour is slightly angular." }, "broth": { "score": 0.51, "reason": "Warm brown shading establishes a soup surface inside the rim, but there is little liquid sheen or surface detail, and ingredients mostly appear laid on top rather than partially immersed." }, "steam": { "score": 0.66, "reason": "Normal frames show changes in wisp positions and opacity, with translucent trails emerging behind the soup without obscuring it. The similarly thick, parallel sinusoidal trails look mechanical and do not visibly spread much." }, "composition": { "score": 0.88, "reason": "A single clearly readable dish anchors the restrained dark scene, with harmonious warm ceramic and green accents, ample breathing room and no distracting text or UI. Mobile framing retains the complete bowl; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn and seaweed with noodles; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.62, "bowl": 0.7033333333333334, "broth": 0.49333333333333335, "steam": 0.5333333333333333, "composition": 0.73 }, "reward": 0.5980000000000001, "sampleStdDev": 0.004330127018922165, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are cream-colored curved strands that read as noodles, but they are sparse, flat and partly hidden under toppings. There is little volume or overlap, and they look pasted on the surface." }, "toppings": { "score": 0.62, "reason": "Nori, bok choy leaves, scallion strips, shiitake mushrooms, tofu cubes, corn and chili bits are all recognizable and varied. The nori is a flat rectangle, the scallions are rigid sticks, and everything floats on top with stylized flatness." }, "bowl": { "score": 0.68, "reason": "The clay bowl has a coherent elliptical rim, a curved body, a foot, highlights and a contact shadow. The rim and inner wall work well and perspective is consistent. The form is somewhat simple and the base is slightly vague." }, "broth": { "score": 0.5, "reason": "The broth is a flat brown-red surface with only a subtle gradient and little sheen or ripple. Ingredients sit on it without much immersion or contact, so it reads as a solid disc." }, "steam": { "score": 0.55, "reason": "Wavy wisps change and fade across frames and drift, which reads as animated. They are thin, uniform line-like strands and look somewhat mechanical, but they stay translucent and do not hide the food." }, "composition": { "score": 0.72, "reason": "A centered bowl with a warm glow on a dark background gives a clear focal point and pleasant color contrast. The scene is clean with no clutter. The bowl is a bit small in frame." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, greens, corn, nori and noodles are all plant-based. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are pale, curved strands that read as noodles, but they are sparse, flat and partly hidden under toppings. They lack volume and overlap, and they do not sit convincingly in the broth." }, "toppings": { "score": 0.62, "reason": "Variety is good: nori, bok choy leaves, scallion strips, mushrooms, tofu cubes, corn kernels and chili bits. The shapes are flat vector icons and the nori is a rigid rectangle. The toppings are readable and stylized but simple." }, "bowl": { "score": 0.75, "reason": "The bowl has a coherent elliptical rim, an inner wall, a curved body and a foot with a ground shadow. Highlights give it a glazed look, and the perspective is consistent. The base is a bit vague and the rim is slightly flat." }, "broth": { "score": 0.48, "reason": "The broth is a flat brown-orange surface with little sheen or depth. Ingredients sit on top rather than being immersed, with no ripples or contact cues." }, "steam": { "score": 0.5, "reason": "Wavy steam lines move and fade between frames, and they start above the bowl. They look like thin, uniform strokes that resemble smoke wires rather than irregular translucent wisps." }, "composition": { "score": 0.75, "reason": "A centered, single focal bowl sits on a dark background with a warm glow, and the colors are harmonious. There is no clutter. The bowl is slightly small in the frame." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, vegetables, corn, nori and noodles are visible. No animal products appear." } }, { "noodles": { "score": 0.5, "reason": "Noodles are pale curved strands that read as noodles with some overlap, but they are sparse and thin, flat in style and partly hidden under toppings. There is no real volume or mass in the broth." }, "toppings": { "score": 0.62, "reason": "Nori, bok choy leaves, scallion strips, mushrooms, tofu cubes, corn and chili bits are all recognizable and varied. The nori is a stiff dark rectangle, the scallions are rigid sticks, and the tofu is blocky, so they look clip-art-like and slightly pasted on." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body, a foot and a contact shadow. Its glossy highlight and rim thickness are convincing, but the rim has a slight inner-lip oddity and the base is a little vague." }, "broth": { "score": 0.5, "reason": "The broth is a flat brown surface with a dark edge. It has little sheen, ripple or depth, and the ingredients sit on top of it with weak immersion." }, "steam": { "score": 0.55, "reason": "Wavy translucent wisps rise and change position and opacity between frames, so they are animated and they do not hide the food. They are thin, uniform line-like strands that look mechanical, and they do not spread or billow." }, "composition": { "score": 0.72, "reason": "The bowl is centered as a clear focal point in a dark atmospheric scene with a warm glow and a cohesive palette. The layout is balanced and uncluttered, though the bowl is somewhat small in frame." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, vegetables, nori, corn and noodles. No animal products are visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7333333333333333, "bowl": 0.8333333333333334, "broth": 0.7, "steam": 0.7666666666666667, "composition": 0.8333333333333334 }, "reward": 0.7483333333333333, "sampleStdDev": 0.03357206179747281, "judgments": [ { "noodles": { "score": 0.6, "reason": "Cream wavy strands have plausible bends and some variation, but they read as flat squiggle strokes pasted on the broth surface with little coherent mass, volume, or immersion; several end abruptly." }, "toppings": { "score": 0.7, "reason": "Good variety and readability: striped nori sheet, green bokchoy leaves with stems, brown shiitake caps with a cross-section slice, tofu cubes, corn-dot clusters, scallion sticks and chili flecks. Shapes are stylized cutouts; tofu cubes are somewhat geometric and corn reads as plain circles." }, "bowl": { "score": 0.8, "reason": "Coherent reddish ceramic bowl: thick elliptical rim with inner edge, tapering walls, supported base hidden in shadow, glossy highlight arcs on the body, and a grounded contact shadow; perspective is consistent, only mild stylized widening." }, "broth": { "score": 0.7, "reason": "Dark miso-toned liquid surface with subtle sheen and depth gradient inside the rim; ingredients mostly rest on it with faint occlusion, though some noodles and corn dots appear to sit atop rather than submerged." }, "steam": { "score": 0.7, "reason": "Translucent tan wisps rise from the soup, fade toward the top, and change shape/position across the desktop frames (1050ms, 3112ms, 5134ms, pointer frames), indicating real animation that stays still under reduced motion; wisps are a bit thin and uniform squiggle-like, slightly mechanical." }, "composition": { "score": 0.8, "reason": "Balanced centered focal dish in a dark atmospheric scene with a warm rim glow, soft vignette and sparse particles; no text, UI or competing elements. On mobile the bowl sits high with a large empty lower half." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-based: nori, mushrooms, bokchoy, tofu, corn, scallions, chili and noodles in broth; no animal-derived toppings visible." } }, { "noodles": { "score": 0.65, "reason": "Pale cream strands with plausible thickness and organic bends peeking around toppings and near the rim; readable as noodles but sparse, with some strands reading as simple repeated arcs rather than a coherent voluminous nest." }, "toppings": { "score": 0.75, "reason": "Good variety: a shiitake with cap and stem, leafy greens with mid-vein detail, beveled tofu cubes, corn kernels, chili flecks, scallion stalks and a dark nori sheet. The nori reads slightly as a flat card with faint ruled lines, the weakest element, but most ingredients have believable stylized shapes and scale." }, "bowl": { "score": 0.85, "reason": "Coherent wide elliptical rim with matching inner cavity, smoothly curving reddish lacquer walls, a subtle ceramic highlight streak on the left, consistent perspective, and grounded contact via a soft shadow beneath. No visible distortion or detached parts." }, "broth": { "score": 0.7, "reason": "Dark miso-toned liquid surface fills the cavity with a lighter ring near the rim and gentle tonal variation; toppings sit within it with slight contact shading. The surface reads a bit flat/solid with minimal sheen or depth cues." }, "steam": { "score": 0.8, "reason": "Translucent warm wisps rise from the bowl and clearly change shape and position across desktop-1, -2, -3 and pointer frames, indicating continuous rising/fading motion; they stay irregular and never hide the food. Wisps are slightly similar in shape to one another, a mild mechanical repetition. Reduced frames correctly freeze it." }, "composition": { "score": 0.85, "reason": "Centered focal bowl against a dark background with a warm behind-glow and ground shadow; balanced framing on desktop and mobile, attractive warm palette, no text, UI or clutter, and the dish remains the clear focus." }, "vegan": { "score": 1, "reason": "Visible ingredients are nori, mushrooms, leafy greens, tofu cubes, corn, chili, scallions and noodles in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Cream strands read as noodles with plausible bends and crossings around the toppings, but they are fairly sparse, flat-filled, lie mostly on the surface without clear broth immersion, and repeat similar S-curve motifs." }, "toppings": { "score": 0.75, "reason": "Good variety: readable nori sheet, shiitake with stem, brown cap, tofu cubes, bok-choy leaves, scallion stalks, corn clusters and chili bits. Shapes are organic enough, though tofu cubes and corn dots are quite geometric and simple." }, "bowl": { "score": 0.85, "reason": "Coherent rim ellipse, thick ceramic wall, cavity, smooth body with highlight arcs, believable low-angle perspective and grounded contact shadow/glow. Minor: small notch-like artifact on the lower rim in the first frame." }, "broth": { "score": 0.7, "reason": "Dark miso-toned surface with subtle gradient and rim sheen; ingredients sit with light contact shading, but the liquid reads somewhat flat and murky with little visible depth or sheen variation." }, "steam": { "score": 0.8, "reason": "Translucent tapered wisps that visibly change shape and position across desktop-1/2/3 and pointer frames, rising and fading without obscuring food; slightly repetitive squiggle forms. Reduced-motion frames are static, consistent with the mode." }, "composition": { "score": 0.85, "reason": "Centered dish as clear focal point, warm rim light and glow against dark background, balanced proportions, no clutter or text; mobile crop also holds together well." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-based: nori, mushrooms, tofu, leafy greens, scallions, corn, chili; no animal-derived toppings apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 99.94999999999982, "frameIntervalP95Ms": 166.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a6f80599626792e27cc91a55516b8df466b8ef6b7884ebb5088f0979412c3bae", "gradedAt": "2026-09-30T23:13:23.284671+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4811111111111111, "toppings": 0.4355555555555556, "bowl": 0.6988888888888889, "broth": 0.51, "steam": 0.7077777777777777, "composition": 0.7477777777777778 }, "reward": 0.5768333333333333, "runId": "openai/gpt-5.6-terra/medium", "rewardStdDev": 0.024444381313049764, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.46, "toppings": 0.4066666666666667, "bowl": 0.73, "broth": 0.43, "steam": 0.7066666666666667, "composition": 0.7566666666666667 }, "reward": 0.5643333333333334, "sampleStdDev": 0.0030550504633038962, "judgments": [ { "noodles": { "score": 0.46, "reason": "Thin continuous yellow strands are readable, but form a narrow band of repeated flat waves with little volume or convincing overlap; several ends extend over the right rim." }, "toppings": { "score": 0.42, "reason": "Green leaves have central veins and tofu-like pieces add variety, but uniformly oval leaves, concentric brown disks and flat yellow blocks remain generic ingredient symbols with little organic texture." }, "bowl": { "score": 0.73, "reason": "The elliptical opening, continuous thick rim and curved shaded body form a coherent supported bowl. Sparse material highlights and a heavy, nearly uniform dark rim limit refinement." }, "broth": { "score": 0.43, "reason": "Warm amber shading distinguishes the soup surface, but it reads as a smooth matte fill without liquid sheen, ripples or convincing immersion around the pasted-on ingredients." }, "steam": { "score": 0.7, "reason": "Normal frames show changing translucent curved wisps rising above the soup without obscuring it. The long, similarly narrow streaks have limited spreading and resemble blurred lines more than varied vapor." }, "composition": { "score": 0.76, "reason": "A single readable dish, restrained warm palette and uncluttered dark background provide a clear focus across desktop and mobile. The low placement and very tall steam column leave the presentation somewhat top-heavy in empty space." }, "vegan": { "score": 0.95, "reason": "Visible noodles, green leaves, tofu-like yellow pieces and brown mushroom-like toppings plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Continuous pale strands are readable, but repeated shallow waves form a flat horizontal band with little volume or convincing immersion; several ends extend over the right rim." }, "toppings": { "score": 0.4, "reason": "Green leaves have simple central veins, but uniformly elliptical shapes, brown concentric disks, and featureless tofu rectangles dominate rather than textured, organic ingredients." }, "bowl": { "score": 0.73, "reason": "The elliptical rim, curved shaded body, and contact shadow form a coherent supported bowl. The heavy dark rim and sparse material highlights make it somewhat graphic and shallow." }, "broth": { "score": 0.43, "reason": "Warm amber shading suggests soup, but the surface lacks liquid sheen or texture, and toppings appear pasted onto it without contact ripples or convincing submersion." }, "steam": { "score": 0.7, "reason": "Normal frames show changing translucent curved wisps rising above the soup without obscuring the food. The narrow, similarly shaped trails offer limited spreading and look somewhat like blurred strokes." }, "composition": { "score": 0.75, "reason": "A single readable dish stands out against a restrained dark, warm background with no UI or text. Excessive space above the low-positioned bowl and very tall steam slightly weaken the balance." }, "vegan": { "score": 0.94, "reason": "Visible noodles, green leaves, pale tofu-like pieces, and brown mushroom-like toppings plausibly form vegan ramen; no clearly animal-derived ingredient is visible." } }, { "noodles": { "score": 0.46, "reason": "Continuous yellow strands are readable, but form a narrow band of repeated flat waves with little volume or convincing immersion; several ends extend across the right rim." }, "toppings": { "score": 0.4, "reason": "Green leaves have simple center veins, but repeated oval shapes, brown concentric disks and flat cream rectangles remain generic ingredient symbols with little organic texture." }, "bowl": { "score": 0.73, "reason": "The elliptical rim, curved red walls and contact shadow form a coherent supported bowl. Restrained edge highlights help, though the heavy black rim and minimally shaded body limit material richness." }, "broth": { "score": 0.43, "reason": "Warm amber shading suggests soup, but the surface lacks liquid sheen or texture, and ingredients appear laid onto it rather than partly immersed." }, "steam": { "score": 0.72, "reason": "Translucent wisps change position and shape across normal frames and leave the food visible. The long, similarly narrow trails show limited spreading and rather uniform strand-like structure." }, "composition": { "score": 0.76, "reason": "A single readable dish sits against a restrained dark warm background without text or clutter, including on mobile. The tall steam column and large upper empty region make the arrangement somewhat bottom-heavy." }, "vegan": { "score": 0.93, "reason": "Visible noodles, green leaves, tofu-like cream pieces and mushroom-like brown toppings plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 0.94 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.45, "bowl": 0.6333333333333333, "broth": 0.55, "steam": 0.6666666666666666, "composition": 0.67 }, "reward": 0.5611666666666667, "sampleStdDev": 0.01850225211517057, "judgments": [ { "noodles": { "score": 0.5, "reason": "Fine golden strands with wavy bends and overlap read as noodles, but they form a thin flat band across the surface with little volume. They look like scribbled wire lines rather than a supple mass sitting in the broth." }, "toppings": { "score": 0.45, "reason": "Green leaves, tofu cubes and mushroom caps are recognizable, but they are small, sparse and simple. The mushrooms are flat brown discs and the tofu is plain blocks. There is no nori and little texture or variety." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a dark inner wall, a red gradient body and a contact shadow. Perspective is consistent. The body is smooth and plain with weak highlights, and the base is missing, so it looks a bit floaty." }, "broth": { "score": 0.55, "reason": "The warm orange radial gradient gives depth and looks appetizing. The surface is flat with little texture or sheen, and the toppings sit on top without ripples or occlusion." }, "steam": { "score": 0.68, "reason": "Translucent wispy strands rise from the bowl and change shape between frames. They fade out toward the top and don't hide the food. The strands look somewhat similar to each other and thin." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, atmospheric scene with a warm glow and sparse particles, and it stays the clear focus. The bowl is small in the frame and the lower half is empty, which leaves the balance somewhat weak." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, leaves and noodles are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands have wavy bends and overlap, but they form a narrow flat band across the bowl. The strands look like scribbled lines with little volume and don't sit in the broth convincingly." }, "toppings": { "score": 0.45, "reason": "Green leaves, tofu cubes and mushroom caps are recognizable and vegan-appropriate, but they are few, small, flat and simplistic. The mushrooms are brown discs and the tofu reads as plain blocks. They look pasted on top, with no nori or other variety." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, dark inner wall, red-gradient body and a ground shadow, with consistent perspective. It is fairly flat and simple, with little ceramic highlight. The base is barely defined and the shadow is wide." }, "broth": { "score": 0.55, "reason": "The warm orange radial gradient gives a glowing, appetizing surface with some depth. It is smooth and flat, with little texture, sheen or ripples. Toppings and noodles show no contact or immersion." }, "steam": { "score": 0.6, "reason": "Translucent wispy strands rise and change shape between frames, and they are not opaque. The wisps look like similar vertical squiggles, and they start faintly above the bowl rather than clearly emerging from the soup." }, "composition": { "score": 0.65, "reason": "The bowl is centered as a clear focal point against a dark, moody background with a warm glow and no distracting text. The bowl is small in the frame and the lower half is empty, but the colors are cohesive." }, "vegan": { "score": 0.92, "reason": "Only noodles, tofu, mushrooms and leaves are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands have wavy bends and overlap, but they form a flat, narrow band across the surface. The strands are scribble-like and have little volume, and they don't read as noodles sitting in broth." }, "toppings": { "score": 0.45, "reason": "Green leaves, tofu cubes and mushroom caps are recognizable, but they are simple and flat. The mushrooms are small brown discs and the tofu is plain blocks. Variety is limited, the pieces sit sparsely around the rim with no nori, and texture is minimal." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a dark inner wall, a red gradient body and a ground shadow. Perspective is consistent. It is fairly plain, with a weak base and little ceramic highlight, and the shadow looks slightly detached." }, "broth": { "score": 0.55, "reason": "The orange radial gradient gives warmth and depth and reads as broth. It is smooth and flat with no surface texture or sheen. Ingredients sit on it with little immersion or ripple." }, "steam": { "score": 0.72, "reason": "Translucent wispy strands rise and change shape from frame to frame, and they fade toward the top without hiding the food. The strands look somewhat uniform and stringy, and they are fairly faint." }, "composition": { "score": 0.68, "reason": "The bowl is centered as the focal point on a dark, warm vignette, with restrained colors and no clutter. The bowl is small relative to the frame, and there is a large empty area below it." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens and noodles are visible. There are no animal ingredients." } } ], "diagnostics": { "vegan": 0.9266666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.48333333333333334, "toppings": 0.45, "bowl": 0.7333333333333333, "broth": 0.55, "steam": 0.75, "composition": 0.8166666666666667 }, "reward": 0.605, "sampleStdDev": 0.010897247358851706, "judgments": [ { "noodles": { "score": 0.45, "reason": "Pale squiggly strands span the bowl with some wave rhythm, but they read as flat wire-like scribbles pasted on the surface \u2014 no volume, shadows, immersion, or coherent strand overlap." }, "toppings": { "score": 0.45, "reason": "Some variety (green scallion-like leaves with vein lines, brown discs, beige tofu squares), but the forms are generic flat shapes: featureless squares and plain disks dominate, reading as simple vector tokens rather than food." }, "bowl": { "score": 0.75, "reason": "Coherent stylized construction: elliptical dark rim with highlight arc, a lighter inner ring, tapering red-orange body, and a soft contact shadow. Perspective is consistent; only mild flatness and a slightly thick uniform rim keep it from higher." }, "broth": { "score": 0.6, "reason": "Warm orange gradient surface with a central glow reads as liquid, and discs/tofu roughly sit on it, but there is no meniscus, texture, or depth, and the noodle squiggles look pasted atop rather than immersed." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise above the bowl and clearly change shape between the time-ordered normal frames (different plumes at 1.6s, 9.2s, 15.9s), while reduced-motion frames are static \u2014 supported ambient motion. Wisps are irregular and never hide the food, though they look slightly scribble-like." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric backdrop with restrained warm glow and subtle ambient particles; no text, UI, or clutter, and the bowl stays the clear focal point. The dish is small in frame but well presented." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are scallion-like greens, brown mushroom-like slices, tofu squares, noodles, and plant broth \u2014 no animal-derived toppings evident." } }, { "noodles": { "score": 0.5, "reason": "Noodles read as thin continuous yellow strands, but they form a single flat horizontal band of uniform repetitive loops with no volume, pile, or nesting; they lie as a ribbon across the surface rather than sitting in broth." }, "toppings": { "score": 0.45, "reason": "Green leaf shapes with midribs are the only convincingly organic pieces; the beige squares are featureless blocks and the brown items are flat plain ellipses, dominating the topping set. Several leaves and squares overlap the back rim looking pasted rather than seated, and variety/detail is minimal." }, "bowl": { "score": 0.75, "reason": "Coherent stylized vector bowl: clean elliptical rim with a subtle left highlight, smooth tapered body, plausible rim-to-body join, and grounded contact shadow. Defects are only simplicity \u2014 a uniformly thick rim and no visible foot or inner-wall depth." }, "broth": { "score": 0.5, "reason": "Broth is a flat orange radial gradient that reads more like a glowing light pool than liquid: no sheen, texture, or edge occlusion, and toppings/noodles appear laid on top with no immersion or contact darkening." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps rise from the bowl, fade with height, and clearly change shape and position across desktop-1/2/3, supporting continuous rising motion (static only in reduced-motion frames, as expected). Wisps are irregular and do not hide the food, though line-squiggle quality is a bit scribbly rather than soft vapor." }, "composition": { "score": 0.8, "reason": "Centered, readable focal dish in an atmospheric dark scene with restrained ember particles and a warm glow behind the bowl; no clutter, text, or competing elements. The dish stays the focus; only minor emptiness in the mobile crop." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are noodles, leafy greens, and pale squares plausibly tofu; no clearly animal-derived items. The brown ellipses are ambiguous (could read as mushroom slices or chashu), which prevents full credit." } }, { "noodles": { "score": 0.5, "reason": "Noodles are a flat horizontal band of thin curly squiggles lying on the broth surface; strand bends are readable but there is no volume, coherent overlap, or immersion, and the dense zigzags at the right edge read scribble/wire-like." }, "toppings": { "score": 0.45, "reason": "Three ingredient types (green leaf shapes with a vein line, brown oval mushroom-like caps, beige square blocks), but the squares are featureless flat cards, the ovals are plain glossy ellipses, and everything lacks texture, scale variation, and contact detail \u2014 recognizable yet artificial." }, "bowl": { "score": 0.7, "reason": "Coherent stylized bowl: dark elliptical rim with a highlight arc, consistent opening/body perspective, smooth red-brown gradient body, and a grounding contact shadow; no broken joins, though it reads shallow and simply modeled." }, "broth": { "score": 0.55, "reason": "The orange surface with warm central glow reads as liquid with sheen, but it looks like a flat glowing disc; toppings and the noodle band sit pasted on top with no occlusion, immersion shadows, or believable contact." }, "steam": { "score": 0.75, "reason": "Normal frames at 1.6s, 9.1s, 15.9s, 38s and 60s show clearly different wisp shapes each time, so the steam animates; it is translucent gray, rises from the soup, spreads and fades without hiding the food, though the wisps are somewhat thin squiggle-lines. Reduced-motion frames are static, consistent with the preference." }, "composition": { "score": 0.8, "reason": "Centered dish in a clean dark atmospheric scene with warm glow, subtle ember particles, no text or clutter; the bowl is a bit small with generous empty space, but the presentation is coherent and focused." }, "vegan": { "score": 0.9, "reason": "Visible ingredients \u2014 noodles, green leaves, brown mushroom-like ovals, and tofu-like beige blocks in plant broth \u2014 plausibly form vegan ramen; no clearly animal-derived toppings appear, though the brown ovals are slightly ambiguous in shape." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9277777777777778, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 808.3499999999985, "frameIntervalP95Ms": 800, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "ec69a8b807df7fd3e34c09cabe42c52d56aba9e4c381541c7f6b36a7bbff4413", "gradedAt": "2026-09-30T23:10:39.716252+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5555555555555556, "toppings": 0.5888888888888889, "bowl": 0.7933333333333333, "broth": 0.5077777777777778, "steam": 0.7022222222222222, "composition": 0.7133333333333334 }, "reward": 0.6444444444444445, "runId": "openai/gpt-5.6-terra/ultra", "rewardStdDev": 0.10290898531838344, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.6166666666666667, "bowl": 0.86, "broth": 0.5233333333333333, "steam": 0.6966666666666667, "composition": 0.66 }, "reward": 0.649, "sampleStdDev": 0.0022912878474778855, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous yellow strands have smooth bends and readable overlaps, but mostly form parallel flat waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.62, "reason": "Veined greens, brown mushroom caps with pale stems, and shaded tofu are recognizable and varied. Leaves look angular and card-like, while tofu and mushrooms have limited organic texture." }, "bowl": { "score": 0.86, "reason": "The elliptical rim, rounded walls, dark base, and contact shadow form a coherent supported bowl. Warm rim highlights and curved lacquer-like reflections convey depth; no material construction defect is observed." }, "broth": { "score": 0.54, "reason": "A visible glossy surface fills the cavity, but multicolored pink and green reflections look unusually iridescent, and ingredients mostly appear laid above the liquid rather than immersed." }, "steam": { "score": 0.7, "reason": "Normal frames show changes in the translucent wisps' bends and brightness, with fading upper ends and unobscured food. The evenly spaced, similarly shaped columns remain somewhat mechanical and show limited spreading." }, "composition": { "score": 0.65, "reason": "The desktop dish is clearly focused against a restrained dark background with warm illumination and no distracting UI. Excess upper empty space and substantial right-side clipping of the bowl on mobile weaken framing." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy vegetables, mushrooms, and tofu plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.5, "reason": "Continuous yellow strands have smooth bends and readable overlaps, but predominantly form parallel flat waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.62, "reason": "Veined green leaves, brown mushroom caps and stems, and shaded tofu pieces are recognizable and varied. Shapes remain geometric and smooth, with sparse organic texture and rather blocky leaves." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim joins rounded burgundy walls and a grounded base; layered highlights and the contact shadow convey lacquer and depth. No material construction defect is observed, though highlights are somewhat broad and graphic." }, "broth": { "score": 0.51, "reason": "The bounded surface has sheen and colorful reflections, but its pink-purple gradients look more decorative than appetizing. Ingredients largely sit above it with limited liquid contact or submerged detail." }, "steam": { "score": 0.69, "reason": "Normal frames show changes in the translucent wisps' bends and brightness, with fading upper ends and unobscured food. The four similarly spaced vertical plumes remain repetitive and show limited spreading." }, "composition": { "score": 0.68, "reason": "Warm food colors stand out against a restrained dark background without text or clutter. Desktop presentation is clear but low-centered; the mobile capture cuts off the bowl's right side and leaves excessive empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, small vegetable garnishes, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous yellow strands have smooth bends and readable overlap, but mostly form parallel, flat waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.61, "reason": "Veined green leaves, brown mushroom caps with stems, and scored tofu are recognizable and varied. Shapes remain geometric and smooth, with stiff leaves and little organic surface texture." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim joins rounded, shaded walls and a grounded base; warm rim highlights and dark lacquer-like bands convey depth. No material construction defect is observed, though the broad highlights are somewhat graphic." }, "broth": { "score": 0.52, "reason": "The surface fills the cavity coherently and has luminous reflections, but its rainbow-like sheen looks artificial, with minimal liquid texture and weak immersion around toppings." }, "steam": { "score": 0.7, "reason": "Normal frames show changes in wisp curvature and opacity; translucent trails emerge above the soup and fade without obscuring food. The evenly spaced, similarly shaped columns remain mechanical with little branching or spreading." }, "composition": { "score": 0.65, "reason": "The desktop dish is clearly focused against a restrained dark background with warm lighting and no clutter. Excess upper empty space and substantial right-side clipping of the bowl on mobile weaken framing." }, "vegan": { "score": 1, "reason": "Visible noodles, tofu-like pieces, mushrooms, leafy greens, and small vegetable garnishes plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.48333333333333334, "toppings": 0.43333333333333335, "bowl": 0.67, "broth": 0.4, "steam": 0.6266666666666667, "composition": 0.6633333333333333 }, "reward": 0.5393333333333333, "sampleStdDev": 0.016327379867367972, "judgments": [ { "noodles": { "score": 0.45, "reason": "The noodles are thin, wavy parallel ribbons that look like repeated flat loops, with a rigid, uniform feel. They are readable as noodles but sit on top of the surface with little volume or overlap." }, "toppings": { "score": 0.4, "reason": "Green leaves with veins and mushroom caps are recognizable. The tofu cubes are rounded beige blocks that look like plastic cards, and the small dots and scallion bits are generic. Variety is limited and there is no nori." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a terracotta body with highlights, a base and a contact shadow. The glassy rim overlay is slightly odd and the wall is fairly flat, but no major distortion is visible." }, "broth": { "score": 0.4, "reason": "The broth is a pink and purple iridescent surface with colored blotches. It reads as a glassy lid or disc rather than appetizing soup, and the toppings seem to sit under or on the film with weak immersion." }, "steam": { "score": 0.6, "reason": "Four thin, translucent wavy wisps rise and change between frames without hiding the food. They are fairly similar to each other and read as mechanical, but they fade softly." }, "composition": { "score": 0.68, "reason": "The bowl is centered and is the clear focus against a dark warm vignette, with no clutter or text. Color harmony is decent, but the broth colors are odd and the bowl sits low on mobile." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, greens and noodles, with no visible meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Wavy cream strands are readable and overlap, but they look like thin flat parallel ribbons lying on the surface, with little volume and few bends, so they read as repeated loops rather than a supple mass." }, "toppings": { "score": 0.45, "reason": "Green bok choy leaves with veins and brown mushroom caps are recognizable, but the tofu cubes are rounded glossy blocks and the mushrooms are partly hidden. There is no nori, and the small dots read as generic confetti." }, "bowl": { "score": 0.68, "reason": "The rim, cavity, tapered body, highlight streaks and ground shadow are coherent, and the terracotta lacquer reads well. The surface is somewhat flat and the glassy lid-like sheen over the soup is slightly odd." }, "broth": { "score": 0.4, "reason": "The broth surface has a pink and purple iridescent glassy look that resembles a lens or disc more than an appetizing broth. Ingredients sit under or on it with little immersion or contact detail." }, "steam": { "score": 0.68, "reason": "Four translucent wavy wisps rise and change shape between frames, fading at the top without hiding the food. They are fairly uniform and thin, and they look a bit mechanical." }, "composition": { "score": 0.66, "reason": "The bowl is centered in a dark warm vignette with good contrast and no clutter. The bowl is small and low in the frame, and the garish broth colors weaken the appeal. On mobile the bowl is cropped." }, "vegan": { "score": 0.92, "reason": "The visible ingredients are noodles, bok choy, mushrooms and tofu-like cubes, with no meat or egg visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are wavy, parallel, thin strands that read as noodles but look repetitive and flat, with a regular sine pattern. They sit on top of the liquid with little volume or overlap." }, "toppings": { "score": 0.45, "reason": "The bok choy leaves, mushroom caps and tofu cubes are recognizable but simplistic. The tofu cubes are rounded, card-like blocks, and the mushrooms are mostly hidden. There is no nori, and little variety or texture." }, "bowl": { "score": 0.65, "reason": "The bowl has a coherent elliptical rim, a dark inner lip, a tapered body with highlights, and a contact shadow. It looks like a competent stylized illustration. The body is somewhat flat and the glassy rim overlay is odd." }, "broth": { "score": 0.4, "reason": "The broth surface shows a pink and purple blotchy iridescent tint that looks murky and unappetizing, more like a glass lid than soup. Ingredients look pasted on top rather than immersed." }, "steam": { "score": 0.6, "reason": "Four translucent wavy wisps rise and change between frames, and they do not hide the food. They are thin and similar to each other, and the plumes look fairly mechanical." }, "composition": { "score": 0.65, "reason": "The single bowl is centered in a dark, warm-lit scene and is clearly the focus. On mobile the bowl is cropped. The colors in the broth are off-putting, but the overall presentation is clean." }, "vegan": { "score": 0.9, "reason": "The toppings are tofu, mushrooms, greens and noodles, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9166666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6833333333333333, "toppings": 0.7166666666666667, "bowl": 0.85, "broth": 0.6, "steam": 0.7833333333333333, "composition": 0.8166666666666667 }, "reward": 0.745, "sampleStdDev": 0.0050000000000000044, "judgments": [ { "noodles": { "score": 0.65, "reason": "Pale cream strands are continuous with rounded bends and coherent layering, and they read as noodles tucked under the tofu cubes; however the wave pattern is noticeably repetitive and flat, more like uniform decorative squiggles than supple varied strands." }, "toppings": { "score": 0.75, "reason": "Good variety: brown shiitake-style caps with stems, veined green leaves, cream tofu cubes with score lines, plus scallion/edamame dots. Shapes are mostly organic and readable, though the leaves are somewhat flat cutouts and the tofu blocks are slightly uniform." }, "bowl": { "score": 0.85, "reason": "Coherent thick orange rim with visible wall thickness, consistent high-angle ellipse, smooth join to the red-brown body with soft highlights and shading, and a grounded contact shadow. Only minor issues: slightly exaggerated rim thickness and a large diffuse base shadow." }, "broth": { "score": 0.6, "reason": "Liquid surface shows sheen and ingredients mostly emerge with some occlusion, but the strong pink/purple iridescent swirl reads more like glass or soap film than appetizing broth, and immersion/contact of the noodle layer is shallow." }, "steam": { "score": 0.75, "reason": "Translucent wisps clearly change shape and position across desktop-1/2/3 and pointer frames, rising from the soup and fading without hiding food. Slight deduction because the plumes are several similar repeated S-curves with a somewhat mechanical look." }, "composition": { "score": 0.8, "reason": "Desktop presentation is polished: centered focal bowl, warm rim glow against a dark vignette, restrained particles, no clutter. Defect: on the mobile frame the bowl is shifted right and clipped by the viewport edge, breaking balance." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, scallion-like dots and noodle broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Cream-colored strands are continuous with gentle bends and layered overlaps across the broth, with plausible thickness and ends tucked behind greens. They read slightly flat and uniform, more a painted squiggle layer than a heaped noodle mass with volume." }, "toppings": { "score": 0.7, "reason": "Good variety: veined dark-green leaves, brown shiitake-like caps, green scallion ovals, and red/yellow specks all read as food. The three beige tofu cubes are readable but blocky, resembling marshmallow-like slabs with score lines; depictions are simple and somewhat flat." }, "bowl": { "score": 0.85, "reason": "Coherent ellipse rim with an orange lip, visible inner cavity shadow, deep maroon body with a curved ceramic highlight, small foot and grounded contact shadow; perspective between rim, body and base is consistent. Only minor stylization-level imperfections." }, "broth": { "score": 0.6, "reason": "Toppings sit in the surface with soft occlusion shadows and a rim highlight gives some depth, but the pink/magenta iridescent sheen reads more like an oily soap-film than appetizing soup, limiting believability of the liquid." }, "steam": { "score": 0.8, "reason": "Translucent pale wisps rise from the bowl, curve irregularly and fade without obscuring food; their shapes and positions differ across desktop-1/2/3 and pointer frames, showing animated rise (static in reduced frames as expected under reduced motion). Wisps are a bit similar/repeated in form." }, "composition": { "score": 0.85, "reason": "Centered dish with warm rim light against a dark atmospheric ground, subtle glow and sparkles, no clutter or UI; food remains the focus. Minor deduction for the mobile frame where the bowl sits low and is slightly cropped at the right edge." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu cubes, leafy greens, mushrooms, scallions and specks in broth; no meat, fish, egg or dairy forms are present." } }, { "noodles": { "score": 0.7, "reason": "Continuous pale-yellow wavy strands spread across the surface with varied bends and crossings; readable as stylized noodles but thin, flat and uniform, lacking volume, thickness variation and immersion depth." }, "toppings": { "score": 0.7, "reason": "Good variety: scored tofu blocks, large veined greens, brown mushroom caps at the back rim, scallion slices and chili/sesame specks; shapes are simplified and the leaves read slightly flat, but all are recognizable as plant ingredients rather than generic plastic forms." }, "bowl": { "score": 0.85, "reason": "Coherent lacquer-style bowl: consistent elliptical rim with warm orange lip, inner edge, tapered body with curved highlight stripes and a grounded cast shadow; no visible distortion or broken joins, though the base is lost in shadow." }, "broth": { "score": 0.6, "reason": "Ingredients show contact and partial submersion with subtle reflections, but the surface is an iridescent purple/magenta/green sheen that reads more like an oil-slick or holographic pool than an appetizing savory broth." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise above the bowl without obscuring food; across desktop-1/2/3 and pointer frames the plumes change height, bend and fade, supporting continuous rising motion (static only in reduced-motion frames, as expected). Plumes are somewhat similar parallel S-curves." }, "composition": { "score": 0.8, "reason": "Desktop presentation is balanced and atmospheric: centered focal bowl, warm under-glow, subtle particles, no text or clutter. Defect: the mobile frame crops the bowl at the bottom-right corner, showing a responsive framing issue." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, leafy greens, mushroom caps, scallions and chili/sesame in a plant broth; no meat, egg, fish or dairy is depicted." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9722222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100, "frameIntervalP95Ms": 183.29999999999927, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a3e8e19f89ac70d6ddce4bccd95e8ea9c0c0e25188e1fd340083f25ff1f7785f", "gradedAt": "2026-09-30T23:11:15.651690+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5444444444444445, "toppings": 0.49444444444444446, "bowl": 0.7077777777777777, "broth": 0.5544444444444444, "steam": 0.5733333333333334, "composition": 0.7133333333333334 }, "reward": 0.5868333333333333, "runId": "openai/gpt-5.6-terra/xhigh", "rewardStdDev": 0.05964641183209973, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5166666666666667, "bowl": 0.76, "broth": 0.5133333333333333, "steam": 0.62, "composition": 0.7266666666666667 }, "reward": 0.5981666666666667, "sampleStdDev": 0.005484827557301481, "judgments": [ { "noodles": { "score": 0.5, "reason": "Golden strands have smooth bends and some overlapping shading, but broad, repeated upright loops read as flat ribbons rather than a supple mass resting in soup." }, "toppings": { "score": 0.5, "reason": "Veined green leaves, brown mushroom-like caps and a marked tofu square are readable, but their flat, repeated shapes and sparse texture make them ingredient symbols rather than convincing food." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, curved walls and shaded underside form a coherent supported vessel. Strong layered bands and a uniformly bright rim make the material somewhat graphic rather than convincingly ceramic." }, "broth": { "score": 0.52, "reason": "Warm amber shading and small surface flecks suggest soup, but concentric linework and limited immersion cues leave the ingredients looking arranged above a flat surface." }, "steam": { "score": 0.62, "reason": "Faint translucent wisps change position and visibility across normal frames without obscuring the food. They remain sparse, narrow strokes, with weak attachment to the soup and little visible spreading." }, "composition": { "score": 0.73, "reason": "A single centered bowl remains readable against a dark background, with harmonious orange and green accents and no distracting UI. Heavy grain, a visible rectangular glow boundary and excessive mobile negative space reduce polish." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushroom-like caps and tofu plausibly form vegan ramen; no visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have smooth bends and some overlapping shading, but broad, isolated upright loops dominate rather than a supple, immersed noodle pile." }, "toppings": { "score": 0.5, "reason": "Veined greens, brown mushroom-like caps and marked tofu are recognizable, but repeated disk-like mushrooms and flat, minimally textured ingredient symbols limit plausibility." }, "bowl": { "score": 0.76, "reason": "The continuous elliptical rim, curved walls and aligned base create coherent depth. Layered highlights suggest lacquer, though the heavy bands and weak ground contact make it somewhat artificial." }, "broth": { "score": 0.52, "reason": "Warm amber shading and small oil-like dots suggest soup, but concentric surface lines and limited immersion make the ingredients appear arranged above a relatively flat liquid surface." }, "steam": { "score": 0.62, "reason": "Faint translucent wisps change position and visibility across normal frames without obscuring food. Their sparse, short, similarly curved forms often appear disconnected from the soup; continuous spreading is not established." }, "composition": { "score": 0.73, "reason": "A centered single bowl and warm orange-green palette remain readable against the dark background with no distracting UI. Heavy grain and a visible rectangular glow boundary reduce polish; mobile retains the dish but has reported horizontal overflow." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, mushroom-like caps and tofu plausibly form vegan ramen, with no discernible animal-derived topping. No material compliance defect is observed." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have smooth bends and some overlap, but broad, repeated upright loops read as flat ribbons rather than a supple noodle pile immersed in soup." }, "toppings": { "score": 0.55, "reason": "Veined green leaves, brown mushroom-like caps and marked tofu are recognizable, but their sparse, flat shapes are highly schematic; mushrooms resemble decorated disks and tofu lacks volume." }, "bowl": { "score": 0.76, "reason": "The continuous elliptical rim, curved walls and aligned base form a coherent vessel. Warm highlights suggest lacquer, though layered front bands and a weak contact shadow flatten its depth." }, "broth": { "score": 0.5, "reason": "Warm amber shading and small surface flecks suggest soup, but concentric outlines and limited liquid sheen make the surface diagrammatic. Most ingredients appear perched above it rather than immersed." }, "steam": { "score": 0.62, "reason": "Faint translucent wisps change position and visibility across normal frames without obscuring food. They remain sparse, similarly narrow curves, often disconnected from the soup, with little visible spreading." }, "composition": { "score": 0.72, "reason": "A centered single dish and warm orange-green palette create a readable focus against the dark background with no text or UI. Heavy grain and a visible rectangular lighting boundary weaken the finish; mobile has substantial empty space and reported overflow." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushroom-like caps and tofu plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.55, "toppings": 0.45, "bowl": 0.6133333333333333, "broth": 0.5166666666666667, "steam": 0.4166666666666667, "composition": 0.63 }, "reward": 0.5223333333333333, "sampleStdDev": 0.003752776749732598, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are tangled, curvy, pale-gold strands with some overlap and tonal variation, which reads as noodles. They are uniform in thickness and look like flat stroked loops, without much volume or immersion." }, "toppings": { "score": 0.45, "reason": "The toppings are two green leaf shapes, two mushroom slices, a tofu square and scallion sticks, all recognizable but simple. The tofu is a rounded card and the scallions are rigid neon sticks. There is little texture and no nori." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, curved walls and a glowing base with a contact shadow. The orange lacquer highlights are decent. The wall reads slightly flat with odd double rim lines, and the base glow is detached from the bowl." }, "broth": { "score": 0.5, "reason": "The broth is a warm orange with a central glow and faint concentric rings that look a bit like a target. Ingredients sit on top rather than being immersed, with little depth or ripple." }, "steam": { "score": 0.45, "reason": "Steam is a few thin, faint wisps that change position between frames, so there is some motion. They are sparse, thin and stringlike, more like lines than translucent plumes." }, "composition": { "score": 0.62, "reason": "The bowl is centered on a dark, grainy background with a warm palette, so the food is the clear focus. The bowl is somewhat small and low in the frame with a lot of empty space. There is no clutter or text." }, "vegan": { "score": 0.85, "reason": "Mushrooms, tofu, greens and noodles all look plant-based. There are no visible animal ingredients." } }, { "noodles": { "score": 0.55, "reason": "Noodles are curly, continuous golden strands with some overlap and volume in the center. They look like uniform-width tubes with repeated loops, somewhat flat and artificial." }, "toppings": { "score": 0.45, "reason": "Toppings include green leaves, mushroom caps, a tofu square and scallion sticks. They are recognizable but simple and flat. The leaves are sail-like triangles, the scallions are rigid sticks, and the tofu is a rounded card. Nori is absent." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a curved body, highlights and a glow underneath. The interior perspective is plausible. The wall layers are a bit muddy, and the base sits on a glow rather than a solid contact." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a central glow and thin ripple lines, so it reads as liquid. It looks flat and somewhat opaque, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.4, "reason": "A few thin, faint wisps rise and change position across frames, so there is some motion. They are sparse, thin lines that look like strokes rather than soft translucent steam." }, "composition": { "score": 0.65, "reason": "The bowl is centered and is the clear focal point against a dark, grainy background. The warm color palette is cohesive and there is no clutter. The bowl is a bit small and the steam is weak." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are noodles, mushrooms, tofu, leaves and scallions, with no animal products visible." } }, { "noodles": { "score": 0.55, "reason": "Noodles are continuous, curvy golden strands with some overlap and a warm gradient, but they are uniform in thickness and look like flat loops. They sit on the surface with little sense of volume or immersion." }, "toppings": { "score": 0.45, "reason": "Green leaves, brown mushroom caps, scallion sticks and a tofu square are recognizable but simplistic. The tofu is a rounded card and the mushrooms are plain ellipses, so the toppings read as flat symbols with little texture. There is no nori." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, curved walls and a glowing base, with warm highlights and a contact shadow. It is smooth and somewhat plain, the wall bands are a little confusing, and the base glow looks detached from the bowl." }, "broth": { "score": 0.5, "reason": "The broth is an orange-red surface with a radial glow and faint ripple lines. It is readable but flat, and the toppings look pasted on top with little occlusion or immersion." }, "steam": { "score": 0.4, "reason": "A few thin wisps rise and change between frames, so there is some motion. They are sparse, thin and stringlike, and the plumes look similar to one another, which is less convincing than soft translucent steam." }, "composition": { "score": 0.62, "reason": "A centered bowl on a dark grainy background has a warm palette and a clear focal point. The bowl is small in the frame with a lot of empty space above and below, and the food detail is modest." }, "vegan": { "score": 0.9, "reason": "The toppings are leaves, mushrooms, tofu and scallion, with noodles in broth. No animal products are visible." } } ], "diagnostics": { "vegan": 0.8833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5833333333333334, "toppings": 0.5166666666666667, "bowl": 0.75, "broth": 0.6333333333333333, "steam": 0.6833333333333333, "composition": 0.7833333333333333 }, "reward": 0.64, "sampleStdDev": 0.017500000000000047, "judgments": [ { "bowl": { "reason": "Coherent shallow bowl: bright orange rim ring, smooth curved body with ceramic highlight streaks, a plausible foot shadow and warm glow seat beneath; rim/body/base perspectives agree with no broken joins or detached parts.", "score": 0.8 }, "broth": { "reason": "Warm glowing broth surface with scattered red specks and soft contact crescents under toppings, but faint concentric ellipse ripple lines cross the surface and noodles/toppings mostly read as lying on a flat plane rather than immersed.", "score": 0.65 }, "composition": { "reason": "Centered, readable focal dish in a cohesive dark noisy scene with a warm under-glow and green accents; no text, clutter or competing elements, though the upper frame is fairly empty except thin wisps.", "score": 0.8 }, "noodles": { "reason": "Cream strands coil and overlap with rounded, fairly consistent thickness and some fill/outline volume, but repeated identical spiral loops and closed donut-like circles look decorative, and capped strand ends float flat on the surface.", "score": 0.6 }, "steam": { "reason": "Thin translucent wavy wisps change position and shape across the normal temporal frames (desktop-1/2/3, pointer frames), showing continuous animation without hiding food; however they are sparse single-stroke curves that begin slightly detached above the rim.", "score": 0.6 }, "toppings": { "reason": "Leaf shapes with vein strokes and scallion rods read acceptably, but the striped rounded-square is a rigid card-like block and the brown ovals are glossy generic disks; variety exists yet generic geometric forms are prominent.", "score": 0.55 }, "vegan": { "reason": "Visible items are noodles, greens, scallion pieces, leafy toppings, a tofu/aburaage-like striped square and brown mushroom-cap-like ovals; nothing is unmistakably animal-derived, though the oval slices and striped square are ambiguous.", "score": 0.75 } }, { "noodles": { "score": 0.55, "reason": "Tan strands form a coherent coiled mass with overlapping arcs, but they are flat uniform-width squiggles with rounded caps that read as repeated flat loops lying on the surface rather than supple volumetric noodles." }, "toppings": { "score": 0.55, "reason": "Readable variety: green rod scallions, striped pointed leaves, a hatched tofu-style block, and two brown striped ovals. Shapes are simple and stylized; the brown ovals read as generic flat glossy disks rather than clearly depicted food." }, "bowl": { "score": 0.75, "reason": "Coherent flat-design bowl: elliptical rim with orange lip, red interior, layered body bands, base shadow and grounded contact with the glow beneath. Perspective is consistent; the body is somewhat simplistic with a decorative stripe but nothing broken or detached." }, "broth": { "score": 0.65, "reason": "Warm liquid surface with oil swirls, chili flecks and a central hotspot gives depth, and toppings overlap into it. The radial glow reads slightly artificial and ingredients look somewhat pasted onto the surface plane rather than immersed." }, "steam": { "score": 0.7, "reason": "Across desktop frames the translucent wisps change position, shape and count (left/center/right drift, fade in and out), showing continuous rising motion. Wisps are thin, irregular, and never obscure the food, though sparse and slightly disconnected." }, "composition": { "score": 0.8, "reason": "Centered focal dish on a dark grainy backdrop with tasteful warm glow and shadow; no text, UI or clutter. Balanced on desktop and mobile, though mobile shows horizontal overflow per diagnostics and the backdrop glow does some of the visual work." }, "vegan": { "score": 0.75, "reason": "Visible items are tofu-like block, scallion, leafy greens, noodle strands and red broth; no eggs, fish or meat clearly shown. The two brown striped ovals are ambiguous (mushroom vs. sliced meat) but plausibly plant-based, so not full credit." } }, { "noodles": { "score": 0.6, "reason": "Yellow-tan strands are continuous, uniform in thickness and readable as noodles, but they render as repeated flat telephone-cord loops lying on the surface with little immersion or volume; ends terminate abruptly at the broth." }, "toppings": { "score": 0.45, "reason": "Sparse set: two flat green triangle leaves, green scallion sticks, and brown oval disks with a stripe plus a tan striped square tile that reads as a generic icon rather than food. Flat, featureless shapes dominate, and variety/texture is minimal." }, "bowl": { "score": 0.7, "reason": "Coherent terracotta bowl: clean elliptical rim with orange highlight, curved wall with highlight band, dark foot and under-glow support. Slightly dish-like wide/low proportions and a simple flat vector finish, but no broken joins or impossible geometry." }, "broth": { "score": 0.6, "reason": "Warm glowing red-orange surface with restrained sheen, oil speckles and subtle ripple ellipses. However the surface is fairly flat and ingredients sit pasted on top of it with little believable submersion or occlusion." }, "steam": { "score": 0.75, "reason": "Thin translucent S-shaped wisps change position, count and shape across desktop frames (fade in/out and drift), showing genuine ambient motion; they are irregular and never hide the food, though they read more as thin curls than volumetric wisps. Reduced frames are correctly static." }, "composition": { "score": 0.75, "reason": "Centered bowl as clear focal point against a dark atmospheric background with warm under-glow; balanced and free of clutter or text. Minor issues: heavy grain texture, slightly sparse topping spread, and mobile horizontal overflow reported." }, "vegan": { "score": 0.55, "reason": "No unmistakable animal ingredients, but appearance is ambiguous: the brown striped ovals could read as chashu or mushrooms and the striped square could read as naruto or fried tofu, so only intermediate compliance credit is warranted." } } ], "diagnostics": { "vegan": 0.6833333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8555555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": true, "reducedHorizontalOverflow": false } }, { "artifactSha256": "62bf9c88b03664b2831e1f2d12e59ae8f0c5f11ea8ed7e7e8a987264f24bcfc1", "gradedAt": "2026-09-30T22:57:33.661179+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5477777777777778, "toppings": 0.6788888888888889, "bowl": 0.8177777777777778, "broth": 0.6477777777777778, "steam": 0.6733333333333333, "composition": 0.8233333333333334 }, "reward": 0.6909444444444445, "runId": "openai/gpt-6-astra/low", "rewardStdDev": 0.07028519943244997, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.46, "toppings": 0.6533333333333333, "bowl": 0.8766666666666667, "broth": 0.61, "steam": 0.6533333333333333, "composition": 0.8633333333333333 }, "reward": 0.676, "sampleStdDev": 0.0036055512754639926, "judgments": [ { "noodles": { "score": 0.46, "reason": "Thin golden strands are readable, but tightly repeated concentric loops and parallel arcs form a flat, maze-like stack rather than supple tangled noodles." }, "toppings": { "score": 0.66, "reason": "Veined greens, scored mushroom caps with visible gills, and scattered scallions are recognizable. Tofu is sharply geometric with sparse surface detail, and leaf silhouettes are overly smooth." }, "bowl": { "score": 0.88, "reason": "A coherent elliptical rim, visible inner wall, rounded body, subtle ceramic speckles, and restrained highlights establish convincing thickness and depth. The grounding shadow supports the bowl; no material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm amber shading and small surface flecks suggest broth, but the surface has little liquid sheen and ingredients appear laid atop it with weak immersion cues." }, "steam": { "score": 0.66, "reason": "Normal frames show translucent plumes changing height, position, and visibility without hiding the food. Their repeated sinusoidal shapes and conspicuous straight-sided fade boundaries look mechanical rather than naturally dispersing." }, "composition": { "score": 0.86, "reason": "The single dish remains clearly readable against a restrained dark background, with harmonious amber, cream, and green colors and ample space for steam. Desktop and mobile framing are clean and uncluttered; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, scallions, tofu-like cubes, and noodles plausibly form vegan ramen, with no visibly animal-derived topping." } }, { "noodles": { "score": 0.46, "reason": "Thin continuous golden strands are readable, but densely repeated nested loops form a flat, mechanical coil rather than supple overlapping noodles." }, "toppings": { "score": 0.66, "reason": "Veined greens, scored mushroom caps with pale gills, and scallion rings provide recognizable variety. Tofu is overly geometric with stripe-like marks, and leaf silhouettes are rather stiff." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, visible inner wall, smoothly shaded rounded body, ceramic speckling, and restrained highlights establish convincing thickness and depth. No material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm amber shading and scattered surface flecks suggest broth, but the surface has little liquid sheen or meniscus detail, and ingredients appear laid on top rather than convincingly immersed." }, "steam": { "score": 0.64, "reason": "Normal frames show changing positions and fading visibility of translucent wisps without obscuring the dish. The plumes remain similarly sinusoidal, with conspicuous straight vertical boundaries and limited spreading." }, "composition": { "score": 0.87, "reason": "A single clearly readable bowl anchors the restrained dark scene, with harmonious amber, cream, and green colors and a soft grounding shadow. Desktop and mobile remain uncluttered; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, leafy greens, scallions, and tofu alongside noodles and plant-compatible broth; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.46, "reason": "Thin golden strands are readable, but tightly repeated nested loops dominate, resembling a flat coil rather than supple, irregular noodles with convincing volume." }, "toppings": { "score": 0.64, "reason": "Veined greens, scored mushroom caps with visible gills, and scattered green garnish provide readable variety. Tofu is overly geometric and smooth, and the leaves have simplified, stiff silhouettes." }, "bowl": { "score": 0.87, "reason": "The continuous elliptical rim, shaded inner wall, rounded body, fine ceramic speckles, and subtle curved highlights create coherent depth and material. No material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm amber shading and scattered surface flecks suggest soup, but the surface lacks liquid reflections and convincing immersion; ingredients appear largely arranged on top." }, "steam": { "score": 0.66, "reason": "Translucent wisps change position and visibility across normal frames without obscuring the food. Their repeated sinusoidal shapes and conspicuously straight vertical boundaries make the effect mechanical rather than naturally dispersing." }, "composition": { "score": 0.86, "reason": "A single clearly framed bowl remains the focus against a restrained dark background, with harmonious warm colors and readable contrast on desktop and mobile. No material compositional defect or distracting UI is observed." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, mushrooms, garnish, and tofu-like blocks plausibly form vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5166666666666667, "toppings": 0.6333333333333333, "bowl": 0.7266666666666667, "broth": 0.5833333333333334, "steam": 0.6, "composition": 0.74 }, "reward": 0.6293333333333333, "sampleStdDev": 0.026006409466386063, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles form a neat coiled nest of fine concentric strands with golden color and some overlap. The loops are fairly uniform and flat, so they read as a repeated spiral pattern rather than supple strands. The nest is small relative to the bowl." }, "toppings": { "score": 0.62, "reason": "Shiitake with cross-scoring, a veined bok choy leaf, tofu cubes with grill lines and chive bits are all recognizable and varied. Tofu cubes are somewhat blocky and rigid, and there is no nori. Shapes are clean but simple." }, "bowl": { "score": 0.75, "reason": "The cream ceramic bowl has a coherent elliptical rim, a dark inner wall, a curved body with speckles and subtle glaze lines, and a soft contact shadow. The base is slightly vague and the bowl looks a bit floaty, but there is no major distortion." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange-brown with speckled oil flecks and reads as liquid. Sheen is restrained. The surface is fairly flat and uniform, and ingredients sit on it with little immersion, ripples or occlusion." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise from the bowl and change shape and position across frames, fading as they rise. They look like similar sinusoidal ribbons with vertical banding edges, somewhat mechanical, but they do not hide the food." }, "composition": { "score": 0.75, "reason": "A single centered bowl in a dark, restrained scene with warm colors and a clear focal point. There is no clutter or text. Lots of empty space and a slightly flat background, but the presentation is balanced." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, bok choy, tofu, chives and noodles are visible. There are no meat or egg items." } }, { "noodles": { "score": 0.5, "reason": "Noodles read as fine concentric coils in a nest, with a nice golden color and some overlap. The coils are fairly regular and flat, looking like a spiral scribble rather than supple strands. They sit on the broth without much immersion." }, "toppings": { "score": 0.6, "reason": "Shiitake with cross scoring, a bok choy leaf with veins, tofu cubes with grill lines and scallion bits are all recognizable and varied. They are stylized and somewhat flat, and the tofu looks like rigid blocks. No nori." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a believable cream ceramic body with speckles and curved lines, and a soft shadow. The base is rounded with no visible foot, so it looks slightly floating, and the rim is a bit thin." }, "broth": { "score": 0.55, "reason": "The broth is a warm amber-brown with speckles and a dark inner rim edge, which is appetizing. It is fairly flat and matte with little sheen or depth, and the toppings sit on it with little immersion." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise and shift across frames, fading softly. They are similar vertical S-curves, somewhat mechanical, and have visible hard vertical edges in places." }, "composition": { "score": 0.72, "reason": "The bowl is centered as a clear focal point on a dark, restrained background with warm colors and no clutter. Bowl and steam are well proportioned, though the scene is a bit plain." }, "vegan": { "score": 0.95, "reason": "Mushrooms, bok choy, tofu, scallions and noodles are all plant-based, and no animal products are visible." } }, { "noodles": { "score": 0.55, "reason": "Noodles form a coiled nest of fine concentric strands with a golden color and some overlap. The coils read as repeated, fairly flat loops, and the nest sits like a pasted disc on the broth with little volume." }, "toppings": { "score": 0.68, "reason": "Shiitake with cross-scoring, a veined bok choy leaf, tofu cubes with grill lines and scallion bits are all recognizable and varied. The tofu is blocky and the mushrooms a bit glossy, and there is no nori." }, "bowl": { "score": 0.75, "reason": "Cream ceramic bowl with a coherent elliptical rim, tapered body, speckle texture, subtle highlight lines and a grounded contact shadow. The base looks slightly pointed and floaty, but no material distortion is seen." }, "broth": { "score": 0.6, "reason": "Warm amber-brown broth with speckled oil spots and a dark inner rim gives depth. The surface is fairly flat and the toppings sit on it with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise from the bowl and change position and shape across frames, fading at the top. The plumes are similar in shape and a bit mechanical, with vertical banding artifacts." }, "composition": { "score": 0.75, "reason": "The bowl is centered on a dark, restrained background with warm colors and the food as the clear focus. There is no clutter or text, though the bowl is smallish and the space below is empty." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, bok choy, tofu, scallions and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.75, "bowl": 0.85, "broth": 0.75, "steam": 0.7666666666666667, "composition": 0.8666666666666667 }, "reward": 0.7675000000000001, "sampleStdDev": 0.049937460888595464, "judgments": [ { "noodles": { "score": 0.6, "reason": "Golden noodle nest with strand detail and scattered scallion flecks is recognizable, but the mass reads as concentric flat swirl loops rather than supple individual strands with volume." }, "toppings": { "score": 0.7, "reason": "Cross-scored shiitake slices, a veined green leaf, scallion bits and two grilled tofu slabs give readable variety; the tofu blocks are somewhat rigid, boxy slabs that border on generic blocks." }, "bowl": { "score": 0.85, "reason": "Coherent speckled ceramic bowl with clean elliptical rim, consistent wall taper, visible foot and soft ground shadow; only minor flatness in the surface detailing." }, "broth": { "score": 0.7, "reason": "Warm orange surface with oil-speckle texture and darker rim edge gives some depth and toppings sit at the surface, but the liquid reads fairly flat with limited immersion shading." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps visibly shift, fade and re-form across the time-ordered frames (desktop-1/2/3, pointer frames) and stay clear of the food; shapes are slightly repetitive stylized S-curls." }, "composition": { "score": 0.85, "reason": "Bowl is a clear focal point, centered with warm rim glow against a restrained dark background; adapts cleanly to mobile; no clutter or competing elements." }, "vegan": { "score": 1, "reason": "Visible ingredients are mushrooms, greens, tofu, scallions, noodles and broth only; no animal-derived toppings appear." } }, { "noodles": { "score": 0.65, "reason": "Noodles form a coherent central nest of fine, continuous golden strands with scallions scattered on top, sitting believably in the broth. However, the nest is largely a uniform concentric spiral with flat, repeated loops and no lifted or draped strands, so it reads somewhat mechanical within the stylized scene." }, "toppings": { "score": 0.7, "reason": "Two shiitake mushrooms with cross-scored caps and visible cream gills read clearly as mushrooms; bok choy leaves have vein detail, and chopped scallions add texture. The two yellow striped blocks read as fried tofu but are somewhat rigid, flat card-like forms with parallel line detailing, which caps the score." }, "bowl": { "score": 0.85, "reason": "Coherent cream ceramic bowl: elliptical rim with a darker inner edge, smoothly tapering walls with curved contour lines, speckled texture, subtle highlight, and a grounded foot with a soft contact shadow. Perspective between rim ellipse and body is consistent with no visible distortion or detached joins." }, "broth": { "score": 0.75, "reason": "Brown broth surface fills the bowl with a darker meniscus ring near the rim, small oil specks, and reasonable contact/occlusion where mushrooms and the noodle nest meet the liquid. Slightly flat and uniform in texture, but appetizing and readable as liquid." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the soup without hiding food. Across the time-ordered desktop frames the wisps clearly change shape, height and opacity (full plumes in frame 1, shifted in frame 2, faded in frame 3), supporting continuous rising motion; the reduced-motion frames show a static pattern, correctly honoring the preference. Slightly icon-like curl shapes keep it from scoring higher." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark atmospheric background with a restrained warm glow behind the bowl and a soft floor shadow; no text, UI or clutter. Mobile framing also fits cleanly without overflow. Only minor shortcoming is generous empty dark space below the bowl on mobile." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, bok choy, tofu/aburaage blocks, scallions, noodles and brown broth; no meat, fish, egg or dairy-like toppings appear." } }, { "noodles": { "score": 0.75, "reason": "Noodles form a clear swirl nest in the broth with thin strand-like stroke texture and scallion garnish on top; stylized and readable, though the concentric loop pattern reads slightly flat and repetitive rather than as individual supple strands with volume." }, "toppings": { "score": 0.85, "reason": "Two shiitake slices with scored caps and cream gills, bok-choy-like green leaves with stems, striped tofu/aburaage blocks, and scattered scallion bits are all recognizable with organic shapes, good scale, and variety; tofu blocks are a bit simple and rigid but nothing reads as generic plastic filler." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl with a clean elliptical rim, visible rim thickness, tapering walls, speckled glaze, soft highlight, and a grounded contact shadow; construction is consistent at the chosen viewing angle with only minor stylized flattening on the body." }, "broth": { "score": 0.8, "reason": "Appetizing amber broth with a darker meniscus ring near the rim, chili/oil speckles, and gentle gradient depth; toppings cast small shadows and sit in the liquid believably, though the surface sheen is quite restrained and slightly flat." }, "steam": { "score": 0.8, "reason": "Translucent wavy wisps rise above the bowl and change shape, count, and opacity across the time-ordered desktop frames (dense at ~1s, sparser and fainter by ~4.4s), showing continuous fade/rise without obscuring the food; the squiggle style is slightly mechanical in shape but clearly animated, and static in reduced-motion frames as expected." }, "composition": { "score": 0.9, "reason": "Centered focal dish with a warm rim glow, soft ground shadow, harmonious amber/cream/green palette, and no clutter, text, or competing elements; responsive tilt on pointer frames is subtle and the food remains the focus." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-based: noodles, mushrooms, leafy greens, tofu, scallions, and vegetable-looking broth; no eggs, meat, fish, or dairy appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.69999999999891, "frameIntervalP95Ms": 16.700000000000728, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a3576b823e006339e8a7fdf90378e2d759d2493a78198dc56ecfd9a626b7ad5a", "gradedAt": "2026-09-30T22:57:04.646269+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6633333333333333, "toppings": 0.7844444444444445, "bowl": 0.8533333333333334, "broth": 0.7077777777777778, "steam": 0.6766666666666667, "composition": 0.8633333333333333 }, "reward": 0.7580555555555556, "runId": "openai/gpt-6-astra/medium", "rewardStdDev": 0.04794161186430997, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.78, "bowl": 0.9166666666666667, "broth": 0.6766666666666667, "steam": 0.6466666666666667, "composition": 0.9 }, "reward": 0.7330000000000001, "sampleStdDev": 0.0030000000000000027, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin golden strands have continuous bends and highlights, but densely repeated horizontal loops form a flat, overly orderly stack rather than a supple tangled volume." }, "toppings": { "score": 0.78, "reason": "Veined greens, scored mushroom caps with stems, radish slices and browned tofu are clearly readable and varied. Shapes are somewhat stiff and uniformly smooth, especially the tofu and mushroom caps." }, "bowl": { "score": 0.92, "reason": "The elliptical rim, inner wall, rounded body and small foot join coherently. Speckled ceramic, subtle horizontal grooves, directional shading and a grounding shadow give convincing material and support; no material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading, scattered oil flecks and restrained curved highlights suggest soup. The surface remains rather flat, with limited meniscus or immersion cues around the ingredients." }, "steam": { "score": 0.64, "reason": "Faint translucent plumes emerge above the soup without obscuring it, and their contours shift between normal desktop frames. They remain broad and blurry, with limited visible evidence of distinct rising, spreading and fading wisps." }, "composition": { "score": 0.9, "reason": "A single well-framed bowl remains the clear focus against a restrained dark backdrop. Warm food colors contrast pleasantly with the pale ceramic and greens, with coherent lighting and no distracting text or clutter; mobile framing retains the entire dish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, leafy vegetables, radish, scallions and tofu alongside noodles. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands have clear highlights and some curved overlaps, but densely repeated horizontal loops dominate, creating a flat, mechanically arranged bundle rather than supple noodle volume." }, "toppings": { "score": 0.78, "reason": "Veined greens, scored mushroom caps with stems, radish slices, scallion rings and textured tofu provide readable variety. Shapes are cleanly illustrated, though tofu resembles spotted blocks and the leaves are somewhat stiff." }, "bowl": { "score": 0.91, "reason": "The continuous elliptical rim, shaded inner wall, rounded body and attached foot establish coherent depth and perspective. Fine speckling, curved ceramic bands and a contact shadow convincingly support the vessel; no material construction defect is observed." }, "broth": { "score": 0.67, "reason": "Warm amber shading, scattered oil-like marks and restrained surface highlights suggest soup. However, immersion is shallowly conveyed, with most ingredients appearing laid atop a relatively flat surface." }, "steam": { "score": 0.64, "reason": "Soft translucent wisps rise above the soup without obscuring it, and their contours vary slightly across normal desktop frames. They remain diffuse and column-like, with limited visible spreading or fading progression." }, "composition": { "score": 0.9, "reason": "A single clearly framed dish anchors the dark, uncluttered scene. Warm food colors contrast attractively with muted ceramic and background tones, with coherent shading and ample space for steam; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy vegetables, radish and scallions alongside noodles, with no visibly animal-derived ingredient." } }, { "noodles": { "score": 0.5, "reason": "Thin highlighted strands have readable bends and overlap, but closely packed, repeated horizontal loops dominate, making the noodles look like a flat arranged bundle rather than supple volume." }, "toppings": { "score": 0.78, "reason": "Veined greens, scored mushroom caps with stems, radish slices and porous golden tofu provide clear variety and detail. Shapes remain somewhat stiff and neatly diagrammatic, especially the tofu and mushroom caps." }, "bowl": { "score": 0.92, "reason": "Consistent elliptical rim, shaded inner wall, rounded body and integrated foot establish convincing depth and support. Fine speckles, curved ceramic bands and restrained highlights enrich the material; no material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading, small oil-like marks and peripheral highlights read as soup. Limited surface reflections and weak immersion cues make the ingredients appear arranged atop a relatively flat liquid plane." }, "steam": { "score": 0.66, "reason": "Faint translucent plumes change shape and position across the normal desktop frames without obscuring the dish. They are rather diffuse and column-like, with limited readable curling or spreading." }, "composition": { "score": 0.9, "reason": "A single clearly framed dish, balanced green and golden accents, soft contact shadow and restrained dark backdrop create a cohesive focal presentation across desktop and mobile. No material clutter or framing defect is observed." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, radish, tofu-like cubes and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.74, "toppings": 0.7733333333333333, "bowl": 0.7933333333333333, "broth": 0.6466666666666667, "steam": 0.55, "composition": 0.8066666666666666 }, "reward": 0.7278333333333333, "sampleStdDev": 0.030233811095085772, "judgments": [ { "noodles": { "score": 0.72, "reason": "Dense mound of fine golden strands with convincing overlap, swirl and volume. Strands look slightly repetitive and mechanical in loops, and the mound sits a bit like a pasted layer on the broth." }, "toppings": { "score": 0.74, "reason": "Shiitake with scored crosses, bok choy leaves with veins, tofu cubes with pores, radish slices and scallion/chili bits are all readable and varied. They look somewhat vector-flat and the tofu is blocky, but no plastic spheres dominate." }, "bowl": { "score": 0.78, "reason": "Coherent elliptical rim, inner wall, tapered body with throwing lines, foot ring and contact shadow. Speckle and highlights read as ceramic. The rim inner wall looks slightly flat and the base is small." }, "broth": { "score": 0.6, "reason": "Warm brown broth with some sheen and ingredients that sit in it. It is fairly flat and uniform, with little depth or reflection, and the contact between toppings and liquid is minimal." }, "steam": { "score": 0.5, "reason": "Faint translucent wisps rise above the bowl and change shape across frames. They are very subtle and somewhat blobby, and the reduced frames show a different curly form." }, "composition": { "score": 0.8, "reason": "Centered bowl on a dark vignette, warm food colors contrasting with the dark background, no clutter or text. Proportions are balanced and lighting is coherent, though a little empty around the bowl." }, "vegan": { "score": 0.95, "reason": "Mushrooms, bok choy, tofu, radish, scallion and noodles with no visible animal products." } }, { "noodles": { "score": 0.72, "reason": "Dense mass of fine, overlapping, swirling strands with highlights and believable volume. The strands are fairly uniform and read slightly as repeated loops, and the pile looks a bit flat atop the broth." }, "toppings": { "score": 0.78, "reason": "Shiitake with cross-cut scoring, bok choy leaves with veins, speckled tofu cubes, radish slices, scallion bits and chili are all readable and varied. They are stylized and somewhat flat, and nori is absent." }, "bowl": { "score": 0.78, "reason": "Coherent elliptical rim, inner wall, rounded body with ridge lines, foot and contact shadow. Matte ceramic speckle and highlights are convincing. Only minor flatness and the inner wall looks slightly translucent." }, "broth": { "score": 0.66, "reason": "Warm brown broth with subtle texture and ingredients sitting in it. It is fairly flat and matte with little sheen or depth, and the contact edges of the toppings are simple." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and change shape across frames, without hiding the food. They are very subtle and vague, and in places read as faint smudges." }, "composition": { "score": 0.8, "reason": "Centered bowl on a dark vignette, with a warm palette that contrasts well against the background. The food is the clear focus, with no clutter or text. A little empty space sits above the bowl." }, "vegan": { "score": 0.97, "reason": "Mushrooms, tofu, bok choy, radish, scallion and noodles in a plant-looking broth. No visible animal products." } }, { "noodles": { "score": 0.78, "reason": "Dense mound of fine golden strands with layered swirls and varied highlights reads as real noodles. The strands are somewhat uniform and parallel, and the mound looks slightly stacked rather than sitting in the broth." }, "toppings": { "score": 0.8, "reason": "Shiitake with cross-cut scoring and gills, bok choy leaves with veins and pale stems, textured tofu cubes with pores, and radish slices are all clearly readable and varied. The scallion and chili bits are tiny and a little scattered. No nori is shown. The tofu edges are slightly rigid." }, "bowl": { "score": 0.82, "reason": "The rim is a coherent ellipse with visible thickness and an inner wall, and the body curves into a foot ring. Ceramic speckle, glaze ridges and a soft contact shadow are present. The perspective is consistent, with only minor flatness in the shading." }, "broth": { "score": 0.68, "reason": "Warm amber-brown broth with subtle ripple lines and a matte sheen looks appetizing. The noodles and toppings sit on it with limited immersion, occlusion or reflection, and the surface is somewhat flat." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise above the bowl and change shape between frames. They are quite faint and sparse, and they start a little above the soup rather than emerging from it. The reduced-motion frames show clearer curved wisps, but the normal frames are subtle." }, "composition": { "score": 0.82, "reason": "A single centered bowl on a dark vignette with a good earthy palette. The food is the clear focal point, with no text or clutter. The bowl is well proportioned, with a lot of empty space above." }, "vegan": { "score": 0.97, "reason": "Only mushrooms, bok choy, tofu, radish, noodles and herbs are visible, with no egg, meat or fish." } } ], "diagnostics": { "vegan": 0.9633333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.75, "toppings": 0.8, "bowl": 0.85, "broth": 0.8, "steam": 0.8333333333333334, "composition": 0.8833333333333333 }, "reward": 0.8133333333333334, "sampleStdDev": 0.020816659994661344, "judgments": [ { "noodles": { "score": 0.8, "reason": "Noodles form a coherent central nest of continuous, gently curving strands with darker gaps and lighter highlights giving volume; they sit believably in the broth. Minor defect: the wavy strand pattern is somewhat repetitive/uniform in amplitude, slightly stylized rather than supple." }, "toppings": { "score": 0.8, "reason": "Clear variety: score-marked shiitake caps with stems, veined leafy greens, speckled tofu cubes, scallion bits and chili flakes. Shapes read organically; tofu cubes are a bit simple/rounded and the pink-rimmed radish slices are flat discs, slightly stylized but recognizable." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: clean elliptical rim with visible thickness and inner wall, speckled glaze with throwing-ridge lines and highlights, tapering body to a visible foot with contact shadow. Consistent perspective, no detached or distorted parts." }, "broth": { "score": 0.85, "reason": "Rich brown liquid surface with restrained sheen arcs and speckled texture; toppings sit in the broth with shadows and occlusion (mushroom stems, greens emerging, tofu partially submerged). No flat target discs or pasted-on look." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps above the soup that change shape and position across the time-ordered desktop frames (1\u21922\u21923 and pointer frames), indicating continuous rising/fading motion; reduced frames freeze identically. Wisps stay faint and never obscure the food." }, "composition": { "score": 0.85, "reason": "Bowl is the clear focal point, well centered and proportioned against a subtle dark vignette with warm rim light; harmonious palette, no text, UI or clutter; mobile crop keeps the dish readable." }, "vegan": { "score": 0.9, "reason": "Visible toppings are mushrooms, leafy greens, tofu, scallions, chili and pink-rimmed radish-style slices; no meat, fish, egg or dairy is evident. Only mild ambiguity from the pink-edged discs, which read as radish rather than fish cake." } }, { "noodles": { "score": 0.75, "reason": "Golden strands form a coherent layered nest with continuous curves, soft shading and a few draped ends over the front edge; slightly repetitive arc patterns and uniform strand thickness keep it below top marks." }, "toppings": { "score": 0.85, "reason": "Well-rendered variety: scored shiitake caps with believable gills, leafy bok-choy greens with stems, speckled fried tofu cubes, pink-edged radish slices, scallion bits and chili flakes; shapes are organic with only mild stylized gloss on the greens." }, "bowl": { "score": 0.85, "reason": "Coherent celadon ceramic bowl with a clean rim ellipse, speckled glazed body, horizontal throwing lines, a proper foot and grounded contact shadow; perspective between rim, wall and base is consistent." }, "broth": { "score": 0.8, "reason": "Rich miso-brown liquid surface with restrained sheen, subtle ripple arcs and darker occlusion halos around mushrooms; toppings read as sitting in the broth rather than pasted on top." }, "steam": { "score": 0.85, "reason": "Normal temporal frames (1.3s, 4.6s, 7.8s, pointer frames) show translucent grey wisps that shift shape, elongate and drift upward, indicating animated rising steam; wisps stay above the rim and never obscure the food." }, "composition": { "score": 0.9, "reason": "Centered, well-proportioned focal bowl on a dark atmospheric background with a soft vignette and contact shadow; no text, UI or clutter, and the mobile crop keeps the dish balanced and readable." }, "vegan": { "score": 0.6, "reason": "Mushrooms, greens, tofu, scallions, chili and noodles are clearly plant-based, but the white discs with pink rims are visually ambiguous between pickled radish and fish-cake (kamaboko) slices, warranting an intermediate score." } }, { "noodles": { "score": 0.7, "reason": "Continuous golden strands with wavy bends layer into a readable nest with slight under-shadow; competent, though the pattern is somewhat repetitive and uniform in line weight, reading a bit mechanical rather than supple individual strands." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: shiitake with score marks and gills, bok choy leaves with veins and stems, speckled tofu cubes, scallion and chili scatter, and pink-rimmed slices. Shapes are organic but slightly flat/stylized, and tofu cubes border on generic blocks." }, "bowl": { "score": 0.85, "reason": "Coherent celadon ceramic bowl: clean elliptical rim, thick wall with inner shadow, subtle speckle texture, visible foot and grounded contact shadow. No distortion or detached parts; perspective is consistent." }, "broth": { "score": 0.75, "reason": "Brown liquid surface with restrained sheen, small bubbles and a darker depth band near the rim; mushrooms and greens sit believably in it. Slightly uniform/flat color across the surface and the noodle nest reads somewhat placed atop rather than immersed." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rise above the bowl, spread and fade without hiding food; wisps change shape and position across desktop-1/2/3 and pointer frames, supporting continuous motion. Static in reduced-motion frames as expected." }, "composition": { "score": 0.9, "reason": "Centered, well-proportioned dish as clear focal point in a dark atmospheric scene with tasteful vignette; no text, UI or clutter, and the mobile frame adapts cleanly with the bowl still balanced." }, "vegan": { "score": 0.5, "reason": "Mushrooms, bok choy, tofu, scallion and chili are clearly plant-based, but the pink-rimmed white discs are ambiguous: they can read as radish slices or as narutomaki fish cakes (they lack the spiral but strongly resemble that topping), so an intermediate score applies." } } ], "diagnostics": { "vegan": 0.6666666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8766666666666666, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 499.9500000000007, "frameIntervalP95Ms": 483.3000000000029, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "d5ae7d55547e63fe15c2deb61d8ae9725e519a8b05a01c0ab659272838db3573", "gradedAt": "2026-09-30T22:57:00.782645+00:00", "scores": { "standalone": 1.0, "procedural": 0.0 }, "reward": 0.0, "runId": "openai/gpt-6-astra/xhigh", "rewardStdDev": null, "judgeResults": [], "standaloneViolations": [], "complianceViolations": [ "Embedded raster artwork; draw the scene with code instead" ], "diagnostics": {}, "performance": {} }, { "artifactSha256": "3ced3902547d1679947aaab1434b27fd3b2834fb900d5ca2631f818667d413c6", "gradedAt": "2026-09-30T23:04:45.786493+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.47888888888888886, "toppings": 0.5055555555555555, "bowl": 0.7744444444444445, "broth": 0.4877777777777778, "steam": 0.6288888888888888, "composition": 0.7711111111111111 }, "reward": 0.5972777777777778, "runId": "openai/gpt-6-luna/high", "rewardStdDev": 0.040614834910594096, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.47, "toppings": 0.53, "bowl": 0.8366666666666667, "broth": 0.44666666666666666, "steam": 0.62, "composition": 0.8233333333333333 }, "reward": 0.6138333333333333, "sampleStdDev": 0.0037859388972001857, "judgments": [ { "noodles": { "score": 0.48, "reason": "Continuous golden strands have smooth bends, but only a few flat, ribbon-like waves are visible, with little overlap, volume or convincing immersion." }, "toppings": { "score": 0.53, "reason": "Leaf clusters, scallion rings, small mushrooms and pale tofu pieces provide readable variety. Angular leaves, tiny disk-like mushrooms and glossy red ovals remain schematic rather than organically textured." }, "bowl": { "score": 0.83, "reason": "The elliptical rim, nested inner edge, rounded walls and attached foot form coherent perspective and depth. Warm shading and a left-side highlight suggest glazed ceramic; the broad highlight and heavy oval shadow are somewhat graphic." }, "broth": { "score": 0.46, "reason": "The warm brown surface is clearly bounded and has a few small specks, but reads as a flat opaque fill with minimal sheen or depth. Ingredients appear laid on top without convincing liquid contact." }, "steam": { "score": 0.62, "reason": "Normal frames show changes in the curvature and height of four translucent trails emerging behind the soup. Faded tips keep the food readable, but the narrow, similarly spaced strokes lack spreading and irregular wispy breakup." }, "composition": { "score": 0.82, "reason": "A single centered dish remains clear against a restrained dark, warm background, with no text or clutter and no cropping on mobile. The oversized bowl body relative to the small food cluster weakens the food's focal prominence." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, scallion rings, mushrooms, pale tofu-like pieces and red vegetables plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Continuous golden strands are readable, but the few broad, nearly parallel sinusoidal ribbons look flat and arranged rather than forming a supple overlapping noodle volume." }, "toppings": { "score": 0.53, "reason": "Leaf clusters, scallion rings, tiny mushrooms and pale tofu pieces provide variety. Angular leaves, glossy red ovals and minimally detailed cubes remain schematic, with little organic texture." }, "bowl": { "score": 0.83, "reason": "Consistent elliptical rim layers, smoothly curved walls, warm ceramic shading and an attached foot create coherent depth and support. The broad graphic highlight and very dark foot are slightly unrefined." }, "broth": { "score": 0.44, "reason": "Warm brown coloring and small surface flecks suggest soup, but the surface is matte and nearly solid-looking; ingredients appear laid on top without convincing immersion or contact ripples." }, "steam": { "score": 0.62, "reason": "Normal frames show changing curves in translucent trails rising from the soup and fading toward their tips. The four narrow, similarly styled strokes lack spreading and irregular wispy breakup." }, "composition": { "score": 0.82, "reason": "A centered, unobstructed dish, restrained warm palette and dark negative space maintain a clear focal point across desktop and mobile. No clutter is visible; the oversized bowl body makes the food area feel comparatively sparse." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, scallion rings, mushrooms, tomato-like red vegetables and tofu-like pale pieces plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.47, "reason": "Continuous golden strands have smooth bends, but only a few flat, similarly wavy ribbons are visible, with little overlap, volume, or submerged depth." }, "toppings": { "score": 0.53, "reason": "Leaf clusters, small mushrooms, scallion rings, and pale tofu pieces are recognizable. Angular leaves, tiny disk-like mushrooms, and glossy red ovals remain schematic rather than convincingly textured food." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, tapered curved walls, attached foot, and contact shadow form a coherent supported bowl. Warm shading and a defined rim convey material and thickness; the broad side highlight is somewhat graphic." }, "broth": { "score": 0.44, "reason": "Warm brown color and sparse speckles suggest broth, but the surface looks mostly flat and matte. Ingredients appear laid on top, with little visible immersion, contact sheen, or liquid depth." }, "steam": { "score": 0.62, "reason": "Normal frames show changing bends and heights in faint translucent trails that emerge behind the soup and fade upward without obscuring it. The narrow, nearly uniform strokes lack spreading and irregular wispy structure." }, "composition": { "score": 0.83, "reason": "A single centered bowl remains clearly framed on desktop and mobile, with restrained warm colors against a dark background and no distracting text or clutter. The oversized bowl body makes the food area feel comparatively sparse." }, "vegan": { "score": 1, "reason": "Visible leafy vegetables, scallion rings, mushrooms, tofu-like pale pieces, and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.42, "bowl": 0.6933333333333334, "broth": 0.5, "steam": 0.6, "composition": 0.6733333333333333 }, "reward": 0.551, "sampleStdDev": 0.0052915026221291225, "judgments": [ { "noodles": { "score": 0.5, "reason": "Wavy yellow strands are smooth and continuous but thin and sparse, like a few flat ribbons on the surface. They have little volume, overlap or immersion in the broth." }, "toppings": { "score": 0.42, "reason": "Bok choy tufts, tiny mushrooms, tofu cubes and scallion rings are recognizable, but they are very small and sparse. Red blobs and ring shapes look generic and plastic-like, and there is no nori." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body with a warm gradient and a highlight, and a grounded shadow. The base is slightly vague, and the flat vector look is simple but consistent." }, "broth": { "score": 0.5, "reason": "The broth has a warm radial gradient and concentric ring lines that read like a target disc. It has little texture, depth or sheen, and the ingredients sit on top rather than in it." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise and sway differently between frames, without hiding the food. They are few, narrow and similar in style, and they fade softly." }, "composition": { "score": 0.68, "reason": "The bowl is centered and is the clear focal point against a dark vignette, with a warm palette and no clutter. The food on the surface is small, so the bowl feels a bit empty." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are noodles, bok choy, mushrooms, tofu and scallions. No animal products appear." } }, { "noodles": { "score": 0.5, "reason": "Noodles are smooth, wavy and continuous with a nice golden color, but they are thin and sparse, sit as flat ribbons on the surface, and look like a few repeated squiggles rather than a volume of noodles." }, "toppings": { "score": 0.42, "reason": "Bok choy clusters, small mushrooms, tofu cubes and scallion rings are recognizable but tiny and symbolic. The red balls look like glossy plastic, the green rings are generic, and there is no nori. They are sparse and sit on top of the surface." }, "bowl": { "score": 0.68, "reason": "Coherent elliptical rim with a cream lip and dark band, a rounded body with smooth shading, a highlight and a ground shadow. The base is hidden and the bowl reads more as a copper soup pot than ramen ceramic, but there is no distortion." }, "broth": { "score": 0.5, "reason": "Warm orange-brown gradient with thin concentric ring lines that read as a flat target disc. Little sheen or depth, and the ingredients look pasted on the surface rather than immersed." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise and sway differently across frames and fade upward without hiding the food. They are thin line-like strands that look somewhat uniform, but they show continuous motion." }, "composition": { "score": 0.66, "reason": "Centered bowl with a dark vignette and warm glow gives a clear focal point with no clutter. The food is small relative to the bowl, which leaves a large empty surface and a somewhat plain presentation." }, "vegan": { "score": 0.88, "reason": "Only noodles, bok choy, mushrooms, tofu cubes and scallion are visible. There is no meat, egg or fish. The red balls are ambiguous but read as tomato or chili." } }, { "noodles": { "score": 0.5, "reason": "Noodles are smooth, wavy, continuous strands with plausible thickness, but they are few, sparse, and sit on the surface as flat ribbons with repeated wave patterns. They lack volume and overlap, and do not read as immersed in broth." }, "toppings": { "score": 0.42, "reason": "Bok choy tufts, tiny mushrooms, tofu cubes and scallion rings are recognizable, but they are very small and sparse. The red glossy blobs look like plastic spheres, and the green rings are generic. No nori is visible." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a double ring, a deep rounded body with a gradient, a highlight and a contact shadow. Perspective is consistent. The base is slightly vague and the shading is somewhat flat vector, but there are no broken joins." }, "broth": { "score": 0.5, "reason": "The warm orange-brown surface has a radial glow and concentric rim lines, which read as liquid. It is fairly flat and target-like, with little ripple or sheen, and the ingredients sit on top with little immersion." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise from the bowl and change shape between frames, drifting and fading. They are few, thin, and rather similar to each other, but they don't obscure the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered, the dark vignette and warm glow keep it the focal point, and the colors are cohesive. The food on top is tiny relative to the big bowl, and the bowl body dominates the frame." }, "vegan": { "score": 0.9, "reason": "Only noodles, bok choy, mushrooms, tofu cubes and scallions are visible. There are no meat, egg or fish items. The red blobs are ambiguous but are likely tomato or chili." } } ], "diagnostics": { "vegan": 0.8933333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.4666666666666667, "toppings": 0.5666666666666667, "bowl": 0.7933333333333333, "broth": 0.5166666666666667, "steam": 0.6666666666666666, "composition": 0.8166666666666667 }, "reward": 0.627, "sampleStdDev": 0.05388181511419228, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles appear as a few thin wavy yellow squiggles lying flat on the broth surface; they lack volume, coherent bundling, overlap and immersion cues, with some short broken segments, reading more as decorative strands than a plausible noodle mass." }, "toppings": { "score": 0.45, "reason": "Two scallion bunches and small brown mushroom slices read well, but the glossy red sphere, featureless white cubes and plain green rings are generic/icon-like; overall the topping set is sparse and flat with little texture or scale variety." }, "bowl": { "score": 0.75, "reason": "Coherent stylized construction: cream rim band with dark broth inset, rounded terracotta body with gradient and highlight arc, small foot and grounding shadow. The very shallow rim ellipse against the deep hemispherical body is slightly mismatched, but it stays within acceptable stylized perspective." }, "broth": { "score": 0.45, "reason": "Flat orange surface with a central radial glow and a single highlight ring near the rim is readable in a flat style, but toppings and noodles cast no shadows or occlusion and show no immersion, so they appear pasted on top of the liquid rather than sitting in it." }, "steam": { "score": 0.6, "reason": "Several translucent, irregular, tapering wisps rise from the soup and fade without hiding the food. Their varied heights and curves avoid a mechanical look, but with only one legible still the rising/fading motion across the temporal frames cannot be confirmed." }, "composition": { "score": 0.8, "reason": "Single centered focal bowl on a dark ground with a warm ambient glow and soft contact shadow; balanced proportions, restrained warm palette, no text, clutter or competing elements. Plain but coherent presentation." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are scallions, mushrooms, tofu-like white cubes, a tomato-like red sphere, green vegetable rings and a thin seaweed-like strip in plant-colored broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.45, "reason": "Only a few smooth yellow strands lie flat on the broth as uniform sine-wave squiggles; they read as rigid ribbon/worm shapes with no volume, overlap, or immersion rather than a believable noodle bed." }, "toppings": { "score": 0.65, "reason": "Readable variety: green scallion-leaf bundles, brown mushroom caps, white tofu cubes, scallion rings, and red spheres. Shapes are simple flat vector icons and the red glossy balls and rings look slightly generic/plastic, but nothing is a featureless card dominating the scene." }, "bowl": { "score": 0.78, "reason": "Coherent stylized bowl: cream rim ring, terracotta hemispherical body with gradient shading and highlight, visible foot and contact shadow. Slight flatness at the very bottom edge and a wide rim ellipse are minor, but construction and perspective join cleanly." }, "broth": { "score": 0.55, "reason": "The broth is a smooth orange ellipse with faint concentric ring strokes near the rim, reading somewhat like a flat filled disc. Ingredients rest on the surface with essentially no occlusion or immersion cues, appearing pasted on top." }, "steam": { "score": 0.65, "reason": "Thin translucent wavy wisps rise above the bowl and clearly change shape between desktop-1, desktop-2, desktop-3 and pointer frames, so genuine ambient motion exists; reduced frames freeze them appropriately. Wisps are a bit string/line-like rather than soft volumetric vapor, but they spread and fade." }, "composition": { "score": 0.8, "reason": "Centered focal bowl with balanced proportions, warm orange-on-dark palette, subtle radial glow and soft ground shadow; no clutter, text, or competing elements. Mobile crops the scene cleanly with no overflow." }, "vegan": { "score": 0.9, "reason": "Visible toppings are scallion greens, mushrooms, tofu cubes, scallion rings, and red spheres plausibly read as tomato/chili; no egg, meat, fish, or dairy forms are visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are a few flat wavy squiggle strands lying on the surface; continuous and smooth but lacking volume, overlap, and broth immersion, reading as pasted ribbons rather than a coherent noodle mass." }, "toppings": { "score": 0.6, "reason": "Readable variety: bok-choy-like greens, small brown mushrooms, white tofu cubes, scallion rings, red spheres, and sprouts. Shapes are simple sticker-like symbols with weak contact shadows and sparse scale; the red glossy spheres are the least convincing." }, "bowl": { "score": 0.85, "reason": "Coherent stylized bowl: cream rim band with dark inner ring, smooth terracotta walls with a subtle sheen streak, visible foot and grounded contact shadow. Rim ellipse and body perspective are consistent for this viewing angle; no broken joins observed." }, "broth": { "score": 0.55, "reason": "Warm orange-brown surface with soft radial glow and light specks reads as soup, but immersion is weak: toppings cast no occlusion and appear pasted on top, and the thin concentric ellipses near the rim read somewhat flat." }, "steam": { "score": 0.75, "reason": "Translucent wavy wisps rise from the bowl and fade without hiding food; their shapes and positions change across desktop-1 through pointer frames, indicating animation. Wisps are thin uniform lines and somewhat parallel, a simplified rather than volumetric representation; static in reduced frames as expected." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric backdrop with a warm glow pool and soft shadow; no text, UI, or clutter; food remains the clear focal point." }, "vegan": { "score": 1.0, "reason": "Visible toppings are greens, mushrooms, tofu cubes, scallions, tomatoes/sprouts, and noodles in plant broth; no animal-derived ingredients apparent." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9533333333333334, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.70000000000073, "frameIntervalP95Ms": 83.39999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "da9780b9e62893c203b32cd4b6ae6fd688ec7a9f213d5fa7afb05c12ba382229", "gradedAt": "2026-09-30T23:05:04.567942+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "runId": "openai/gpt-6-luna/low", "rewardStdDev": 0.0, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0, "reason": "No noodles are visible; all captures show only a dark gradient." }, "toppings": { "score": 0, "reason": "No toppings or recognizable ingredients are visible." }, "bowl": { "score": 0, "reason": "No bowl, rim, cavity, or base is visible." }, "broth": { "score": 0, "reason": "No soup or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam is visible in any time-ordered frame." }, "composition": { "score": 0, "reason": "The dark background has a faint warm glow, but no ramen dish or focal subject appears." }, "vegan": { "score": 0.5, "reason": "No ingredients are discernible, so vegan compliance cannot be assessed; no animal-derived ingredients are visibly present." } }, { "noodles": { "score": 0, "reason": "No noodles are visible in any frame; only a dark background gradient appears." }, "toppings": { "score": 0, "reason": "No toppings or recognizable food ingredients are visible." }, "bowl": { "score": 0, "reason": "No bowl, rim, cavity, or supporting base is visible." }, "broth": { "score": 0, "reason": "No soup or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam is visible across the time-ordered frames." }, "composition": { "score": 0, "reason": "The scene is blank apart from a faint warm gradient, with no focal ramen dish." }, "vegan": { "score": 0.5, "reason": "No ingredients are discernible, so vegan compliance cannot be established; no animal-derived ingredients are visible either." } }, { "noodles": { "score": 0, "reason": "No noodles are visible in any frame." }, "toppings": { "score": 0, "reason": "No toppings or discernible food appear." }, "bowl": { "score": 0, "reason": "No bowl, rim, cavity, or base is visible." }, "broth": { "score": 0, "reason": "No soup or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam appears in the time-ordered frames." }, "composition": { "score": 0, "reason": "Only a dark background with a faint warm glow is shown; the focal ramen dish is entirely absent." }, "vegan": { "score": 0.5, "reason": "No ingredients are discernible, so vegan compliance cannot be assessed; no animal-derived ingredients are visibly present." } } ], "diagnostics": { "vegan": 0.5 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0, "reason": "All frames show only a dark radial gradient; no noodles are visible." }, "toppings": { "score": 0, "reason": "No toppings are visible; the page is blank apart from the background gradient." }, "bowl": { "score": 0, "reason": "No bowl is rendered. The capture reports JavaScript errors on every viewport." }, "broth": { "score": 0, "reason": "No broth or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam is visible, and the desktop frames are identical over time." }, "composition": { "score": 0, "reason": "The scene is an empty dark vignette with no focal dish, so it is a non-ramen scene." }, "vegan": { "score": 0, "reason": "No ramen ingredients are discernible, so vegan compliance cannot be credited." } }, { "noodles": { "score": 0, "reason": "All frames show only a dark radial gradient; no noodles are visible." }, "toppings": { "score": 0, "reason": "No toppings are present in any frame." }, "bowl": { "score": 0, "reason": "No bowl is rendered. The page errors (JavaScript error) likely stopped the scene from drawing." }, "broth": { "score": 0, "reason": "No broth or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam is visible, and the frames are identical across time." }, "composition": { "score": 0, "reason": "The frame is an empty dark vignette with a faint brown glow and no focal dish, so there is no ramen scene." }, "vegan": { "score": 0.1, "reason": "There are no discernible ramen ingredients, so vegan compliance cannot be assessed. No animal products are visible either." } }, { "noodles": { "score": 0, "reason": "All frames show only a dark radial gradient; no noodles are visible." }, "toppings": { "score": 0, "reason": "No toppings are present in any frame." }, "bowl": { "score": 0, "reason": "No bowl is rendered. Page errors suggest the scene failed to draw." }, "broth": { "score": 0, "reason": "No broth or liquid surface is visible." }, "steam": { "score": 0, "reason": "No steam is visible, and the frames are identical across time." }, "composition": { "score": 0, "reason": "The scene is an empty dark vignette with no dish, so there is no focal subject." }, "vegan": { "score": 0, "reason": "No ramen ingredients are visible, so vegan compliance cannot be credited." } } ], "diagnostics": { "vegan": 0.03333333333333333 }, "veganFailed": true }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0, "reason": "No noodles are visible; the scene is entirely blank apart from a faint central glow." }, "toppings": { "score": 0, "reason": "No toppings of any kind appear in any frame; the canvas is empty." }, "bowl": { "score": 0, "reason": "No bowl, rim, cavity or vessel is rendered anywhere in the frames." }, "broth": { "score": 0, "reason": "No liquid surface or broth is present; only a dark background with a diffuse orange glow." }, "steam": { "score": 0, "reason": "No steam wisps appear in any temporal frame; there is nothing but static ambient glow." }, "composition": { "score": 0, "reason": "Blank/non-ramen scene: just a dark gradient with a faint central orange glow and no focal dish." }, "vegan": { "score": 0.25, "reason": "No ingredients are visible at all, so no animal-derived items appear, but no vegan ramen ingredients can be verified either; ambiguous earns only partial credit." } }, { "noodles": { "score": 0, "reason": "No noodles are visible; every frame shows only a dark screen with a faint central glow. The scene is effectively blank." }, "toppings": { "score": 0, "reason": "No toppings of any kind appear in any captured frame; the canvas is empty apart from a dim warm gradient." }, "bowl": { "score": 0, "reason": "No bowl, rim, cavity, or vessel is present or readable in any frame." }, "broth": { "score": 0, "reason": "No liquid surface or soup is visible; the only color is an ambient background glow, not broth." }, "steam": { "score": 0, "reason": "No steam wisps appear in any frame, and frames show no temporal change other than nothing at all; steam is entirely absent." }, "composition": { "score": 0, "reason": "No focal dish or ramen scene exists \u2014 only a near-black empty canvas with a subtle glow. JavaScript errors in all contexts suggest the render failed." }, "vegan": { "score": 0.5, "reason": "No ingredients are visible at all, so no clearly animal-derived toppings appear, but there is also no evidence of vegan ramen ingredients; an empty scene cannot earn full compliance credit." } }, { "noodles": { "score": 0, "reason": "No noodles or any ramen scene are visible; all frames show only a near-black background with a faint warm glow. The page reports JavaScript errors and renders nothing." }, "toppings": { "score": 0, "reason": "No toppings, ingredients, or any food imagery appear in any captured frame; the scene failed to render." }, "bowl": { "score": 0, "reason": "No bowl, rim, or vessel of any kind is present in the frames; only an empty dark gradient with a dim central glow." }, "broth": { "score": 0, "reason": "No broth surface or liquid is visible; the frames contain no ramen scene whatsoever." }, "steam": { "score": 0, "reason": "No steam wisps appear in any time-ordered frame; there is no soup to emit steam, only a static glow unchanged across frames." }, "composition": { "score": 0, "reason": "The composition is an empty dark frame with a faint central glow and no focal dish; effectively a blank non-ramen scene." }, "vegan": { "score": 0, "reason": "No ingredients are visible at all, so vegan plausibility cannot be established; a scene with no discernible ramen ingredients earns no credit." } } ], "diagnostics": { "vegan": 0.25 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.2611111111111111, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.699999999999818, "frameIntervalP95Ms": 16.700000000000728, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "46a4d3b4bafd7244df280e6bc6ab2a78427805c97d5163235ea94e3af7b323fe", "gradedAt": "2026-09-30T23:02:45.630515+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5222222222222223, "toppings": 0.62, "bowl": 0.741111111111111, "broth": 0.5966666666666667, "steam": 0.6844444444444444, "composition": 0.7977777777777778 }, "reward": 0.6497777777777778, "runId": "openai/gpt-6-luna/max", "rewardStdDev": 0.07794056377446525, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.66, "bowl": 0.7733333333333333, "broth": 0.54, "steam": 0.6866666666666666, "composition": 0.8266666666666667 }, "reward": 0.6593333333333333, "sampleStdDev": 0.005033222956847171, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and some topping occlusion, but mostly form repeated flat waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.66, "reason": "Scored mushroom caps, shaded tofu cubes, scallion rings and veined greens provide readable variety. Shapes remain very schematic, with smooth leaf silhouettes and nearly identical mushrooms lacking organic texture." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inner cavity and curved shaded walls form a coherent lacquer-like vessel with restrained highlights. Ground contact is weak, especially on mobile where the bowl visibly floats above its shadow." }, "broth": { "score": 0.55, "reason": "Warm amber shading defines a readable soup surface, but there is little liquid sheen or contact detail; ingredients largely appear laid over a flat surface rather than partly immersed." }, "steam": { "score": 0.68, "reason": "Normal frames show translucent wisps changing height and opacity without obscuring the food. The narrow, similarly sinusoidal strokes remain mechanically uniform and show limited spreading." }, "composition": { "score": 0.82, "reason": "A single clearly readable bowl anchors a restrained dark scene, with warm gold and brown contrasted by green toppings and no distracting text or UI. Mobile leaves excessive empty space, but the desktop focal balance is strong." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, scallions, corn-like kernels and noodles plausibly depict vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and some overlap, but repeated sinusoidal ribbons lie in a flat layer with little convincing noodle volume." }, "toppings": { "score": 0.66, "reason": "Scored mushroom caps, shaded tofu cubes, scallion rings and veined green leaves are readable and varied. Shapes remain highly schematic, with nearly identical mushrooms, smooth tofu and a stiff leaf cluster." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, rounded walls and layered warm highlights form a coherent lacquer-like vessel. Ground contact is weak, especially on mobile where the bowl appears suspended well above its shadow." }, "broth": { "score": 0.53, "reason": "Warm amber shading defines a readable soup surface, but there is little liquid sheen or surface texture, and ingredients largely appear laid onto it rather than immersed." }, "steam": { "score": 0.68, "reason": "Normal frames show translucent trails shifting upward and changing opacity without obscuring the food. The narrow, similarly wavy lines remain mechanically shaped, with limited spreading and some disconnected lower ends." }, "composition": { "score": 0.82, "reason": "A single uncluttered dish is clearly emphasized by warm gold and green accents against a restrained dark background, with no distracting text or UI. Mobile framing leaves substantial empty space, slightly weakening the balance." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, scallions, corn-like kernels and noodles plausibly constitute vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and readable crossings, but repeated sinusoidal loops lie in a flat band with little tangled volume or convincing immersion." }, "toppings": { "score": 0.66, "reason": "Scored mushroom caps, shaded tofu cubes, scallion rings and veined leaves provide recognizable variety. Shapes remain schematic, with nearly matching mushrooms, smooth tofu and stiff, oversized leaves." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inset cavity and curved walls form coherent construction, with warm edge highlights and lacquer-like shading. Ground contact is weak, especially on mobile where the bowl visibly floats above its shadow." }, "broth": { "score": 0.54, "reason": "Warm amber shading clearly defines the soup surface, but there is little liquid sheen, surface texture or contact detail; ingredients largely appear laid on top rather than immersed." }, "steam": { "score": 0.7, "reason": "Normal frames show translucent wisps changing height and opacity without obscuring the food. The narrow, similarly wavy strokes lack much spreading or turbulent variation, and some lower ends detach from the soup." }, "composition": { "score": 0.84, "reason": "A single readable dish, restrained dark background and harmonious amber, brown and green palette create a clear focal point without text or clutter. Mobile framing leaves substantial unused space, but overall presentation remains balanced." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn-like kernels and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.48333333333333334, "toppings": 0.5, "bowl": 0.6333333333333333, "broth": 0.55, "steam": 0.6166666666666667, "composition": 0.7166666666666667 }, "reward": 0.5675, "sampleStdDev": 0.02317865397299851, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin wavy golden strands read as noodles and overlap a little, but they are flat, uniform and wire-like, with little volume or immersion in the broth." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, shiitake-like mushrooms, a leafy green sprig, scallion rings and corn-like dots are all recognizable. They are small and stylized, the cubes look like generic blocks, and the dark nori is barely readable." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body, copper highlights and a shadow. The broth sits in a recessed inner ring. The base is pointed and seems to float above the shadow, which weakens the support and contact." }, "broth": { "score": 0.6, "reason": "The orange-red gradient broth reads as warm and appetizing, with a dark inner rim giving some depth. It is fairly flat, with little sheen or surface texture, and the ingredients look laid on top of it." }, "steam": { "score": 0.65, "reason": "Thin wavy wisps rise and change position and fade across frames. They are translucent and leave the food visible. They look like similar line plumes and are not very wispy or spreading." }, "composition": { "score": 0.75, "reason": "A single centered bowl in a dark, warm-glow setting keeps the focus on the dish. The colors are harmonious. The shadow is slightly detached and the ingredients are small relative to the bowl." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, greens, scallions and noodles are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin, wavy golden strands that overlap, but they are uniform sine-like ribbons with repeated flat loops. They read as noodles but look schematic and sparse." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, shiitake-like mushroom caps, a leafy green sprig, corn kernels and scallion rings are all recognizable. The tofu cubes look like rigid blocks and the mushrooms are flat. The nori is a dark blocky shape, and the topping set is small and symbolic." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, inner wall, curved body, copper highlights and a ground shadow. The base comes to a pointed bottom and seems to float above the shadow with no foot ring. The glaze is fairly flat." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange gradient with a dark inner rim. It reads as soup, but the surface is flat with little sheen or ripple. Ingredients sit on it more than in it." }, "steam": { "score": 0.6, "reason": "Thin translucent wavy wisps rise, shift and fade between frames, and they do not cover the food. They are narrow, similar line-like plumes, so they look a bit mechanical." }, "composition": { "score": 0.7, "reason": "The bowl is centered on a dark, moody background with a warm glow and no distracting UI. Colors are cohesive and the food is the focus. The bowl sits a little low and the toppings are small relative to it." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are tofu, mushrooms, greens, corn, scallion and noodles. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.45, "reason": "Noodles are thin wavy golden lines that read as strands, but they are flat, repetitive sine-like waves with little volume or overlap, and they look like scribbles on the surface." }, "toppings": { "score": 0.5, "reason": "Mushrooms with gill lines, tofu cubes, a leafy green sprig and scallion rings are recognizable and varied. The forms are flat and stylized, the tofu cubes look like generic blocks, and the nori is a dark square-ish shape." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a wall with highlights and a copper lip, and a soft base shadow. The base looks slightly pointed and floats above the shadow, and the rim has a double-ring look." }, "broth": { "score": 0.5, "reason": "The orange gradient broth reads as warm and appetizing. It is flat and the toppings sit on top with little immersion, ripples or sheen." }, "steam": { "score": 0.6, "reason": "Thin translucent wavy wisps rise and change between frames, fading at the top and not hiding the food. They are thin, similar in shape and somewhat mechanical." }, "composition": { "score": 0.7, "reason": "A centered bowl on a dark, warm vignette, with no distracting text or UI. The palette is cohesive and the food is the focus, though the bowl seems to float and the soup surface is small." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, tofu, greens, scallions and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.9266666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5833333333333334, "toppings": 0.7, "bowl": 0.8166666666666667, "broth": 0.7, "steam": 0.75, "composition": 0.85 }, "reward": 0.7225, "sampleStdDev": 0.006614378277661503, "judgments": [ { "noodles": { "score": 0.55, "reason": "Continuous pale-yellow strands weave between toppings with coherent crossings and subtle highlights, but they render as uniform flat squiggle lines painted on the broth surface \u2014 little volume, clumping, or depth; readable yet stylized-flat." }, "toppings": { "score": 0.7, "reason": "Good variety: two shiitake slices with cross-hatched gills, tofu cubes with believable 3D faces, scallion rings, chili bits, a dark seaweed piece and a stylized leaf cluster. All read as food; minor defects are the vague red chili blobs and somewhat sparse, decorative arrangement." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with inner lip, smoothly curved brown ceramic body, believable thickness, ground shadow and support; consistent perspective between rim and body. Only minor blemish is the thin scratch-like highlight strokes on the body wall." }, "broth": { "score": 0.65, "reason": "Warm appetizing orange gradient with a darker inner edge giving some depth, and small specks for texture. However, toppings sit on the surface with minimal occlusion or immersion shadows, so the liquid reads somewhat flat." }, "steam": { "score": 0.8, "reason": "Translucent thin wisps clearly morph and drift across the desktop frames (desktop-1 through pointer-left show different curvatures and lengths), rising from the soup and fading without obscuring food. Slightly squiggle-like cartoon linework, but genuinely animated and irregular; reduced frames correctly hold static." }, "composition": { "score": 0.85, "reason": "Balanced centered dish against a dark atmospheric background with restrained warm glow, subtle particles and soft shadow; no text, UI or competing elements, and the food remains the focal point." }, "vegan": { "score": 1, "reason": "Visible toppings are mushrooms, tofu, scallions, chili, leafy greens and seaweed in an orange plant-based-looking broth; no animal-derived ingredients are apparent." } }, { "noodles": { "score": 0.6, "reason": "Cream wavy strands read as noodles and stay continuous across the broth, but they are flat schematic sine-wave ribbons with repetitive, uniform bends and little volume or coherent overlap, lying painted-on rather than clumped in the soup." }, "toppings": { "score": 0.7, "reason": "Stylized but recognizable set: dark cross-hatched shiitake caps, 3D tofu cubes with small shadows, scallion rings, a corn/scallion cluster, a leafy herb sprig and a nori sheet behind it. Shapes are readable though somewhat simplified; mushrooms are plain domes and the cubes are geometrically crisp." }, "bowl": { "score": 0.8, "reason": "Coherent lacquer-style bowl: light copper rim, dark inner shadow band, smoothly curved brown body with soft highlights and a grounded elliptical shadow. Rim and body perspective are consistent with the slightly elevated view; only mild stylized thickening of the rim." }, "broth": { "score": 0.75, "reason": "Appetizing amber-orange surface with center glow and darker edge shading, plus chili-oil specks; toppings sit on the surface with small contact shadows. Slight flatness and a schematic feel keep it from higher marks." }, "steam": { "score": 0.75, "reason": "Translucent thin wisps rise from the soup and fade before obscuring food; their shapes and positions clearly change across desktop-1/2/3, indicating animated, non-static steam. Wisps are simple line squiggles rather than volumetric plumes." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric background with subtle warm glow and sparse dust motes; no text, UI or competing elements, and the mobile frame adapts cleanly. The dish remains the clear focal point." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu cubes, mushroom caps, scallion rings, corn/veg dots, herb leaves, nori and chili oil; no meat, fish, egg or dairy forms appear." } }, { "noodles": { "score": 0.6, "reason": "Noodles read as smooth continuous pale-yellow strands with consistent thickness and highlights, but they are flat surface squiggles in a few clusters with visibly repeated sine-wave shapes and no sense of volume or immersion." }, "toppings": { "score": 0.7, "reason": "Good variety: two scored shiitake caps with highlights, shaded tofu cubes, a green herb-leaf cluster, scallion rings, corn dots, chili slivers and a dark nori piece. Shapes are organic and readable, though tofu cubes are slightly simplistic and garnish coverage is sparse." }, "bowl": { "score": 0.8, "reason": "Coherent wide shallow donburi: clean elliptical rim with tan ceramic band, smooth maroon body with gradient shading and highlight streaks, grounded by a soft contact shadow. Minor shortcoming: broth sits nearly flush with the rim and interior wall depth is minimal." }, "broth": { "score": 0.7, "reason": "Warm orange gradient surface reads as miso-style broth viewed at an angle with a darker tare ring at the wall. Toppings rest believably on the surface, but occlusion/shadow integration is light, so items look slightly laid on top rather than sitting in liquid." }, "steam": { "score": 0.7, "reason": "Thin translucent wavy wisps rise from the soup, fade toward the top, and change shape and position across desktop-1/2/3 and pointer frames, confirming animation; reduced frames are static as expected. Wisps are slightly formulaic similar curves and a bit wire-thin." }, "composition": { "score": 0.85, "reason": "Centered focal dish in a restrained dark scene with a subtle warm glow, dust specks, and no text, UI or clutter; scales cleanly to mobile with no overflow. Only minor weakness: generous empty dark space around the bowl." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu, scallion, corn, leafy herbs, chili and seaweed in plant broth; no meat, fish, egg or dairy appears." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9755555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.699999999999818, "frameIntervalP95Ms": 33.400000000000546, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "e56611460e9b47992c6e74fbe19417f61cc1623741cedf6ea2017c4520595b28", "gradedAt": "2026-09-30T23:05:04.778633+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5277777777777778, "toppings": 0.6288888888888888, "bowl": 0.6877777777777778, "broth": 0.5844444444444444, "steam": 0.6755555555555556, "composition": 0.79 }, "reward": 0.6391111111111112, "runId": "openai/gpt-6-luna/medium", "rewardStdDev": 0.06782664449686206, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.61, "bowl": 0.7633333333333333, "broth": 0.48666666666666664, "steam": 0.6933333333333334, "composition": 0.85 }, "reward": 0.6428333333333334, "sampleStdDev": 0.0005773502691896262, "judgments": [ { "noodles": { "score": 0.5, "reason": "Readable golden strands curve around and behind toppings, but sparse, flat loops with inconsistent thickness provide little convincing noodle volume or immersion." }, "toppings": { "score": 0.61, "reason": "Veined green leaves, scored mushroom caps, tofu cubes and scallion rings are recognizable and varied. Shapes remain very simplified, with smooth leaves and textureless geometric tofu." }, "bowl": { "score": 0.76, "reason": "The elliptical layered rim and rounded shaded walls form a coherent vessel with readable thickness. Extra sweeping wall outlines and a weak contact shadow make the lower construction feel slightly artificial." }, "broth": { "score": 0.48, "reason": "Warm orange shading suggests broth, but the surface is largely a matte, featureless ellipse. Ingredients appear laid above it with little liquid contact, reflection or partial submersion." }, "steam": { "score": 0.7, "reason": "Normal frames show changes in the positions and visibility of soft translucent wisps rising above the soup without obscuring it. The narrow, similarly shaped sinusoidal trails look somewhat mechanical rather than spreading naturally." }, "composition": { "score": 0.85, "reason": "A single centered dish stays clearly readable against a restrained dark brown background, with coordinated warm food colors, cream rim highlights and ample negative space. No distracting text or clutter is present; the mobile dish is somewhat small." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushroom caps, scallion rings and noodles plausibly form vegan ramen. No visibly animal-derived topping is observed." } }, { "noodles": { "score": 0.5, "reason": "Readable golden strands have smooth bends and some overlap, but sparse, flat looping strokes and inconsistent thickness provide little convincing noodle volume or immersion." }, "toppings": { "score": 0.61, "reason": "Veined green leaves, scored mushroom caps, scallion rings and shaded tofu cubes provide recognizable variety. Shapes remain overly regular and smooth, with minimal organic texture." }, "bowl": { "score": 0.77, "reason": "Coherent elliptical rim, rounded walls and layered shading establish a substantial vessel. Extra exterior contour strokes and a slightly floating relationship to the broad shadow weaken material and contact." }, "broth": { "score": 0.49, "reason": "Warm reddish shading makes the soup readable, but the surface is nearly featureless and solid-looking; ingredients lack convincing wet contact, reflections or partial immersion." }, "steam": { "score": 0.68, "reason": "Normal frames show changing positions and opacity of thin translucent wisps above the soup without obscuring food. The similarly shaped sinusoidal strokes remain mechanical, with limited visible spreading." }, "composition": { "score": 0.85, "reason": "A centered, clearly readable dish, restrained warm palette and dark textured surround create a balanced focal scene without text or clutter. Mobile framing preserves the dish; no material composition defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu cubes, mushrooms, leafy greens, scallion rings and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.5, "reason": "Yellow strands have smooth bends and readable overlaps, but sparse flat loops and inconsistent thickness convey little noodle volume or immersion." }, "toppings": { "score": 0.61, "reason": "Veined green leaves, scored mushroom caps, scallion rings and shaded tofu cubes provide recognizable variety. Shapes remain simplified and smooth, with little organic texture." }, "bowl": { "score": 0.76, "reason": "Nested elliptical rim edges and a shaded curved body give coherent thickness and depth. Stray lower-side highlight strokes and a slightly floating contact shadow weaken the finish." }, "broth": { "score": 0.49, "reason": "Warm orange-brown shading suggests soup, but the surface is largely a smooth solid ellipse with minimal liquid sheen or contact detail; ingredients appear laid over it." }, "steam": { "score": 0.7, "reason": "Normal frames show changing positions and visibility of translucent wisps above the soup without obscuring food. The narrow, similarly sinusoidal strokes look mechanical and show limited spreading." }, "composition": { "score": 0.85, "reason": "A single centered dish, restrained warm lighting, dark textured surroundings and ample negative space keep the food clearly focal across desktop and mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu cubes, mushrooms, leafy greens, scallion rings and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5766666666666667, "bowl": 0.5333333333333333, "broth": 0.6, "steam": 0.6, "composition": 0.6866666666666666 }, "reward": 0.5695, "sampleStdDev": 0.01814524731162405, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are thin looping strands that read as noodles with some overlap. They are flat, sparse and scribble-like, and barely sit in the broth." }, "toppings": { "score": 0.6, "reason": "Spinach leaves with veins, shiitake with cross-cut marks, and tofu cubes are recognizable and varied. Shading is flat and stylized, the green rings look odd, and there is no nori." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a curved body and a shadow. The translucent glassy body looks odd, with faint stray line artifacts at the base and a floating, weakly grounded look." }, "broth": { "score": 0.6, "reason": "The orange-red broth has a radial gradient and a rim highlight that give depth. It is fairly flat and the toppings sit on it rather than in it." }, "steam": { "score": 0.6, "reason": "Thin wavy wisps rise, shift and fade across frames, so the motion is supported. They are thin line squiggles, rather mechanical and similar to each other." }, "composition": { "score": 0.7, "reason": "A centered bowl on a dark warm background keeps the food as the focus, with no clutter and a pleasant palette. The bowl is small in frame and the ground shadow looks slightly detached." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, spinach and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin, flat looping ribbons with repeated similar loops. They are clustered in a small area and read as noodles, but they look like rigid strokes with little volume and sit on the surface rather than in the broth." }, "toppings": { "score": 0.55, "reason": "Spinach leaves with veins, shiitake caps with scoring, and tofu cubes are all recognizable and varied. The forms are clean vector icons with little texture, the cubes are block-like, and the green rings look generic. There is no nori." }, "bowl": { "score": 0.5, "reason": "The bowl has a coherent elliptical rim and a rounded body with highlights. The body looks translucent and glassy rather than ceramic, and the stray foot lines at the sides look odd. The bowl floats above a detached shadow ellipse, so the base contact is weak." }, "broth": { "score": 0.6, "reason": "The orange-red gradient broth has a rim shadow and a pleasing glow. It looks fairly flat and like a disc, and the ingredients sit on top with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Thin wavy translucent wisps rise and change position across frames. They are thin line-like strokes that look somewhat mechanical and similar to each other, but they do not hide the food." }, "composition": { "score": 0.68, "reason": "A centered bowl on a dark warm background gives a clear focal point with no clutter and pleasing warm colors. The bowl is small in the frame with a lot of empty space, and the shadow looks detached." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, spinach and noodles with a plant-looking broth. No animal ingredients are visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin, smooth, curvy strands with some overlap, but they are small, sparse and flat-looking. They read as ribbons lying on the surface rather than a voluminous tangle immersed in broth." }, "toppings": { "score": 0.58, "reason": "Spinach leaves with veins, shiitake caps with scoring and tofu cubes are all recognizable and vegan. The shapes are simple and vector-flat, the tofu cubes are plain and the green rings are odd. There is no nori, and variety and texture are limited." }, "bowl": { "score": 0.5, "reason": "The rim ellipse and body are coherent and have highlights. The body looks translucent and glassy, which does not read as ceramic, and stray lines at the base look like odd feet. The floating shadow is detached and the rim-to-body join looks awkward." }, "broth": { "score": 0.6, "reason": "The warm orange-red gradient broth has a rim shadow and a plausible depth glow and sheen. It is fairly flat and lacks ripples or oil, and the toppings sit on top more than in it." }, "steam": { "score": 0.6, "reason": "Thin wavy translucent wisps rise from the bowl and change position and shape across frames, which shows motion. They are thin line-like curves, somewhat uniform and mechanical, and they do not spread much." }, "composition": { "score": 0.68, "reason": "The bowl is centered on a dark, warm vignette and is the clear focal point, with cohesive colors and no clutter. The bowl is smallish in the frame and the food is somewhat sparse." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, spinach, noodles and a green garnish are visible. There is no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5833333333333334, "toppings": 0.7, "bowl": 0.7666666666666667, "broth": 0.6666666666666666, "steam": 0.7333333333333333, "composition": 0.8333333333333334 }, "reward": 0.7050000000000001, "sampleStdDev": 0.03750000000000003, "judgments": [ { "noodles": { "score": 0.6, "reason": "Cream-colored strands with gentle bends and a light core highlight are continuous and coherent, but they read as flat ribbons lying on the broth surface with little volume or submersion, and the looping shapes repeat." }, "toppings": { "score": 0.65, "reason": "Recognizable variety: organic dark-green leaves, scallion rings, two tofu cubes and brown mushroom slices. Defects: tofu is very rigid isometric geometry, and the mushroom caps carry odd crosshatch marks rather than believable gills; texture is otherwise minimal flat shading." }, "bowl": { "score": 0.7, "reason": "Coherent metallic-gray bowl with cream rim, consistent rim ellipse, shaded body with vertical highlights, and a grounded contact shadow. Minor flaws: thin stray curved stroke artifacts around the rim and lower sides, and fairly simple shading." }, "broth": { "score": 0.6, "reason": "Glossy orange-red surface with radial gradient, darker oil blotches and an edge highlight reads as liquid, but it is shallow and disc-like; noodles and toppings sit atop the surface with weak immersion and occlusion cues." }, "steam": { "score": 0.7, "reason": "Translucent thin S-curved wisps change shape and position across the desktop frames (desktop-1 through pointer-left), rising and fading without hiding food. They are more hairline squiggles than voluminous plumes, slightly mechanical in style." }, "composition": { "score": 0.8, "reason": "Centered, readable focal dish in a clean dark scene with a warm subtle glow and vignette; no text, UI or clutter, harmonious color palette. Presentation is restrained and coherent, though the dish fills only a modest portion of the frame." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, tofu cubes, mushrooms, leafy greens, scallions and a plant-looking broth; no meat, fish, egg or dairy forms appear." } }, { "noodles": { "score": 0.55, "reason": "Pale yellow strands curve across the broth with a few overlaps and rounded tapered ends, but they read as flat, uniform-width ribbons/squiggles lying on the surface rather than volumetric strands with believable depth." }, "toppings": { "score": 0.7, "reason": "Recognizable tofu cubes with soft shading, cross-hatched mushroom halves, leafy greens with veins, and scallion rings; organic-enough silhouettes and good variety, though shading is somewhat flat/clip-art and the cubes are a bit blocky." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a clean inner lip, ceramic-like gradient body with rim highlight, consistent perspective between rim and walls, and a grounding shadow ellipse; only minor shortcoming is the rather shallow, wide proportions." }, "broth": { "score": 0.7, "reason": "Warm orange-red liquid surface with radial depth, restrained sheen arc and darker edge meniscus; noodles and toppings appear to sit in/on it with small contact cues, though the surface is slightly flat and uniform." }, "steam": { "score": 0.7, "reason": "Thin translucent wisps clearly change shape and position across the desktop frames (rising, drifting, fading) and hold static in the reduced-motion frames; they never hide the food, but the repeated S-curve line style is a bit mechanical." }, "composition": { "score": 0.85, "reason": "Centered, readable focal dish on a dark warm atmosphere with soft glow and grounding shadow; balanced framing, appetizing palette, no clutter or UI, and the mobile crop scales cleanly." }, "vegan": { "score": 1, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, scallion, noodles and plant-toned broth; no meat, fish, egg or dairy-like toppings appear." } }, { "noodles": { "score": 0.6, "reason": "Noodles read as pale wavy strands curving around toppings with coherent bends, but they are sparse flat ribbons with rounded ends and little volume or overlap depth." }, "toppings": { "score": 0.75, "reason": "Readable variety: shaded tofu cubes, leafy greens with stems, scored mushroom caps and scallion rings; the mushrooms' cross-hatch texture looks slightly patterned but still reads as shiitake." }, "bowl": { "score": 0.8, "reason": "Coherent cream rim, smooth curved body with a soft ceramic sheen and consistent elliptical perspective; the foot is hidden by the shadow ellipse, which is acceptable stylization." }, "broth": { "score": 0.7, "reason": "Appetizing orange-red gradient surface with a subtle sheen streak; ingredients rest on the surface believably, though occlusion shadows under toppings are minimal." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise, bend and fade upward; their shapes shift noticeably between the time-ordered desktop and pointer frames, confirming motion, and they never obscure the food." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark warm-glow background, balanced proportions, restrained palette, no clutter or distracting UI; mobile framing also stays clean and focused." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, leafy greens, scallions and noodles in a plant-style broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.39999999999964, "frameIntervalP95Ms": 116.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "283de87228170997f45e86d28b86f95b151061afe6c6bc9f2776b0e2a75881f3", "gradedAt": "2026-09-30T23:03:27.339193+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5733333333333334, "toppings": 0.59, "bowl": 0.8, "broth": 0.63, "steam": 0.5055555555555555, "composition": 0.8111111111111111 }, "reward": 0.6421111111111112, "runId": "openai/gpt-6-luna/xhigh", "rewardStdDev": 0.0756786578848968, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.62, "toppings": 0.6533333333333333, "bowl": 0.87, "broth": 0.6233333333333333, "steam": 0.4666666666666667, "composition": 0.86 }, "reward": 0.6796666666666666, "sampleStdDev": 0.004645786621588821, "judgments": [ { "noodles": { "score": 0.62, "reason": "Continuous golden strands have readable bends and some topping occlusion, but remain sparse, similarly outlined surface curves with little tangled volume or immersion." }, "toppings": { "score": 0.66, "reason": "Mushrooms have curved silhouettes and fine gill-like lines; tofu, scallion rings and greens provide variety. The greens resemble a rigid geometric fan, and tofu lacks surface texture." }, "bowl": { "score": 0.87, "reason": "A coherent elliptical rim, inset cavity, rounded tapered walls and attached foot establish convincing construction. Warm highlights and a contact shadow support the lacquer-like finish; wall streaks are somewhat mechanical." }, "broth": { "score": 0.64, "reason": "Warm reddish broth has subtle ripples, scattered highlights and partial ingredient occlusion. Its broad smooth gradient feels relatively flat, with limited wet contact around toppings." }, "steam": { "score": 0.48, "reason": "Three translucent plumes connect to the soup and change curvature across normal frames without obscuring food. Their nearly identical parallel wave shapes look mechanical, with little visible spreading or dispersal." }, "composition": { "score": 0.86, "reason": "The single dish is clearly framed against a restrained dark background, with harmonious copper, gold and green colors and no distracting UI. Mobile framing remains clean, though generous empty space and a large bowl body reduce food prominence." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict mushrooms, tofu, leafy greens and scallions alongside noodles; no visibly animal-derived ingredient is present. No material vegan-appearance defect is observed." } }, { "noodles": { "score": 0.62, "reason": "Continuous golden strands have consistent thickness and curved bends, with some topping occlusion. They remain sparse, mostly parallel surface strokes with limited tangled volume or convincing immersion." }, "toppings": { "score": 0.65, "reason": "Brown mushrooms have readable curved gill details, alongside tofu pieces and scattered green rings. The leafy greens form a stiff geometric fan, and tofu lacks porous texture, limiting organic food plausibility." }, "bowl": { "score": 0.87, "reason": "The elliptical rim, thick highlighted lip, curved walls and small integrated foot form coherent depth and perspective. Warm material shading and a contact shadow support the vessel convincingly; no material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm reddish broth has subtle surface streaks and small highlights, with partially occluded toppings. Its broad smooth gradient reads rather flat, and ingredient contact lacks convincing menisci or liquid depth." }, "steam": { "score": 0.46, "reason": "Three translucent plumes emerge from the soup and visibly shift between normal frames without obscuring the dish. Their nearly identical sinusoidal outlines and bright central strokes look mechanical, with little visible spreading or dissolution." }, "composition": { "score": 0.86, "reason": "A single readable bowl anchors the uncluttered dark scene, with harmonious copper, gold and green colors and restrained illumination. Mobile framing preserves the complete dish; the desktop arrangement is slightly bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, tofu-like cubes, scallion rings and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.62, "reason": "Continuous golden strands curve around and behind toppings with readable thickness, but remain sparse, evenly outlined surface strokes with little tangled volume or immersion." }, "toppings": { "score": 0.65, "reason": "Mushrooms have curved gill details, tofu has shaded facets, and scallion rings add variety. The greens form a stiff geometric fan rather than supple organic leaves, and ingredient textures are limited." }, "bowl": { "score": 0.87, "reason": "The elliptical rim, inset cavity, tapering walls and small foot join coherently. Warm shading, fine vertical marks and a contact shadow establish material and support; no material construction defect is observed." }, "broth": { "score": 0.62, "reason": "Warm orange-brown color, small flecks and restrained surface ripples suggest soup. The broad surface is quite flat, with limited contact menisci and depth around the ingredients." }, "steam": { "score": 0.46, "reason": "Three translucent trails originate near the soup and visibly shift across normal frames without obscuring food. Their nearly identical, outlined sinusoidal shapes remain mechanical, with little visible spreading or breakup." }, "composition": { "score": 0.86, "reason": "A single clearly readable bowl anchors the dark scene, with warm ceramic and broth balanced by green toppings. Desktop and mobile views remain uncluttered and fully framed; the mobile layout has somewhat excessive empty space." }, "vegan": { "score": 1, "reason": "Visible noodles, mushrooms, tofu-like cubes, greens and scallions plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.4666666666666667, "bowl": 0.73, "broth": 0.55, "steam": 0.43333333333333335, "composition": 0.7233333333333333 }, "reward": 0.555, "sampleStdDev": 0.026024027359346263, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are thin, smooth golden strands with some bends and overlap, but they are sparse and flat on the surface, with clear outlines that read as drawn lines rather than a supple, voluminous mass." }, "toppings": { "score": 0.45, "reason": "A bok choy fan is recognizable, but the mushrooms are dark leaf-like ovals, the tofu reads as faceted cream pentagons, and the corn/scallion bits are tiny dots. Variety is present but the forms are generic and cut off at the rim." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, curved walls with a lacquer-like gradient, a foot, and a soft shadow. The rim is thick and flat and the contact with the ground is faint, but no broken geometry is visible." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a gradient and faint sheen streaks, but the surface is fairly flat and opaque, and the toppings sit on it without much immersion or ripple." }, "steam": { "score": 0.4, "reason": "Three wavy S-shaped lines sway slightly between frames, but they are thin, nearly identical and mechanical, and they read as ribbons rather than translucent wisps that spread and fade." }, "composition": { "score": 0.7, "reason": "The bowl is centered and is the clear focal point on a dark, restrained background with warm lighting. The top of the soup is cramped, and the scene is otherwise simple and clean." }, "vegan": { "score": 0.85, "reason": "Only noodles, mushroom-like pieces, bok choy, tofu-like cubes and corn/scallion are visible, with no obvious animal products." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin golden strands with some curves and overlap, but they are sparse, flat and partly cut off, so they read as scribbles on the surface more than a supple, voluminous mass." }, "toppings": { "score": 0.5, "reason": "A bok choy fan, mushroom caps, tofu cubes, corn and scallion rings are recognizable. The shapes are small, flat and generic, and the tofu looks like faceted polygons. Variety is fair but the detail is limited." }, "bowl": { "score": 0.75, "reason": "The elliptical rim is coherent, with a cream lip, a copper-lacquer body, vertical gradients and highlights, a foot and a contact shadow. The form is convincing. The rim interior is slightly flat and the base is a little vague." }, "broth": { "score": 0.6, "reason": "The warm orange-brown broth has a gradient and subtle ripples, and it reads as liquid. It looks fairly flat, toppings sit on it with little immersion or occlusion, and the dark inner ring is heavy." }, "steam": { "score": 0.5, "reason": "Three wavy wisps are translucent and move slightly between frames. They are thin, near-identical sine-shaped strands that look mechanical, and they do not spread or fade much, so they read as ribbons rather than steam." }, "composition": { "score": 0.75, "reason": "The bowl is centered and is a clear focal point against a dark background, with warm colors and soft glow. The presentation is clean with no clutter. The bowl sits slightly low, and the large empty space above is only partly filled by the steam." }, "vegan": { "score": 0.85, "reason": "The visible toppings are mushrooms, bok choy, tofu, corn, scallions and noodles. There is no visible meat, egg or fish. The dark mushroom caps are slightly ambiguous." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin golden curved strands with some overlap and bends, but they are sparse, flat and partly cut off by toppings. They read as scribbled lines on the surface rather than a voluminous mass sitting in the broth." }, "toppings": { "score": 0.45, "reason": "The bok choy fan is recognizable, and the mushrooms have gill lines. The tofu cubes are flat faceted shapes, and the corn and scallion dots are tiny generic dots. There is little variety, no nori, and the pieces look pasted on the surface." }, "bowl": { "score": 0.72, "reason": "Coherent elliptical rim with a cream lip, a tapering body, a base and a ground shadow. The copper gradient and vertical streaks suggest glaze. The rim perspective is slightly inconsistent with the body and the foot is faint, but there is no major distortion." }, "broth": { "score": 0.5, "reason": "Warm orange-red gradient with faint sheen marks and a dark inner rim. It is fairly flat and opaque, and the toppings sit on top without visible immersion or ripples." }, "steam": { "score": 0.4, "reason": "Three wavy steam wisps are translucent and drift a little between frames. They look like identical mechanical sine-wave lines, thin and uniform, and they don't spread or fade much." }, "composition": { "score": 0.72, "reason": "Centered bowl on a dark background with warm lighting and a clear focal point. There is no clutter or text. The large empty space around the bowl and the simple staging keep it from scoring higher." }, "vegan": { "score": 0.85, "reason": "Mushrooms, bok choy, tofu-like cubes, corn and scallions with noodles, and no visible meat or egg. The broth appears plant-based, though the mushrooms are somewhat ambiguous." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.65, "bowl": 0.8, "broth": 0.7166666666666667, "steam": 0.6166666666666667, "composition": 0.85 }, "reward": 0.6916666666666667, "sampleStdDev": 0.033757715167548434, "judgments": [ { "noodles": { "score": 0.55, "reason": "Yellow strands with outlines curve and cross believably, but they read as sparse flat squiggles resting on the surface rather than a coherent noodle nest; little volume, submergence, or thickness variation." }, "toppings": { "score": 0.6, "reason": "Good variety: gilled shiitake-like caps, a large stylized leafy green fan, pale tofu cubes, scallion rings and yellow specks. However scallion rings are repetitive mechanical donuts, and tofu/greens are simple flat outlined shapes." }, "bowl": { "score": 0.8, "reason": "Coherent cream rim ellipse with consistent inner ring, smoothly tapered terracotta body, small foot ring, and vertical ceramic sheen streaks; perspective stays consistent across frames with only a minor stray glow blob at the upper-left rim." }, "broth": { "score": 0.7, "reason": "Warm orange-red gradient with sheen swirls and sesame specks reads as appetizing liquid in this flat style, though ingredients sit on top with minimal occlusion/immersion and the surface is fairly flat." }, "steam": { "score": 0.55, "reason": "Translucent wisps do change curvature across the normal frames (morphing between desktop-1/2/3 and pointer frames), and reduced frames show them frozen, confirming animation. But they are three near-identical vertical S-plumes that sway in place rather than rise, spread and fade." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a clean dark atmospheric ground with subtle contact shadow, restrained warm palette, no clutter or UI; only a faint misplaced glow blob behind the left rim and tiny background specks." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, leafy greens, tofu cubes, scallion rings and corn-like specks in plant broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.7, "reason": "Yellow strands are continuous with varied bends and darker under-outlines giving some depth, lying plausibly across the broth; they read clearly as noodles, though they sit somewhat flat and sparse on the surface with a few repetitive hook curls." }, "toppings": { "score": 0.7, "reason": "Good variety: dark shiitake-like caps with radial gill stripes, a bok choy/napa leaf with radiating veins, tofu cubes, scallion rings and corn-like dots, all readable. Shapes are slightly flat/graphic and mushrooms are somewhat ellipse-like, but nothing generic or plastic dominates." }, "bowl": { "score": 0.75, "reason": "Coherent ceramic bowl: thick cream/gold rim band, dark inner ring, smoothly joined curved body with warm gradient and glaze streaks, visible foot with contact shadow. Tilt in pointer frames reads as an intentional interaction, not broken geometry; body streak texture is slightly odd." }, "broth": { "score": 0.7, "reason": "Rich orange-red surface with soft sheen gradient, darker meniscus at the edge and scattered chili/oil specks; ingredients sit with visible contact, though toppings rest a bit pasted-on rather than deeply immersed." }, "steam": { "score": 0.7, "reason": "Three translucent wisps rise above the bowl without hiding food, and their wave shapes visibly shift across desktop-1/2/3, showing animated rising motion. The plumes are quite similar to one another and stylized as uniform squiggles, slightly mechanical." }, "composition": { "score": 0.85, "reason": "Centered focal dish with warm copper/orange palette against a clean dark background, subtle glow and shadow grounding, no text, UI or clutter; presentation is balanced and appetizing." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, leafy greens, tofu, scallions, corn-like bits, noodles and plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.55, "reason": "Yellow strands are continuous with rounded, uniform ends and lie flat on the broth surface like painted squiggles; they read as noodles but lack volume, immersion, and vary little from wire-like curves." }, "toppings": { "score": 0.65, "reason": "Readable variety: gill-striped mushroom halves, pale tofu cubes, scallion rings, corn dots, and a green leek/nori-like fan, all appropriately scaled. Shapes are simple flat vector forms and the green fan is slightly ambiguous, but nothing reads as plastic or featureless." }, "bowl": { "score": 0.85, "reason": "Coherent terracotta bowl with a wide cream rim band, matching interior ellipse, tapering walls, subtle foot, vertical highlight streaks, and a soft contact shadow; perspective is consistent, though the rim band is somewhat heavy." }, "broth": { "score": 0.75, "reason": "Rich orange-red liquid surface with oil swirls, sesame specks, and a glossy sheen; toppings sit within it with light occlusion. It reads slightly flat/2D with minimal depth beyond the surface gradient." }, "steam": { "score": 0.6, "reason": "Three translucent gray wisps clearly undulate/shift phase across the time-ordered frames, so motion is real and they do not hide the food. However, the form is three near-identical parallel squiggles, a mechanical rather than organic plume treatment." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a dark atmospheric backdrop with a warm rim glow; balanced proportions, no text, UI, or clutter, and the mobile crop keeps the dish legible." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu cubes, scallion, greens, corn, noodles, and plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.95, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "17be720d6cb8892fdabb96fe1f1713b6caadb8e04213306b46b34c84b235c8c4", "gradedAt": "2026-09-30T23:01:16.181326+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6155555555555555, "toppings": 0.7611111111111111, "bowl": 0.8211111111111111, "broth": 0.7100000000000001, "steam": 0.69, "composition": 0.8511111111111112 }, "reward": 0.7372222222222222, "runId": "openai/gpt-6-sol/high", "rewardStdDev": 0.06593817249964239, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5466666666666666, "toppings": 0.72, "bowl": 0.8733333333333333, "broth": 0.6466666666666667, "steam": 0.6433333333333333, "composition": 0.8866666666666667 }, "reward": 0.7138333333333333, "sampleStdDev": 0.008098353742170898, "judgments": [ { "noodles": { "score": 0.5, "reason": "Golden strands are readable and overlap beneath toppings, but mostly form parallel flat arcs with blunt ends and little tangled volume or supple variation." }, "toppings": { "score": 0.72, "reason": "Mushrooms have scored caps and readable gills, greens have curved stems and leaf veins, and tofu has shaded faces. Texture is sparse, and the thick, rectangular nori looks more like rigid cards than thin seaweed." }, "bowl": { "score": 0.87, "reason": "The elliptical rim, curved walls, small attached foot and contact shadow form coherent supported geometry. Layered rim highlights and soft wall shading convincingly suggest dark ceramic; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm amber shading, scattered flecks and restrained curved highlights suggest appetizing soup. Immersion is weak, with fully outlined tofu and mushrooms appearing placed atop the surface rather than partially submerged." }, "steam": { "score": 0.65, "reason": "Translucent tapering wisps emerge above the soup without obscuring it, and their contours shift across normal frames. They remain similarly spaced line-like plumes, with limited visible spreading or dissipation." }, "composition": { "score": 0.88, "reason": "A single prominent dish, warm gold-green food colors and restrained dark surroundings create a clear focal hierarchy. Desktop and mobile framing keep the bowl readable without clutter or text; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings read as mushrooms, tofu, leafy greens, scallions and seaweed alongside noodles; no visibly animal-derived ingredients are present." } }, { "noodles": { "score": 0.56, "reason": "Golden strands curve beneath the toppings with readable thickness, but mostly form parallel, flat sweeps with little tangled overlap or convincing noodle volume." }, "toppings": { "score": 0.72, "reason": "Mushroom caps have scoring and readable gills, greens have curved stems and leaf veins, and tofu has shaded faces. Texture remains sparse, and the thick, rectangular nori looks more like rigid cards than thin seaweed." }, "bowl": { "score": 0.87, "reason": "The elliptical rim, deep curved walls, small attached foot and contact shadow form a coherent vessel. Restrained highlights convincingly suggest dark glazed ceramic; no material construction defect is observed." }, "broth": { "score": 0.64, "reason": "Warm amber shading and scattered surface highlights make the soup readable and appetizing, but the fully outlined tofu and mushrooms appear perched on the surface with limited immersion or contact cues." }, "steam": { "score": 0.66, "reason": "Translucent wisps emerge near the soup and change curvature across normal frames without obscuring food. They remain similarly spaced, narrow sinusoidal columns, with limited visible spreading or dissolution." }, "composition": { "score": 0.89, "reason": "The single centered dish is clearly framed on desktop and mobile, with warm food colors contrasting against the dark bowl and restrained background glow. No distracting text, clutter or material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, scallions and seaweed with noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.58, "reason": "Golden strands curve continuously beneath toppings with readable thickness, but mostly form sparse, parallel shallow arcs rather than a supple, overlapping noodle volume." }, "toppings": { "score": 0.72, "reason": "Mushroom caps have scoring and readable gills, bok choy has curved stems and leaf veins, and tofu has shaded faces. Texture remains schematic, and the nori resembles thick, regular cards." }, "bowl": { "score": 0.88, "reason": "Nested elliptical rim, curved walls, attached foot and contact shadow establish coherent depth and support. Restrained warm highlights convincingly describe dark ceramic; no material construction defect is observed." }, "broth": { "score": 0.65, "reason": "Warm amber shading, small flecks and curved highlights suggest appetizing soup, but ingredient bases lack clear immersion or menisci, leaving the surface somewhat flat and toppings perched above it." }, "steam": { "score": 0.62, "reason": "Translucent wisps extend from the soup and fade without obscuring food; their bends visibly change across normal frames. The persistent, evenly spaced line-like plumes show limited spreading or dissipating variation." }, "composition": { "score": 0.89, "reason": "A single readable dish dominates a restrained dark scene, with warm broth contrasting green vegetables and ample room for steam. Desktop and mobile framing remain uncluttered and balanced; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings read as mushrooms, tofu, leafy greens, scallions and seaweed, with noodles and plant-compatible broth; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.7133333333333333, "bowl": 0.74, "broth": 0.65, "steam": 0.6433333333333333, "composition": 0.7833333333333333 }, "reward": 0.6861666666666666, "sampleStdDev": 0.014545904349105751, "judgments": [ { "noodles": { "score": 0.6, "reason": "Golden noodle strands swirl under the surface with highlights and read as noodles. They are thin and fairly flat, with repetitive parallel curves, and they are partly hidden by the toppings." }, "toppings": { "score": 0.72, "reason": "The shiitake halves show gill lines and crosshatch caps, the tofu cubes have edge highlights and pores, the bok choy has layered leaves, and the scallion rings are readable. There is good variety. The nori looks like rigid gridded panels, and the tofu is a bit blocky." }, "bowl": { "score": 0.72, "reason": "The dark ceramic bowl has a coherent elliptical rim, a curved wall with highlights, a foot and a ground shadow, and it reads as a solid object. The rim has a slight notch at the sides and the body is somewhat flat." }, "broth": { "score": 0.62, "reason": "The warm amber broth has a gradient and some sheen lines, and the toppings sit in it plausibly. It looks a bit flat and opaque, with little depth or reflection." }, "steam": { "score": 0.6, "reason": "Thin translucent wavy wisps rise and change shape across the frames without covering the food. They are narrow, similar to each other and somewhat mechanical in look." }, "composition": { "score": 0.75, "reason": "The bowl is centered and is the clear focal point against a dark, moody background with a warm glow. The palette is cohesive, and there is no clutter or text. The lower space is a bit empty." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, bok choy, nori, scallions and noodles are visible, with no animal ingredients." } }, { "noodles": { "score": 0.6, "reason": "Golden noodle strands with highlights flow in curved bundles under the toppings and read as noodles. They are thin and somewhat flat and sketchy, with limited volume and overlap, so they lack depth." }, "toppings": { "score": 0.72, "reason": "Shiitake halves with gill and cross-hatch detail, tofu cubes with edge highlights, a bok choy with layered leaves and pale stalks, scallion rings and nori sheets. All are recognizable and varied. The tofu is a bit blocky and the nori is stiff and rectangular." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, curved walls, a foot ring, a cast shadow and subtle highlights. The dark ceramic reads well. The rim has small notch artifacts at the sides and the body is slightly flat in material." }, "broth": { "score": 0.68, "reason": "Warm amber broth with a darker gradient at the edge, floating sesame specks and ripple lines. Toppings sit on it believably. It is fairly flat and slightly opaque, with little reflective depth." }, "steam": { "score": 0.68, "reason": "Thin translucent wavy wisps rise, shift position across frames and fade at the top. They are subtle and do not hide the food. They look somewhat uniform and line-like, and some start above the rim rather than from the soup." }, "composition": { "score": 0.8, "reason": "Centered bowl with a warm glow against a dark vignette, a clear focal point and a balanced palette. No distracting clutter. The bottom shadow is a little heavy." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, bok choy, scallions, nori and noodles are visible. No egg, meat or fish." } }, { "noodles": { "score": 0.6, "reason": "Noodles are pale golden strands with curved highlights and some overlap, but they read as thin flat stripes submerged in the broth. There is little volume and few individual bends, so the mass feels somewhat flat and repetitive." }, "toppings": { "score": 0.7, "reason": "Shiitake halves with gill detail and cross-hatched caps, tofu cubes with edge highlights, bok choy with pale stems and veined leaves, scallion rings and nori sheets are all recognizable and varied. The tofu cubes are a bit rigid and uniform, and the nori reads as stiff panels rather than thin irregular seaweed." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, curved walls, small foot and cast shadow, with highlights on the dark ceramic. The rim has a small notch at the sides, and the body is slightly plain, but the perspective is consistent and it reads as a solid bowl." }, "broth": { "score": 0.65, "reason": "Warm amber broth with a gradient and subtle sheen lines, and the toppings sit in it with a believable feel. The surface is fairly flat and stylized, with little depth or occlusion at the contact points." }, "steam": { "score": 0.65, "reason": "Thin translucent wavy wisps rise from the bowl and shift between frames, fading at the top. They are fairly uniform and thin, so they look somewhat mechanical, but they do not hide the food." }, "composition": { "score": 0.8, "reason": "The bowl is centered and is the clear focal point against a dark vignette, with a warm glow and a cohesive palette. There is no text or clutter, and the food stays the focus." }, "vegan": { "score": 0.97, "reason": "Only tofu, mushrooms, bok choy, scallions, nori and noodles are visible. There is no egg, meat or fish." } } ], "diagnostics": { "vegan": 0.97 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7, "toppings": 0.85, "bowl": 0.85, "broth": 0.8333333333333334, "steam": 0.7833333333333333, "composition": 0.8833333333333333 }, "reward": 0.8116666666666666, "sampleStdDev": 0.01376892636821524, "judgments": [ { "noodles": { "score": 0.7, "reason": "Golden wavy noodle strands with darker outlines are grouped in believable curved bundles across the broth, but they read as flat ribbons with repeated similar sweeps rather than supple strands with volume." }, "toppings": { "score": 0.85, "reason": "Well-drawn shiitake caps with scored tops and gill detail, glossy tofu cubes, bok choy with leaf veining, scallion rings, and patterned nori sheets behind the rim; good variety, scale and readability, with tofu slightly box-like." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim matches the tapering body, footed base with contact shadow, left-side highlight; small lug-like tabs at the rim edges are slightly odd but consistent and not broken." }, "broth": { "score": 0.8, "reason": "Warm miso-toned liquid surface with lighter swirls, oil specks and restrained sheen; toppings sit in the broth with contact occlusion rather than floating, though the surface texture is somewhat simplified." }, "steam": { "score": 0.8, "reason": "Translucent thin wisps rise from the broth and visibly change shape and position across desktop-1/2/3, fading near the top without hiding food; wisps are a bit similar in wiggle form. Reduced frames show appropriately static steam." }, "composition": { "score": 0.9, "reason": "Bowl is centered and clearly the focal dish against a dark atmospheric backdrop with a warm glow and soft floor shadow; no text, UI or clutter, and the mobile framing stays balanced." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, bok choy, scallion and nori in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.7, "reason": "Cream wavy strands with rounded ends read as noodles and coil believably around the mushrooms, but they are sparse, mostly flat ribbon strokes with little layered volume, and large areas of the broth show only swirl lines rather than a coherent noodle mass." }, "toppings": { "score": 0.85, "reason": "Shiitake halves show scored caps and gill texture, tofu cubes have beveled shading, bok choy leaves read as leafy greens, scallion rings and nori sheets are recognizable; minor deduction for slightly box-like tofu and the tiled-looking nori grid." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with visible wall thickness, tapering dark ceramic body, pedestal foot and grounded cast shadow with warm rim reflections; the small lug tabs at the rim's left/right edges are slightly odd but read as stylized handles, not broken geometry." }, "broth": { "score": 0.85, "reason": "Rich orange-brown miso surface with restrained oil swirls, sesame specks and clean contact/occlusion where mushrooms, tofu and greens sit in the liquid; appetizing and readable without flat target discs." }, "steam": { "score": 0.85, "reason": "Translucent irregular wisps rise from the soup and fade without hiding food; comparing desktop-1/2/3 and pointer frames the plume shapes and positions change, showing continuous ambient motion rather than static smoke." }, "composition": { "score": 0.9, "reason": "Centered bowl as clear focal point, warm broth glowing against a dark atmospheric vignette, balanced proportions, soft ground shadow, no text, UI or competing clutter." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu cubes, bok choy, scallions, nori, noodles and plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Yellow wavy strands with highlight streaks read as noodles lying in broth, with consistent thickness and layered overlap; however they render as fairly flat painted ribbons on the surface without much strand volume or partial submersion." }, "toppings": { "score": 0.85, "reason": "Well-drawn shiitake caps with cross scores and gill lines, speckled tofu cubes, layered bok choy, scallion rings, and grid-textured nori sheets tucked at the back rim; only minor rigidity in the tofu cubes and nori rectangles." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with believable thickness, smooth dark ceramic body with highlights, footed base with contact shadow, and stable perspective; small symmetric side tabs read as handles rather than broken joins." }, "broth": { "score": 0.85, "reason": "Warm miso-toned surface with restrained sheen, speckles, and depth; toppings sit on the liquid with contact rather than floating, and the glow reads as surface reflection." }, "steam": { "score": 0.7, "reason": "Translucent wavy wisps rise from the bowl without obscuring food, and their curvature shifts subtly across the timed desktop frames, indicating animation; the plumes are somewhat thin, uniform squiggles rather than volumetric wisps." }, "composition": { "score": 0.85, "reason": "Single focal bowl centered on a dark atmospheric background with warm vignette, coherent lighting, no text or clutter, and readable presentation at both desktop and mobile sizes; background glow does not overshadow the dish." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu, bok choy, scallions, and nori in plant broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.99, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.29999999999927, "frameIntervalP95Ms": 100, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "4b41cd273979aacadf75d12eb968177271f403ab56c1ff4569ba448f4a3d0cc6", "gradedAt": "2026-09-30T23:02:02.609112+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.62, "toppings": 0.6588888888888889, "bowl": 0.7999999999999999, "broth": 0.6722222222222222, "steam": 0.7, "composition": 0.8277777777777778 }, "reward": 0.7037222222222222, "runId": "openai/gpt-6-sol/low", "rewardStdDev": 0.08523893433124287, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.49333333333333335, "toppings": 0.5933333333333333, "bowl": 0.85, "broth": 0.6, "steam": 0.6633333333333333, "composition": 0.8666666666666667 }, "reward": 0.6631666666666667, "sampleStdDev": 0.0063311399710741885, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends and consistent thickness, but repeated parallel waves form a flat pattern with little overlap, volume, or submersion." }, "toppings": { "score": 0.6, "reason": "Veined greens, scallion rings, and a shaded mushroom cap are recognizable. The pale pieces are overly smooth and generic, while the nori resembles a rigid rectangular panel rather than irregular thin seaweed." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, visible inner wall, curved dark ceramic body, and contact shadow create coherent depth and support. Restrained highlights and curved exterior bands reinforce the form; small abrupt rim joins are a minor defect." }, "broth": { "score": 0.62, "reason": "Warm amber shading and subtle surface highlights suggest appetizing broth, but the broad smooth surface lacks liquid texture and ingredients appear mostly laid on top rather than immersed." }, "steam": { "score": 0.66, "reason": "Normal frames show translucent wisps changing height and visibility above the soup without obscuring it. Their similar narrow S-curves and sharply clipped upper ends make the motion look mechanical rather than naturally spreading." }, "composition": { "score": 0.87, "reason": "A single centered dish remains clearly readable against a restrained dark background, with warm food colors contrasting the green ceramic. Desktop and mobile framing are balanced and free of distracting text or clutter; no material compositional defect is observed." }, "vegan": { "score": 0.93, "reason": "Visible greens, mushroom, seaweed, scallions, and noodles plausibly form vegan ramen. The pale pieces are not clearly identifiable but do not show unmistakable animal-derived features." } }, { "noodles": { "score": 0.48, "reason": "Golden strands have smooth continuous bends, but mostly form evenly spaced parallel waves with exposed blunt ends and little tangled volume or immersion." }, "toppings": { "score": 0.58, "reason": "Scored mushroom cap, veined greens and scallion rings are readable. However, the nori resembles a rigid striped card, and the pale chunks are overly smooth and rounded." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim, visible cavity, curved dark ceramic walls and restrained decorative bands establish convincing depth and support. Minor abrupt joins at the rim sides interrupt the otherwise polished construction." }, "broth": { "score": 0.59, "reason": "Warm amber shading and subtle highlights suggest soup, but broad nested surface bands and limited contact shadows make the ingredients appear laid over a relatively flat surface." }, "steam": { "score": 0.66, "reason": "Normal frames show changing heights and opacity in translucent wisps emerging behind the soup without obscuring it. The plumes retain very similar narrow wave shapes and conspicuously abrupt upper cutoffs." }, "composition": { "score": 0.87, "reason": "A single clearly readable dish anchors the dark scene, with harmonious amber and green colors, soft grounding shadow and ample negative space across desktop and mobile. No material compositional defect or distracting clutter is observed." }, "vegan": { "score": 1, "reason": "Visible noodles, mushroom, leafy greens, scallions, seaweed and pale plant-like chunks plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends, but mostly form parallel flat waves with exposed ends and little tangled overlap, volume, or submersion." }, "toppings": { "score": 0.6, "reason": "Veined leaves, scallion rings, and a shaded mushroom cap are recognizable. Pale pieces are overly smooth and rounded, while the nori resembles a rigid striped card rather than irregular seaweed." }, "bowl": { "score": 0.85, "reason": "The elliptical rim, inner cavity, rounded walls, subtle exterior bands, and grounding shadow form coherent depth and support. Rim thickness is readable; minor abrupt side joins and subdued material highlights limit polish." }, "broth": { "score": 0.59, "reason": "Warm amber shading and restrained highlights suggest soup, but broad nested surface bands look graphic and ingredients have little liquid contact or partial immersion." }, "steam": { "score": 0.67, "reason": "Normal frames show changes in plume height and opacity, with translucent wisps rising above the soup without obscuring it. Repeated narrow S-shaped columns and abruptly clipped upper ends look mechanical." }, "composition": { "score": 0.86, "reason": "A single clearly framed bowl remains the focus against a restrained dark background, with harmonious amber and green colors and ample breathing room on desktop and mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, scallions, seaweed, mushroom, and pale plant-compatible pieces plausibly form vegan ramen; no clearly animal-derived topping is visible." } } ], "diagnostics": { "vegan": 0.9766666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.6, "bowl": 0.7, "broth": 0.6333333333333333, "steam": 0.6533333333333333, "composition": 0.75 }, "reward": 0.6463333333333333, "sampleStdDev": 0.012220201853215585, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles are smooth, wavy golden strands with decent overlap and taper, but they look like thin parallel flat curves with limited volume and little sense of being submerged." }, "toppings": { "score": 0.6, "reason": "Nori with vein lines, a shiitake cap with gills, tofu cubes, a bok choy leaf, scallion rings and a mushroom stem are all readable. They are simple flat vector shapes, and the tofu has a slightly plastic look." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a dark teal body with subtle highlight lines, a base and a contact shadow. Perspective is consistent. The body is somewhat flat and dark, and the rim has a slight seam." }, "broth": { "score": 0.66, "reason": "The warm brown broth has a radial glow and subtle ripple highlights that give a sense of depth. Ingredients sit in it fairly well, but the surface is smooth and lacks oil sheen or texture." }, "steam": { "score": 0.68, "reason": "Translucent wavy wisps rise and change positions across frames, fading at the top without hiding the food. They are a bit uniform and thin, like repeated squiggles." }, "composition": { "score": 0.75, "reason": "The bowl is a centered, clear focal point on a dark vignette with warm glow and no clutter. Colors are harmonious. The bowl is a bit small in frame, leaving a lot of empty space." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushroom, nori, greens, scallions and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.6, "reason": "Wavy golden strands with varied bends and overlap read as noodles and sit in the broth. They are thin, fairly parallel and flat, and they look a bit like stylized ribbons with little volume." }, "toppings": { "score": 0.6, "reason": "Nori sheet with veins, shiitake cap with gills, tofu cubes, bok choy leaves, scallion rings and a mushroom stem are all recognizable and vegan. The tofu cubes are smooth and rounded, somewhat plastic, and the items are sparse and flat." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, dark teal body with subtle highlights and a grounded shadow. The rim has a slight seam and the wall reads a bit flat, but no major distortion." }, "broth": { "score": 0.62, "reason": "Warm brown broth with a soft glow and faint ripple rings looks appetizing. It is somewhat flat with concentric rings, and ingredients sit on it with limited immersion cues." }, "steam": { "score": 0.68, "reason": "Translucent wavy wisps change position and fade across frames, and they do not hide the food. The wisps are similar thin sine-like curves, so they look slightly mechanical." }, "composition": { "score": 0.75, "reason": "Centered bowl with a dark vignette, warm glow and a harmonious palette keeps the focus on the food. There is no clutter, though the bowl is small in the frame and the empty space is large." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushroom, nori, greens, scallion and noodles are visible, with no egg or meat." } }, { "noodles": { "score": 0.6, "reason": "Wavy golden strands with varied bends and some overlap read as noodles. They are thin, fairly parallel and flat-looking, with little volume and only partial immersion in the broth." }, "toppings": { "score": 0.6, "reason": "A nori sheet with vein lines, a shiitake cap with gills, tofu cubes, scallion rings and a bok choy leaf are all recognizable and varied. The tofu cubes look slightly glossy and plastic, the nori is a rigid rectangle, and the leaf changes shape between frames." }, "bowl": { "score": 0.68, "reason": "The bowl has a clear rim, a dark teal body, a base and a contact shadow, plus subtle highlight lines. The rim and body perspective are consistent. The bowl is somewhat flat and lacks a visible foot, but there are no distortions." }, "broth": { "score": 0.62, "reason": "The warm brown broth has a radial glow and a subtle sheen ring that give it some depth. The toppings sit on the surface with little immersion or occlusion, and the surface is fairly smooth." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise and change position across the frames, and they do not hide the food. The wisps are similar, thin, line-like plumes, so they look somewhat mechanical." }, "composition": { "score": 0.75, "reason": "A centered bowl on a dark vignette with warm colors keeps the food as the focus. There is no clutter or text. The bowl is small in the frame, with lots of empty space." }, "vegan": { "score": 0.95, "reason": "Tofu, shiitake, nori, bok choy, scallion and noodles are all plant-based. No animal ingredients are visible." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7666666666666667, "toppings": 0.7833333333333333, "bowl": 0.85, "broth": 0.7833333333333333, "steam": 0.7833333333333333, "composition": 0.8666666666666667 }, "reward": 0.8016666666666667, "sampleStdDev": 0.020966242709015273, "judgments": [ { "noodles": { "score": 0.8, "reason": "Golden strands are continuous and supple with plausible wavy bends, passing behind the tofu cubes and under the mushroom with believable layering; ends tuck rather than dangle. Slight weakness: the parallel S-curves are quite uniform and repetitive, giving a stylized flat-band feel rather than varied strand clusters." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: a striped nori sheet standing upright, leafy greens with veins, scallion rings, carrot slivers, a ribbed brown mushroom cap, and two pale tofu cubes. Defects: the tofu cubes are somewhat glossy and marshmallow-like with a single dot, slightly toy-like; the mushroom is a smooth dome with radial ribs but minimal texture." }, "bowl": { "score": 0.85, "reason": "Coherent deep teal bowl with a thick light glazed rim, consistent rim/body ellipse perspective, horizontal banding suggesting curvature, and a soft grounded contact shadow; it stays intact when the parallax tilts in pointer frames. Minor shortcoming: the body reads slightly shallow/wide and material texture is minimal." }, "broth": { "score": 0.8, "reason": "Warm brown liquid surface with a restrained sheen, subtle highlight swirl and concentric tone rings giving depth; toppings and noodles sit on the surface with sensible occlusion rather than floating pasted above. The oil swirl stroke is a bit clean/graphic, and broth texture is otherwise simple." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 the translucent gray wisps visibly change shape, position and fade while never obscuring the dish, so steam is genuinely animated, not static. The wisps are drawn as thin squiggle ribbons with soft edges rather than volumetric vapor, which caps the score slightly." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a dark atmospheric scene with a warm halo glow, steam leading the eye upward, balanced margins on desktop and mobile, no text, UI or clutter, and gentle pointer parallax that keeps the dish legible." }, "vegan": { "score": 0.9, "reason": "Visible toppings are all plausibly plant-based (nori, tofu cubes, mushroom, leafy greens, scallions, carrot, noodles); no egg, meat, fish or dairy imagery. The small pale cylinder is ambiguous but reads as garlic/tofu puffed piece rather than narutomaki (no pink spiral), hence not full credit." } }, { "noodles": { "score": 0.7, "reason": "Continuous wavy cream-colored strands with consistent thickness and some overlap read clearly as noodles in a flat-illustration style; however, they sit as parallel surface squiggles with little volume or immersion, and the S-curve repetition feels slightly mechanical." }, "toppings": { "score": 0.75, "reason": "Good variety: striped nori sheets, a shiitake cap with radial gill lines, tofu cubes, scallion rings, carrot slices and leafy greens are all recognizable. Weaknesses: the tofu cubes look glossy and marshmallow-like with rounded plastic highlights, the nori is two rigid slabs, and forms are very simplified." }, "bowl": { "score": 0.85, "reason": "Coherent dark-teal ceramic bowl with a clean elliptical rim band, tapered body, subtle glaze stripes and a soft contact shadow; rim and body perspective agree, and the pointer frames show a plausible gentle tilt. Only minor simplification of the join between rim and wall." }, "broth": { "score": 0.75, "reason": "Warm brown glossy surface with restrained sheen, a soft radial glow and believable immersion\u2014nori leans into the liquid, noodles and toppings rest on it. Slightly flat: depth comes mostly from a uniform glow plus a light ring near the rim rather than texture." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps visibly morph, shift and drift across the time-ordered desktop frames (shapes at 1010ms, 2765ms, 4512ms all differ), rising above the bowl without obscuring food; reduced-motion frames are correctly static. A couple of wisps have hard straight rectangular edges that read slightly mechanical." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with balanced proportions, attractive warm-on-dark palette, subtle background glow and vignette, no text, UI or clutter; mobile crop also keeps the dish well framed. The glow supports but does not substitute for the dish." }, "vegan": { "score": 1.0, "reason": "All visible toppings are plant-derived: nori, tofu cubes, mushroom, scallions, carrots, leafy greens and noodles in a plant broth; no eggs, meat, fish or dairy appear anywhere." } }, { "noodles": { "score": 0.8, "reason": "Supple, continuous yellow strands with consistent thickness and gentle bends that dip behind the mushroom and tofu for believable layering in the flat-illustration style. Minor flaw: several strand ends show rounded caps terminating mid-broth, slightly artificial." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: textured nori sheets, speckled tofu cubes, a domed shiitake-style mushroom, bok choy leaves with veins, scallion rings, carrot slices and a menma stick. Forms are stylized but organic; tofu cubes read slightly marshmallow-like." }, "bowl": { "score": 0.85, "reason": "Coherent dark-teal ceramic bowl with a pale glazed rim, consistent elliptical rim/body perspective, subtle wall shading and a supporting ground shadow. No broken joins or perspective conflicts observed." }, "broth": { "score": 0.8, "reason": "Rich brown liquid with a warm central glow, restrained swirl highlights and small seasoning specks; ingredients sit within the surface with plausible occlusion rather than floating above it." }, "steam": { "score": 0.75, "reason": "Translucent irregular wisps that clearly change shape and position across the time-ordered normal frames, rising and fading without hiding the food; the reduced-motion frames hold them correctly static. Wisps are somewhat uniform S-curves and start slightly detached above the rim." }, "composition": { "score": 0.9, "reason": "Centered focal dish on a dark atmospheric backdrop with a soft warm halo, balanced proportions, no text or clutter, and subtle pointer-driven parallax that does not disrupt the read." }, "vegan": { "score": 1.0, "reason": "Visible toppings are nori, tofu, mushroom, leafy greens, scallion, carrot and bamboo-style menma in plant broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9644444444444444, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.39999999999964, "frameIntervalP95Ms": 100, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "c10182568ff3fdc06ff5bb2101ac23d3748ab07bda1573c0649c969c09f6ef23", "gradedAt": "2026-09-30T22:59:37.469145+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6388888888888888, "toppings": 0.7000000000000001, "bowl": 0.8244444444444444, "broth": 0.6677777777777778, "steam": 0.7088888888888889, "composition": 0.8222222222222222 }, "reward": 0.723, "runId": "openai/gpt-6-sol/max", "rewardStdDev": 0.041462164814576534, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6033333333333333, "toppings": 0.67, "bowl": 0.8666666666666667, "broth": 0.6233333333333333, "steam": 0.69, "composition": 0.8633333333333333 }, "reward": 0.7136666666666667, "sampleStdDev": 0.0022546248764114206, "judgments": [ { "noodles": { "score": 0.61, "reason": "Golden strands have consistent thickness, curved ends and readable overlaps, but mostly form shallow, similarly directed strokes with limited supple tangling or submerged volume." }, "toppings": { "score": 0.67, "reason": "Mushrooms have shaded caps and readable gills, tofu has distinct faces and small surface marks, and greens show veins and pale stems. Leaves remain angular and stiff, while nori resembles a straight-edged card." }, "bowl": { "score": 0.87, "reason": "The elliptical rim, continuous rounded walls and integrated foot form coherent depth and perspective. Layered rim highlights, subtle speckling and curved reflections convey dark glazed ceramic; no material construction defect is observed." }, "broth": { "score": 0.62, "reason": "Warm amber shading and scattered garnish make the surface readable and appetizing, but broad graphic streaks provide little liquid sheen, and ingredients lack convincing wet contact or partial immersion." }, "steam": { "score": 0.68, "reason": "Normal frames show changes in wisp height and opacity, with translucent blurred plumes rising above the soup without concealing it. Alternating narrow, clean sinusoidal lines look mechanical, and some plume bases appear disconnected." }, "composition": { "score": 0.86, "reason": "A single focal bowl, restrained dark background and warm soup against cool ceramic create a balanced atmospheric presentation without text or clutter. Mobile framing remains readable, though its bowl sits very close to the side edges." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.59, "reason": "Golden strands have readable thickness, curved ends and some overlap, but mostly form short, nearly parallel strokes with limited supple tangling or submerged volume." }, "toppings": { "score": 0.67, "reason": "Mushroom caps have highlights and readable gills, tofu has shaded faces, and greens show veins and pale stems. Leaves remain stiff and angular, while the nori is an unusually straight rectangular panel." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim joins rounded walls and a centered foot; layered rim highlights and subtle speckling convincingly suggest dark glazed ceramic. No material construction defect is observed, though wall reflections are somewhat schematic." }, "broth": { "score": 0.63, "reason": "Warm amber shading and scattered surface details make the soup readable, but broad flat patches and minimal contact ripples leave ingredients looking placed atop the surface rather than convincingly immersed." }, "steam": { "score": 0.7, "reason": "Normal frames show changing wisp positions and opacity, with translucent blurred plumes rising above the soup without obscuring food. Several narrow strands retain repetitive sinusoidal shapes and occasionally appear disconnected from the surface." }, "composition": { "score": 0.86, "reason": "The single dish is clearly framed against a restrained dark background, with warm soup contrasting effectively against the cool bowl and ample space for steam. No text or clutter competes with the food; mobile framing is slightly tight at the sides." }, "vegan": { "score": 1, "reason": "Visible tofu cubes, mushrooms, leafy greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.61, "reason": "Golden strands have consistent thickness, curved ends and readable overlaps, but mostly form shallow, similarly directed strokes with limited supple tangling or submerged volume." }, "toppings": { "score": 0.67, "reason": "Mushroom caps and gills, shaded tofu cubes, leaf veins and scallion rings provide recognizable variety. Leaves are angular and stiff, while the nori is an overly straight rectangular panel." }, "bowl": { "score": 0.87, "reason": "A coherent elliptical rim, rounded walls, attached foot and layered restrained highlights convincingly establish a dark glazed bowl. Fine speckling adds material texture; ground contact is slightly weak." }, "broth": { "score": 0.62, "reason": "Warm brown gradients and scattered surface accents suggest soup, but the surface is fairly matte and ingredients largely appear laid over it, with little visible immersion or contact detail." }, "steam": { "score": 0.69, "reason": "Normal frames show wisps changing height and opacity without obscuring the food. Soft translucent trails are effective, but alternating narrow lines retain repetitive sinusoidal shapes and some visibly detached lower ends." }, "composition": { "score": 0.87, "reason": "The single dish remains the clear focus against a restrained dark background, with warm soup contrasting the cool bowl and ample space for steam. No distracting text or clutter appears; mobile framing is slightly tight at the sides." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, scallions, seaweed and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6466666666666667, "toppings": 0.6466666666666667, "bowl": 0.7566666666666667, "broth": 0.6133333333333333, "steam": 0.72, "composition": 0.7533333333333333 }, "reward": 0.687, "sampleStdDev": 0.016822603841260737, "judgments": [ { "noodles": { "score": 0.68, "reason": "Noodles are thin, cream-colored strands that curve and overlap plausibly, but they are clustered in one small central patch, and the thin strands look a bit flat and sparse. No broken or maze-like lines." }, "toppings": { "score": 0.66, "reason": "Tofu cubes, shiitake caps, bok choy leaves, nori, scallion rounds and chili flecks are recognizable and varied. The tofu cubes are fairly blocky and uniform, and the mushrooms and leaves are simplified, with a few odd cutoffs at the rim." }, "bowl": { "score": 0.74, "reason": "The dark glossy bowl has a coherent elliptical rim, a curved body, soft highlights and a footed base. It looks slightly floating with a faint shadow, and the broth surface seems a bit tilted relative to the rim." }, "broth": { "score": 0.64, "reason": "The warm brown broth has a gradient, faint ring lines and some depth. The surface is fairly flat, and the toppings look slightly pasted on, with little immersion or ripple." }, "steam": { "score": 0.74, "reason": "Several translucent wavy wisps and softer blurred plumes rise, drift and fade across the frames, and the pattern changes over time. The curly line wisps are somewhat uniform, but they do not hide the food." }, "composition": { "score": 0.76, "reason": "The bowl is centered in a dark, moody scene with a warm glow, and there is no clutter or text. The palette is cohesive and the food is the focus, though the bowl is a bit large and the lower area is empty." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, nori, scallions and noodles in a brown broth. No visible animal products." } }, { "noodles": { "score": 0.66, "reason": "Noodles are supple, tan, overlapping curved strands with believable bends and some volume in the center. They look a bit thin and flat, and the mass is small relative to the bowl, but there are no rigid maze-like loops." }, "toppings": { "score": 0.66, "reason": "Tofu cubes, shiitake caps with gills, bok choy leaves, nori, scallion rings and chili flecks are all readable and varied. The tofu cubes are fairly generic and blocky, the nori reads as a dark rectangle, and the placement is slightly pasted on." }, "bowl": { "score": 0.78, "reason": "The dark glazed bowl has a coherent elliptical rim, a curved body with soft highlights, and a foot ring with a subtle shadow. The rim perspective matches the body. The bowl looks slightly floating and the glaze is flat, with no major distortion." }, "broth": { "score": 0.6, "reason": "The warm brown broth has a gradient and subtle ripple lines, and it reads as liquid. It looks fairly flat, with little sheen or immersion depth, and the ingredients sit on top more than in it." }, "steam": { "score": 0.7, "reason": "Translucent wavy wisps in a mix of thin lines and soft plumes rise, sway and fade, and they change between frames. They are somewhat uniform and sinuous, and some begin detached above the rim." }, "composition": { "score": 0.75, "reason": "The bowl is centered and the focal dish is clear against a dark vignette with a warm glow. The palette is cohesive and there is no clutter. The food is small within the bowl and the scene is somewhat empty." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, bok choy, nori, scallions and noodles. No animal products are visible." } }, { "noodles": { "score": 0.6, "reason": "Thin cream strands curve and overlap in a readable cluster with soft bends. They are fairly flat and sparse, and they sit on the surface rather than forming a voluminous mass. The loops look slightly repetitive." }, "toppings": { "score": 0.62, "reason": "Tofu cubes, shiitake caps, bok choy leaves, nori, scallion rings and chili flecks are all recognizable, and the variety is good. The tofu cubes are rather blocky and uniform, the nori is a rigid rectangle, and the mushrooms look a bit cupcake-like." }, "bowl": { "score": 0.75, "reason": "The dark glossy bowl has a coherent elliptical rim, a curved body with soft highlights and a small foot ring. Perspective is consistent. The bowl appears to float above the dark background with little contact shadow, and the foot is slightly flat." }, "broth": { "score": 0.6, "reason": "The warm brown broth has a gradient and subtle ring lines, and it reads as liquid. Ingredients sit on it without much immersion, ripples or reflections, so it looks somewhat flat." }, "steam": { "score": 0.72, "reason": "Translucent wavy wisps of different widths rise, drift and fade across the frames, and they do not hide the food. The thin line plumes look somewhat mechanical and similar to one another." }, "composition": { "score": 0.75, "reason": "The bowl is centered as the clear focal point against a dark, warm-glow background. Colors are appealing and cohesive, and there is no clutter or text. The bowl looks slightly floaty, and the food occupies a small area." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, scallions and noodles are all plant-based. No egg, meat or fish is visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.7833333333333333, "bowl": 0.85, "broth": 0.7666666666666667, "steam": 0.7166666666666667, "composition": 0.85 }, "reward": 0.7683333333333333, "sampleStdDev": 0.008036375634160854, "judgments": [ { "noodles": { "score": 0.65, "reason": "Noodles read as wavy cream-yellow strands with consistent thickness and soft bends, but they are flat squiggle ribbons lacking bundle volume, and several strands appear broken into separate segments rather than continuous." }, "toppings": { "score": 0.8, "reason": "Well-drawn shiitake caps (one showing gill lines), bok choy leaves with lighter midribs, tofu cubes with small dash texture, a dark nori sheet, scallion rings and chili flakes. Minor shortcoming: tofu cubes are uniform geometric blocks and scales are somewhat repetitive." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with a clean elliptical rim, inner-wall ring, smooth curved body with subtle specular highlights, a visible foot ring and grounded contact shadow; perspective is consistent with only slightly soft stylized highlights." }, "broth": { "score": 0.75, "reason": "Rich brown-orange liquid surface with gradient depth, floating scallions and chili, and darker shadow pools under toppings suggesting immersion; slightly flat and smooth overall with restrained sheen." }, "steam": { "score": 0.7, "reason": "Steam clearly changes between desktop-1/2/3 and pointer frames (wisp positions and shapes shift), rising translucent wisps that never obscure the food; deduction for the repetitive squiggle shapes and a few soft blob plumes that read slightly detached." }, "composition": { "score": 0.85, "reason": "Centered focal dish against a dark atmospheric backdrop with a warm glow and faint backlight halo; no clutter or UI. Mobile view scales coherently though the bowl nearly touches the screen edges." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, bok choy, nori, scallions and chili in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.7, "reason": "Yellow strands curve coherently across the broth with varied bends and overlaps, reading as noodles in the stylized illustration, but they lie flat on the surface with uniform ribbon-like thickness and limited volume or submersion." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: shiitake caps (one with gill detail), speckled tofu cubes, bok choy leaves, a dark nori sheet, scallion rings and chili flakes. Shortcomings: leaves are flat angular shapes, the nori is a plain featureless rectangle, and the mushroom cap crosshatching looks patterned rather than organic." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with a consistent rim ellipse, visible rim thickness, tapering walls, a footed base with contact shadow, and soft glaze highlights; perspective holds together across frames with no detached or broken joins." }, "broth": { "score": 0.75, "reason": "Warm brown liquid surface with restrained sheen, darker edge shading and believable occlusion shadows under the tofu and mushrooms; garnishes float plausibly. Somewhat flat overall, relying on a simple radial gradient for depth." }, "steam": { "score": 0.7, "reason": "Translucent wisps rise, spread and clearly change shape across desktop-1/2/3, so the steam is animated and does not hide the food. Defects: several similar S-curl squiggle strokes read slightly mechanical, and plume shapes repeat." }, "composition": { "score": 0.85, "reason": "Well-balanced centered dish against a dark atmospheric background with a warm glow behind the bowl; steam and lighting are coherent, no text, UI or clutter, and the food stays the clear focal point on both desktop and mobile." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu cubes, leafy greens, nori, scallions and chili in a plant-based-looking broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.65, "reason": "Tan strands weave across the broth with plausible thickness and curves, but they read as flat surface squiggles with limited volume and some repetitive loop shapes rather than supple submerged strands." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: shiitake caps (one showing gills), tofu cubes with speckle detail, bok choy leaves, scallion coins, chili bits and a nori sheet. Shapes are organic; tofu and nori are slightly plain/flat but still convincing in the stylized look." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim with thickness, smooth walls, curved highlights, subtle speckles, a foot with contact shadow beneath. Perspective between rim and body is consistent; no visible joins or distortion." }, "broth": { "score": 0.8, "reason": "Appetizing mahogany broth with tonal depth, floating scallions and chili, and shadow pooling under toppings so they read as seated in the liquid. Nori leans at the back rim plausibly; minor pasted feel on a few pieces." }, "steam": { "score": 0.75, "reason": "Wisps are translucent, irregular and change shape/position across the normal frames (desktop-1/2/3 and pointer frames all differ), rising without hiding the food; reduced frames are sensibly static. A few wisps look like thin rigid S-lines rather than soft vapor." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a restrained dark scene with a warm glow behind, balanced proportions, and no text or clutter; mobile crop keeps the dish readable. Steam occupies the upper space cleanly." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: mushrooms, tofu, bok choy, scallions, chili, nori, noodles and broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.30000000000109, "frameIntervalP95Ms": 183.29999999999927, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "62269faa33d98902be9bbbd7548f6e53098c4e7f25ec16399445003d0ea9fd48", "gradedAt": "2026-09-30T23:01:43.400319+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5611111111111111, "toppings": 0.6466666666666666, "bowl": 0.7622222222222222, "broth": 0.5933333333333334, "steam": 0.6355555555555557, "composition": 0.7377777777777778 }, "reward": 0.6547777777777778, "runId": "openai/gpt-6-sol/medium", "rewardStdDev": 0.06335796304712991, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.51, "toppings": 0.6033333333333333, "bowl": 0.83, "broth": 0.5133333333333333, "steam": 0.6566666666666667, "composition": 0.7533333333333333 }, "reward": 0.644, "sampleStdDev": 0.0021794494717703515, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands have rounded bends and some occlusion, but mostly form parallel, flat waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.6, "reason": "Veined greens, faceted tofu, scored mushroom caps and scattered garnishes provide readable variety. Shapes remain schematic and minimally textured, especially the straight-edged nori card and smooth tofu blocks." }, "bowl": { "score": 0.83, "reason": "A coherent elliptical rim, rounded deep body, attached foot and contact shadow establish convincing construction. Muted teal shading and curved highlights suggest glazed ceramic, though the broad outline highlights are somewhat graphic." }, "broth": { "score": 0.52, "reason": "Warm brown coloration and subtle tonal variation establish soup, but the surface feels matte and flat, with weak liquid contact around toppings and little visible sheen or depth." }, "steam": { "score": 0.66, "reason": "Translucent wisps emerge near the soup and change position and visibility across normal frames without obscuring food. They remain similarly narrow, parallel sinusoidal trails with limited spreading." }, "composition": { "score": 0.76, "reason": "The single dish is cleanly framed against a restrained dark background with no clutter or text. Warm food and cool ceramic are harmonious, but low contrast mutes the focal details and the large upper negative space makes the desktop framing bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy vegetables, mushrooms, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have rounded bends, but mostly form parallel, flat waves with little overlap or convincing submerged volume." }, "toppings": { "score": 0.61, "reason": "Veined greens, shaded tofu cubes and marked mushroom caps provide readable variety. Shapes remain schematic, with smooth blocky tofu and an overly straight, rigid nori sheet." }, "bowl": { "score": 0.83, "reason": "Coherent elliptical rim, curved walls, small grounded foot and contact shadow establish convincing construction. Muted ceramic texture and curved highlights add depth, though the broad outline highlights are somewhat graphic." }, "broth": { "score": 0.53, "reason": "Warm brown coloring and subtle tonal patches suggest soup, but the surface has little liquid sheen or depth; ingredients largely appear arranged on top rather than immersed." }, "steam": { "score": 0.66, "reason": "Translucent wisps emerge behind the soup and change position and visibility across normal frames without obscuring food. Their evenly spaced, similarly narrow sinusoidal shapes feel mechanical rather than irregularly spreading." }, "composition": { "score": 0.76, "reason": "A single uncluttered dish anchors the restrained dark scene, with harmonious green, amber and charcoal colors. Excessive dimness mutes the food, and the desktop layout leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushrooms, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredients are present." } }, { "noodles": { "score": 0.53, "reason": "Continuous golden strands have smooth bends and some topping occlusion, but mostly form parallel surface waves with little tangled volume or convincing immersion." }, "toppings": { "score": 0.6, "reason": "Veined leaves, shaded tofu cubes and scored mushroom caps provide readable variety. Shapes remain schematic and smooth, while the nori resembles a rigid rectangular card." }, "bowl": { "score": 0.83, "reason": "A coherent elliptical rim, curved walls, small aligned foot and contact shadow establish a supported vessel. Subtle surface texture and curved highlights suggest dark ceramic; decorative lines and broad highlights are somewhat flat." }, "broth": { "score": 0.49, "reason": "Warm brown color and sparse flecks identify soup, but broad flat color bands lack liquid sheen and depth. Ingredients largely appear arranged atop the surface rather than partially immersed." }, "steam": { "score": 0.65, "reason": "Normal frames show changes in the visibility and positions of translucent wisps rising above the soup. They fade without obscuring food, but remain a fairly regular set of narrow, similar-height sinusoidal plumes with little spreading." }, "composition": { "score": 0.74, "reason": "The isolated bowl, restrained palette and uncluttered dark setting maintain a clear focal dish across desktop and mobile. Low contrast mutes food detail, and the large empty upper region makes the desktop framing bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushrooms, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.5866666666666667, "bowl": 0.6733333333333333, "broth": 0.56, "steam": 0.5833333333333334, "composition": 0.66 }, "reward": 0.5975, "sampleStdDev": 0.007053367989832952, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are smooth wavy strands with some overlap, but they are thin, flat and sparse, and read as a few repeated curves rather than a voluminous tangle in the broth." }, "toppings": { "score": 0.6, "reason": "Bok choy leaves with veins, tofu cubes, shiitake halves with gill lines, a nori sheet, scallion rounds and chili bits are all recognizable and varied. They are stylized and flat, the tofu is boxy, and the nori is a rigid rectangle." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a rounded body, a foot, a contact shadow and subtle highlight lines. It is dark and low in contrast, and the base looks slightly floaty." }, "broth": { "score": 0.58, "reason": "The broth is a warm brown with subtle concentric rings and reads as liquid. The surface is fairly flat with little sheen or depth, and the toppings sit on top more than in it." }, "steam": { "score": 0.55, "reason": "Thin translucent wavy wisps rise and change shape between frames, and they do not hide the food. They are fairly uniform and mechanical, and very faint." }, "composition": { "score": 0.62, "reason": "A centered bowl in a dark atmospheric scene keeps the focus on the food, with no clutter. The image is very dark and dim, and the bowl is small within the frame." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, scallion and noodles are all visible, with no animal ingredients." } }, { "noodles": { "score": 0.5, "reason": "Noodles are thin wavy strands with some overlap and read as noodles, but they are flat, sparse and repetitive, and sit as surface lines rather than a voluminous mass." }, "toppings": { "score": 0.58, "reason": "Bok choy leaves with veins, tofu cubes, shiitake halves, a nori sheet, scallion rounds and chili bits give good variety and readable shapes. They are stylized and flat, and the tofu blocks are somewhat rigid." }, "bowl": { "score": 0.66, "reason": "The bowl has a coherent elliptical rim, curved walls, a foot and a contact shadow, plus subtle highlight lines. It is dark and low contrast, and the shadow is slightly detached and flat." }, "broth": { "score": 0.55, "reason": "The broth is a warm brown with subtle concentric rings and reads as soup. It is fairly flat, with little sheen, and the ingredients look somewhat pasted on the surface." }, "steam": { "score": 0.6, "reason": "Several thin translucent wisps rise and change position across frames without hiding the food. They are fairly uniform, sinuous lines and very faint." }, "composition": { "score": 0.68, "reason": "A centered, single focal bowl on a dark atmospheric background with no clutter. The overall image is very dark, and the bowl is small within the frame." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, greens, nori, scallions and noodles are all plant-based, with no visible animal products." } }, { "noodles": { "score": 0.55, "reason": "Noodles are wavy, continuous tan strands with some overlap and bends. They read as noodles but are thin, flat and sparse, with repeated parallel curves and little volume." }, "toppings": { "score": 0.58, "reason": "Toppings are varied and recognizable: bok choy leaves with veins, tofu cubes, shiitake halves, a nori sheet, scallion rings and chili bits. The style is flat and vector-like with little texture, and the tofu blocks are somewhat rigid." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a rounded body, a foot ring, thin glaze lines, highlights and a contact shadow. It is dark and slightly flat, and the base shadow looks a little detached." }, "broth": { "score": 0.55, "reason": "The broth is a warm brown surface with subtle concentric rings and a reasonable sheen. It is fairly flat, and the toppings sit on it with little immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Several translucent wavy wisps rise from the bowl. Their positions and shapes change between frames, and they are subtle and don't hide the food. They look fairly uniform and thin, like mechanical squiggles." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm and cool color contrast and no clutter. The overall image is very dark and the bowl is small in the frame, especially on mobile." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, nori, scallions and noodles are visible. No animal ingredients appear." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.64, "toppings": 0.75, "bowl": 0.7833333333333333, "broth": 0.7066666666666667, "steam": 0.6666666666666667, "composition": 0.8 }, "reward": 0.7228333333333333, "sampleStdDev": 0.013475286020464754, "judgments": [ { "noodles": { "score": 0.65, "reason": "Tan strands are continuous with plausible thickness and cross naturally, but the wavy pattern is repetitive and flat, lying on the surface with little volume or immersion depth." }, "toppings": { "score": 0.75, "reason": "Good variety: green leaves with midribs, tofu cubes with top faces, shiitake slices with gill lines, a nori sheet, edamame/green discs and chili flecks. Tofu cubes read slightly box-like and the green discs are a bit generic." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a highlight band, visible inner wall crescent, tapering body with subtle glaze lines, foot ring and soft contact shadow; construction is consistent, only a slightly squat stylized proportion." }, "broth": { "score": 0.7, "reason": "Brown surface with mottled tonal variation and restrained sheen; ingredients mostly sit believably, though toppings show minimal occlusion shadows and read slightly pasted on top." }, "steam": { "score": 0.65, "reason": "Translucent wisps rise and fade without hiding food, and their shape/position clearly changes across desktop-1/2/3 and pointer frames; however the curls are similar repeated S-shapes, a bit mechanical. Reduced frames are correctly static." }, "composition": { "score": 0.75, "reason": "Centered focal dish, warm under-glow, clean dark scene with no clutter or text. The scene is quite dim, muting food readability, and the mobile frame leaves a large empty dark area below the bowl." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, leafy greens, nori, edamame/scallion, chili and noodles in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.62, "reason": "Tan noodle strands with gentle S-bends lie coherently in the broth on both sides and tuck under toppings, but they read as flat ribbons painted on the surface with limited volume and somewhat repetitive curves; no maze-like scribbles or broken strands." }, "toppings": { "score": 0.75, "reason": "Good variety and readable identities: leafy greens with mid-ribs, tofu cubes with 3D top faces, mushroom slices with gill hatching, scallion rings, red chili bits and a dark nori sheet. Shapes are slightly simple and flat (the nori is a near-rectangular block), but nothing reads as glossy plastic disks." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with an inner-rim highlight, body tapering to a visible foot ring, decorative bands, soft ceramic-like shading and a grounded contact shadow; perspective between rim ellipse and body is consistent with no broken joins." }, "broth": { "score": 0.72, "reason": "Brown liquid surface with tonal variation and a restrained sheen near the rim; ingredients sit in the broth with contact rather than floating above it, though surface texture is fairly flat and depth cues are modest." }, "steam": { "score": 0.55, "reason": "Translucent, irregular wisps rise from the soup, vary in length and fade without hiding the food; only a single labeled frame is present, so ris/fade motion cannot be verified and a mid score is used rather than claiming fluidity." }, "composition": { "score": 0.85, "reason": "Centered, readable focal dish in a restrained dark atmospheric scene with a soft table line, contact shadow and muted harmonious palette; no text, UI, clutter or competing elements, and background glow does not distract from the food." }, "vegan": { "score": 1, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, scallions, chili and nori in plant broth; no meat, fish, egg or dairy-like toppings are discernible." } }, { "noodles": { "score": 0.65, "reason": "Tan strands with consistent thickness and gentle waves peek out on both sides of the bowl and read as noodles in broth, but the mass is sparse, largely hidden under toppings, and the wave pattern is somewhat repetitive rather than individually varied." }, "toppings": { "score": 0.75, "reason": "Good variety of readable ingredients: green leafy vegetables with organic vein shapes, tofu cubes with subtle 3D faces, fan-shaped mushroom slices with gill lines, a dark nori sheet, edamame and red chili flecks. Shapes are simple flat-style but plausible; the nori is a rather rigid rectangle." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl: rim ellipse matches the body curve, inner cavity visible, subtle decorative band lines follow the curvature, foot ring and contact shadow present, consistent perspective across all frames. Slightly stubby proportions and modest highlights keep it from higher marks." }, "broth": { "score": 0.7, "reason": "Brown broth surface with tonal variation, scattered chili and seed specks, and a highlight arc where it meets the inner wall; ingredients generally sit in the liquid rather than floating, though the surface reads a bit flat with little sheen or depth." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the translucent wisps clearly change shape, position and length, rising from the soup and fading near the top without obscuring the food; wisps are somewhat uniform S-curves in character. Reduced-motion frames appropriately hold static wisps." }, "composition": { "score": 0.8, "reason": "Bowl is the clear focal point, centered with a grounded contact shadow, warm glow behind the dish, vignette and faint table line; no text or clutter. The overall scene is intentionally dark, which slightly mutes the food's appetizing color but keeps balance and atmosphere coherent on desktop and mobile." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are leaves, tofu, mushrooms, edamame, chili, nori, and noodles in plant broth; no egg, meat, fish or dairy appears anywhere." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 49.899999999999636, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "180a1a71d3014c1787498af0da9299303a5038616ce4600e944e499a3c990109", "gradedAt": "2026-09-30T22:59:26.702587+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6644444444444444, "toppings": 0.6611111111111111, "bowl": 0.7622222222222222, "broth": 0.6577777777777778, "steam": 0.6588888888888889, "composition": 0.8099999999999999 }, "reward": 0.6962222222222223, "runId": "openai/gpt-6-sol/ultra", "rewardStdDev": 0.08113608823954241, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6233333333333333, "toppings": 0.58, "bowl": 0.78, "broth": 0.5566666666666668, "steam": 0.61, "composition": 0.8433333333333333 }, "reward": 0.6571666666666667, "sampleStdDev": 0.010299676370320286, "judgments": [ { "noodles": { "score": 0.61, "reason": "Continuous golden strands have smooth bends and some overlap, but mostly run as parallel, flat waves with limited tangled volume or immersion." }, "toppings": { "score": 0.58, "reason": "Tofu cubes, veined greens, mushroom caps and scallion rings are readable and varied. Leaves are stiff and paddle-like, mushrooms lack organic texture, and nori resembles rounded gridded cards." }, "bowl": { "score": 0.78, "reason": "Elliptical rim, shaded curved walls and supporting foot form a coherent vessel with convincing depth. Side joins are abrupt, and the outlined foot feels less solid than the body." }, "broth": { "score": 0.55, "reason": "Warm amber color and scattered garnish suggest soup, but broad opaque color bands look more like a solid painted surface; ingredient contact and submersion are weak." }, "steam": { "score": 0.57, "reason": "Translucent blurred wisps rise behind the soup without obscuring it, and their contours change across normal frames. They remain fairly columnar and repetitive, with limited visible spreading and fading." }, "composition": { "score": 0.84, "reason": "A single readable dish dominates the uncluttered dark setting, with restrained amber-green colors and soft grounding shadows. No distracting text or UI appears; the mobile bowl sits very close to the side edges." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.65, "reason": "Continuous golden strands have smooth bends and some readable overlaps, but largely parallel waves and broad flat shading limit tangled volume and believable immersion." }, "toppings": { "score": 0.58, "reason": "Veined greens, scored mushroom caps, shaded tofu cubes and scallion rings provide readable variety. Greens remain stiff and paddle-like, while nori resembles thick rounded grid cards rather than irregular thin seaweed." }, "bowl": { "score": 0.78, "reason": "Elliptical rim, deep curved walls and warm edge highlights form a coherent vessel with grounded shadow. The outlined foot is sparsely modeled, and abrupt lateral rim joins slightly weaken the construction." }, "broth": { "score": 0.56, "reason": "Warm brown and amber tones suggest an appetizing broth, but broad graphic bands lack liquid sheen and depth; ingredient contact and partial submersion are weakly expressed." }, "steam": { "score": 0.64, "reason": "Soft translucent wisps rise behind the soup without obscuring it, and their contours and brightness change across normal frames. The broad, similar vertical plumes offer limited evidence of spreading and continuous fading." }, "composition": { "score": 0.85, "reason": "A clear single-dish focal point, restrained dark background and coordinated amber-green palette create a polished atmospheric illustration without text or clutter. Desktop spacing is balanced; mobile framing leaves very little lateral clearance." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.61, "reason": "Continuous golden strands have smooth bends and some overlap, but mostly follow parallel waves in a shallow, ribbon-like layer with limited tangled volume." }, "toppings": { "score": 0.58, "reason": "Tofu cubes, veined greens, mushroom caps and scallion rings are readable and varied. Leaves are stiff and overly smooth, mushrooms have sparse texture, and nori resembles thick rounded grid cards." }, "bowl": { "score": 0.78, "reason": "Elliptical rim, shaded curved walls and ground shadow establish coherent depth and support. The outlined foot is minimally connected and the side rim joins are abrupt, reducing material plausibility." }, "broth": { "score": 0.56, "reason": "Warm amber coloration and scattered garnish suggest soup, but broad flat color bands lack liquid sheen, and ingredients show little immersion or contact detail." }, "steam": { "score": 0.62, "reason": "Soft translucent wisps remain above the soup without obscuring food, and their contours and brightness change across normal frames. Similar upright plumes and limited visible spreading make the motion somewhat mechanical." }, "composition": { "score": 0.84, "reason": "A single prominent dish, restrained dark background, warm food colors and subtle grounding shadow create a clean atmospheric focus without text or clutter. Mobile framing leaves the bowl very close to the side edges." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, scallions and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.66, "toppings": 0.62, "bowl": 0.6733333333333333, "broth": 0.6333333333333333, "steam": 0.5666666666666667, "composition": 0.72 }, "reward": 0.642, "sampleStdDev": 0.02146508793366567, "judgments": [ { "noodles": { "score": 0.62, "reason": "Flowing wavy cream strands with overlap and tapered ends read as noodles and sit in the broth. They are fairly uniform, parallel and flat-looking, with little volume and no lifted bundle." }, "toppings": { "score": 0.6, "reason": "Bok choy with pale stems and veined leaves is recognizable, as are the tofu cubes, the shiitake cap and the scallion rings. The nori is two flat grid-patterned cards standing behind the bowl, and the mushroom is half hidden. Overall the set is stylized and somewhat generic, but varied." }, "bowl": { "score": 0.66, "reason": "The bowl has a coherent elliptical rim, a dark lacquer body with copper outline highlights, a foot ring and a ground shadow. The rim has a slightly odd double-band edge, and the foot looks small and loosely attached. Perspective is consistent." }, "broth": { "score": 0.62, "reason": "Warm brown broth with layered gradients and chili-oil specks looks appetizing and has depth. The surface is fairly flat, and the toppings sit on it with minimal ripple or immersion cues." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and change shape between frames, which shows motion. They are very subtle, somewhat blobby and confined to a small area, and they don't connect clearly to the soup surface." }, "composition": { "score": 0.72, "reason": "A centered bowl on a dark background with warm lighting keeps the food as the clear focal point. The palette is cohesive and there is no clutter. The bowl sits a little high, leaving a large empty lower area." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushroom, bok choy, scallion, nori and noodles are visible, with no egg, meat or fish. The broth appears plant-based." } }, { "noodles": { "score": 0.68, "reason": "Flowing, wavy tan strands with soft bends, overlap and highlights read as noodles sitting in the broth. They are fairly uniform and parallel, and they end abruptly on the right side, so they look slightly flat and repetitive." }, "toppings": { "score": 0.66, "reason": "Bok choy with veined leaves and pale stems, cubed tofu with shading, a shiitake cap with gill lines, scallion rings and nori squares are all readable. The nori reads as stiff grid-patterned cards standing behind the bowl. The tofu and mushroom are simplified, and the variety is decent." }, "bowl": { "score": 0.7, "reason": "A coherent elliptical rim with a dark inner lip, a curved body with rim highlights, a foot ring and a contact shadow. The rim has a slightly odd brown notch at the sides, and the dark body is a bit flat and murky, but the perspective is consistent." }, "broth": { "score": 0.62, "reason": "The warm brown gradient with layered bands and small chili oil flecks gives depth, and the ingredients sit in it believably. The surface is flat and stylized, with little sheen and no occlusion at the edges of the ingredients." }, "steam": { "score": 0.6, "reason": "Faint translucent wisps rise above the bowl and shift between frames, and they do not hide the food. They are subtle and somewhat blurry, with a few small hard-edged streaks that look mechanical." }, "composition": { "score": 0.72, "reason": "A centered bowl on a dark vignette with warm glow and a good color palette keeps the focus on the food. There is no clutter or text. The bowl sits a bit low, leaving a lot of empty space above it." }, "vegan": { "score": 0.93, "reason": "Tofu, mushroom, bok choy, scallions, nori and noodles are all plant-based. There is no visible meat or egg." } }, { "noodles": { "score": 0.68, "reason": "Flowing, wavy tan strands overlap with coherent volume and sit believably in the broth. They are a bit flat and uniform, and the strand ends look cut off, so they read as stylized rather than highly convincing." }, "toppings": { "score": 0.6, "reason": "Bok choy with a veined leaf and pale stem, tofu cubes, a shiitake cap, scallion rings and nori are all recognizable and varied. The nori reads as stiff grid-lined cards, the tofu is blocky, and the mushroom is mostly hidden." }, "bowl": { "score": 0.66, "reason": "The rim ellipse, curved body, foot ring and cast shadow are coherent, with a copper edge highlight. The body is dark and a little flat. The foot ring looks slightly detached, and the lower contour comes to an odd point." }, "broth": { "score": 0.66, "reason": "The warm brown broth has subtle tonal banding and a soft sheen, and the ingredients sit in it fairly well. It is somewhat flat, with faint concentric bands and little surface texture or immersion detail." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and vary between frames, so there is motion. The steam is very subtle and low contrast, and it sits mostly behind the nori rather than emerging from the soup surface." }, "composition": { "score": 0.72, "reason": "A centered, well-lit bowl on a dark atmospheric background, with a warm palette and green accents. The food is the clear focus and there is no clutter. The bowl sits slightly low and the bowl body is dark." }, "vegan": { "score": 0.95, "reason": "Tofu, mushroom, bok choy, scallion, nori and noodles. No visible animal products." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.71, "toppings": 0.7833333333333333, "bowl": 0.8333333333333334, "broth": 0.7833333333333333, "steam": 0.8, "composition": 0.8666666666666667 }, "reward": 0.7895, "sampleStdDev": 0.03175295261861491, "judgments": [ { "noodles": { "score": 0.78, "reason": "Continuous wavy cream-colored strands flow coherently across the broth with plausible bends and layering, shaded for slight volume. They read clearly as noodles rather than wires, though they lie in one flat band on the surface with fairly uniform rhythm and minimal immersion into the liquid." }, "toppings": { "score": 0.8, "reason": "Good variety: bok choy with pale stems, a sliced shiitake-style mushroom with gill lines, tofu cubes with seared-face detail, scallion rings, corn kernels, chili flecks and nori sheets leaning at the rim. Shapes are mostly organic and readable; the nori reads as slightly rigid grid-patterned cards and tofu cubes are a bit geometric." }, "bowl": { "score": 0.8, "reason": "Coherent lacquer-look bowl: elliptical rim with visible wall thickness, tapering body, foot ring with contact shadow and warm rim highlights. Perspective is consistent rim-to-base. Minor flaws: small notch-like seams at the left and right rim edges and a somewhat flat body shading." }, "broth": { "score": 0.75, "reason": "Appetizing rich brown surface with lighter oil swirls, darker patches, chili oil dots and sesame specks giving texture and depth. Most ingredients sit believably on/in the surface with soft contact shadows, though tofu cubes and noodles read slightly pasted-on with little occlusion." }, "steam": { "score": 0.8, "reason": "Across the time-ordered desktop frames the translucent gray wisps visibly change shape, rise and fade, with pointer frames showing altered plume positions. Wisps are irregular, translucent and never obscure the food; they are somewhat evenly spaced parallel plumes but clearly animated, not static." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric ground with a subtle warm glow behind the bowl, coherent lighting and a soft floor shadow. The food remains the clear focal point with no text, UI or competing elements; presentation is restrained and attractive." }, "vegan": { "score": 1, "reason": "All visible ingredients are plant-based: noodles, tofu cubes, mushroom, bok choy, scallion, corn, chili and nori in what appears to be a miso-style broth. No animal-derived toppings are visible." } }, { "noodles": { "score": 0.6, "reason": "Noodles read as a continuous cream-colored bundle with plausible bends, but the strands form a repeating flat wave pattern with little volume or overlap depth, sitting on the surface rather than showing believable immersion." }, "toppings": { "score": 0.75, "reason": "Good variety: recognizable shiitake-style mushroom cap with gill lines, tofu cubes with bevels, bok choy stalks with lighter ribs, scallion rings, corn and chili flecks. The nori sheets are rigid grid-patterned rounded cards that look slightly stiff, and tofu forms are simple." }, "bowl": { "score": 0.85, "reason": "Coherent rounded ceramic body with a thick dark rim band, elliptical rim in consistent perspective, visible foot ring, contact shadow and soft highlights; only a small notch/break in the right rim edge and a very dark body reduce polish." }, "broth": { "score": 0.8, "reason": "Warm brown gradient surface with darker depth swirls, restrained sheen, oil specks and seasoning; toppings sit in the liquid with soft shadows, though the surface is fairly smooth and slightly uniform." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps rise and change shape across the normal frames (plume shifts and brightens between desktop-1/2/3 and pointer frames), spreading without hiding food; wisps are a bit sparse and blob-like at the top." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with warm food tones against a clean dark atmospheric backdrop, subtle glow and shadow, no clutter or UI; presentation is balanced and readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushroom, bok choy, nori, scallion, corn and plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.75, "reason": "Continuous wavy strands with highlight/edge shading and plausible bends spread across the broth; coherent overlaps, but they read as a fairly flat single layer of drawn ribbons with limited volume/submergence." }, "toppings": { "score": 0.8, "reason": "Readable variety: bok choy with pale stems, a scored shiitake cap, beveled tofu cubes, scallion rings, corn/chili specks and angled nori sheets; nori's rounded grid-patterned squares look slightly card-like but stay stylized-plausible." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic construction: thick elliptical rim, dark cavity, curved body with glaze highlight and swirl, visible foot with contact shadow. Minor oddity: small tab-like rim ends at left/right edges read slightly detached." }, "broth": { "score": 0.8, "reason": "Appetizing orange-brown surface with chili-oil swirls, red oil dots and specks; tofu and greens sit with soft occlusion shadows. Slightly flat sheen and toppings mostly float on the surface rather than sinking in." }, "steam": { "score": 0.85, "reason": "Across desktop-1/2/3 the translucent grey wisps change shape and position, with bright wisp tips appearing at shifting spots, so genuine rising/fading motion is supported; wisps stay thin, irregular and never hide the food." }, "composition": { "score": 0.9, "reason": "Centered focal bowl in a restrained dark scene with warm under-glow and vignette; steam frames the dish, no text/UI/clutter, and the mobile crop keeps the bowl balanced and fully readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushroom, bok choy, scallions, nori, corn and chili in a plant broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9788888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50.0, "frameIntervalP95Ms": 66.60000000000036, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8f2a03502778ac0f0fe9ae75c308aafe360c205ce41c6095bee376d9c96a6f43", "gradedAt": "2026-09-30T23:00:11.496952+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5700000000000001, "toppings": 0.6577777777777778, "bowl": 0.7633333333333333, "broth": 0.6, "steam": 0.7255555555555555, "composition": 0.7944444444444444 }, "reward": 0.6793888888888889, "runId": "openai/gpt-6-sol/xhigh", "rewardStdDev": 0.07593534042806374, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.49333333333333335, "toppings": 0.65, "bowl": 0.8733333333333333, "broth": 0.5666666666666667, "steam": 0.76, "composition": 0.8533333333333333 }, "reward": 0.6918333333333333, "sampleStdDev": 0.0045092497528228985, "judgments": [ { "noodles": { "score": 0.48, "reason": "Golden strands have readable thickness and overlapping bends, but numerous short, angular segments form a flat, wire-like tangle rather than supple immersed noodles." }, "toppings": { "score": 0.65, "reason": "Veined greens, scallion rings, shaded mushroom caps and porous tofu cubes provide recognizable variety. Leaves are stiff and repetitive, mushrooms lack organic detail, and nori resembles regular gridded cards." }, "bowl": { "score": 0.87, "reason": "Coherent elliptical rim, visible ceramic thickness, smoothly shaded curved walls and subtle surface grain convincingly establish depth. Small angular joins at the side rim and indistinct base contact are minor shortcomings." }, "broth": { "score": 0.56, "reason": "Warm brown coloration and scattered surface flecks suggest soup, but the surface has little liquid sheen or depth; ingredients mostly appear layered above it with limited immersion." }, "steam": { "score": 0.76, "reason": "Translucent wisps change curvature and position across normal frames, emerge behind the soup and fade upward without obscuring food. Their similarly spaced, softly blurred columns remain somewhat mechanical." }, "composition": { "score": 0.85, "reason": "A single clearly readable bowl dominates the restrained dark setting, with warm food colors contrasting green toppings and softly lit ceramic. No distracting text or clutter is visible; mobile framing leaves the left rim very close to the edge." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, scallions, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have consistent thickness and readable overlaps, but many short, angular bends and exposed ends form a flat wire-like tangle rather than supple immersed noodles." }, "toppings": { "score": 0.65, "reason": "Veined greens, scallion rings, shaded mushroom caps and porous tofu cubes provide readable variety. Leaves are rather stiff and symmetrical, mushrooms lack organic detail, and nori resembles regular gridded cards." }, "bowl": { "score": 0.87, "reason": "Coherent elliptical rim, visible wall thickness, smoothly shaded ceramic body and subtle ground shadow give convincing depth and support. Small abrupt highlight joins at the rim sides are the main visible refinement needed." }, "broth": { "score": 0.56, "reason": "Warm brown coloration and scattered surface flecks suggest broth, but the surface has little liquid sheen or translucency; ingredients largely appear laid on top rather than partially immersed." }, "steam": { "score": 0.76, "reason": "Normal frames show changing translucent wisps connected to the soup, with softened fading tips that leave the food readable. The few similarly shaped vertical plumes remain somewhat repetitive; sampled frames do not establish continuous fluid motion." }, "composition": { "score": 0.85, "reason": "A large, clear focal bowl, restrained dark background and balanced green-and-gold accents create an attractive atmospheric illustration without text or clutter. Mobile framing puts the bowl very close to the left edge." }, "vegan": { "score": 1, "reason": "Visible toppings read as tofu, mushrooms, leafy greens, scallions and seaweed alongside noodles; no visibly animal-derived ingredients are present." } }, { "noodles": { "score": 0.5, "reason": "Golden strands have readable outlines and overlapping bends, but numerous short angular segments form a flat, wire-like tangle rather than supple noodle volume." }, "toppings": { "score": 0.65, "reason": "Veined greens, scallion rings, shaded mushroom caps and porous tofu provide readable variety. Leaves are stiff and geometric, mushrooms lack organic detail, and nori resembles regular gridded cards." }, "bowl": { "score": 0.88, "reason": "Coherent elliptical rim, consistent wall curvature, cream edge highlights and softly textured ceramic shading establish convincing depth. No material construction defect is observed; ground contact is somewhat faint." }, "broth": { "score": 0.58, "reason": "Warm brown shading and small surface flecks suggest soup, but the surface lacks convincing liquid sheen and immersion; most ingredients appear layered atop a flat field." }, "steam": { "score": 0.76, "reason": "Normal frames show changing translucent curling wisps anchored above the soup, with fading upper ends and unobscured food. The plumes remain similarly spaced, broad blurred columns with limited fine irregularity." }, "composition": { "score": 0.86, "reason": "The single bowl is clearly emphasized against a restrained dark background, with harmonious warm food colors and cool ceramic shading and no distracting UI. Desktop framing is balanced; mobile placement is slightly cramped at the left edge." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, seaweed and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.6066666666666667, "bowl": 0.5833333333333334, "broth": 0.5166666666666667, "steam": 0.6, "composition": 0.68 }, "reward": 0.598, "sampleStdDev": 0.010000000000000009, "judgments": [ { "noodles": { "score": 0.6, "reason": "Fine, wavy golden strands overlap with some volume and read as noodles, but they look like short, thin, crispy scribbles and lack long supple continuity. They sit in a pile and do not clearly submerge in the broth." }, "toppings": { "score": 0.6, "reason": "Bok choy leaves with veins, tofu cubes, shiitake-like caps and scallion rings are all recognizable and vegan-friendly. The mushrooms are glossy ovals that look plastic, the tofu blocks are simple, and the scallion rings are sparse outlines." }, "bowl": { "score": 0.6, "reason": "The ceramic bowl has a readable rim, a curved body, a soft gradient and a base highlight. The rim is a flat ellipse with odd side notches, and the inner cavity looks pasted on, with no wall thickness visible inside. The base has no visible foot or contact shadow." }, "broth": { "score": 0.5, "reason": "The broth is a warm reddish-brown with a gradient, but it reads fairly flat and opaque. The ingredients have little immersion, ripple or occlusion, and they look laid on top of it." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise above the bowl and change shape between frames, which shows continuous motion. They are thin and somewhat uniform, but they do not hide the food." }, "composition": { "score": 0.68, "reason": "A centered bowl sits on a dark background with warm colors and good focus. The food is dark-on-dark at the edges and the bowl takes up a lot of the frame, but there is no clutter." }, "vegan": { "score": 0.95, "reason": "The bowl shows only tofu, mushrooms, bok choy, scallions and noodles, with no visible animal products." } }, { "noodles": { "score": 0.6, "reason": "Fine, wavy golden strands overlap with some volume and read as noodles. They are short and thin, more like crispy or shredded noodles, and they sit as a flat mat with little depth." }, "toppings": { "score": 0.62, "reason": "The bok choy leaves with veins, shiitake caps with highlights, tofu cubes and scallion rings are all recognizable and varied. The mushrooms are smooth glossy ovals and the scallion rings are thin outlines. The nori is dark and ambiguous." }, "bowl": { "score": 0.6, "reason": "The bowl has a clear rim, a rounded body with gradient shading and a base highlight. The rim has odd small notches at the sides and looks like a lid or flange. The rim is wider than the body, and the interior perspective looks flat." }, "broth": { "score": 0.55, "reason": "The broth is a warm reddish-brown with a slight glow, and the ingredients sit in it plausibly. The surface is fairly flat and murky, with little sheen or depth, and the toppings look slightly pasted on." }, "steam": { "score": 0.6, "reason": "Translucent wispy plumes change shape and position between frames, so they read as moving and fading. They are faint, somewhat mechanical in their sine shape, and detached from the surface." }, "composition": { "score": 0.68, "reason": "A centered bowl on a dark background makes the food the focal point, with a warm palette and no clutter. The bowl sits a little high and the lighting is fairly flat." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, greens, scallions and noodles are all plant-based, and no animal ingredients are visible." } }, { "noodles": { "score": 0.6, "reason": "Fine, wavy golden strands overlap with a believable tangled volume, but they read as short, thin, somewhat scribbly and repetitive. They lie flat on the surface rather than sitting in the broth." }, "toppings": { "score": 0.6, "reason": "Bok choy leaves with veins, shiitake caps with highlights, tofu cubes and scallion rings are all recognizable. The mushrooms look like glossy ovals, the scallions are thin outlined rings, and the dark green nori is hard to read. Variety is decent, but the arrangement is sparse and symmetric." }, "bowl": { "score": 0.55, "reason": "The bowl has a clear rim, a rounded body and a base highlight, and its gray ceramic shading is pleasant. The rim is seen as a flat ellipse with a odd notch or tab at each side, and the broth inset looks like a pasted-on oval. The perspective between rim and body is slightly incompatible." }, "broth": { "score": 0.5, "reason": "The warm reddish-brown broth is appetizing in color and has some gradient, but it reads as a flat painted disc. Immersion and contact with the ingredients are weak, and the toppings seem to sit on top." }, "steam": { "score": 0.6, "reason": "Translucent, wispy plumes rise above the bowl, and their shapes change between frames in a soft, continuous way. They are fairly faint and somewhat uniform in style, but they do not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered on a dark, restrained background with warm glow, good color harmony and a clear focal point, and there is no clutter. The bowl is slightly low in the frame, and the mobile view leaves large empty areas." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, scallions and noodles are visible, with no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6166666666666667, "toppings": 0.7166666666666667, "bowl": 0.8333333333333334, "broth": 0.7166666666666667, "steam": 0.8166666666666667, "composition": 0.85 }, "reward": 0.7483333333333334, "sampleStdDev": 0.011814539065631543, "judgments": [ { "noodles": { "score": 0.65, "reason": "Tan strands have supple wavy bends and consistent thickness, but the mass reads as loose flat scribbles lying on the surface: small loop clusters repeat, some strands end abruptly, and they lack convincing immersion in the broth." }, "toppings": { "score": 0.7, "reason": "Good variety: glossy brown shiitake-like caps, bokchoy-style greens with pale stalks, dark nori sheets at the back, and scattered scallion rings all read correctly. Tofu cubes are slightly boxy with crate-like orange outlines and the mushrooms a bit like simple glossy ovals." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: clean elliptical rim with a bright highlight edge, consistent rim-to-body perspective, smooth gray glaze with soft gradients, believable foot shadow and contact. Minor quibble: the small side lugs are slightly ambiguous, though not broken." }, "broth": { "score": 0.7, "reason": "Rich brown-red liquid with a warm central glow, darker edges and restrained sheen, plus floating specks. Ingredients mostly sit believably, but the noodle tangle and tofu look somewhat pasted on top, and the surface is a bit flat with faint concentric arcs near the rim." }, "steam": { "score": 0.85, "reason": "Translucent gray wisps rise, curl, drift and fade, changing shape and position clearly across the 1040ms/3015ms/4779ms frames without obscuring the food; reduced-motion frames correctly hold a static pose. Wisps are slightly repetitive S-curves." }, "composition": { "score": 0.85, "reason": "Balanced dark scene with a soft warm radial glow behind a centered, well-scaled focal bowl; no text, UI or clutter; mobile viewport crops cleanly. The dish stays the clear focus." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, leafy greens, tofu cubes, scallions, seaweed, noodles and plant broth; no animal-derived toppings are visible." } }, { "noodles": { "score": 0.6, "reason": "Tan strands are continuous with varied wavy bends and a plausible thin gauge, but they lie as a flat tangle on the surface with little volume or visible immersion; lower strands collapse into scribbly loop clusters rather than coherent nested noodles." }, "toppings": { "score": 0.75, "reason": "Good variety: veined dark-green bok choy leaves, glossy brown shiitake caps with highlights, tofu cubes, dark nori strips at the back rim, scallion rings and sesame specks. Shapes are simplified and the tofu cubes read slightly box-like, but all are recognizable with appropriate scale." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: elliptical rim with a pale lip, dark inner-wall ring, smoothly tapering gray body with soft gradient shading, and a grounded contact shadow beneath. Minor quibbles: tiny tab-like flanges at the 3 and 9 o'clock rim edges and a fairly featureless glaze, but no broken joins or perspective mismatch." }, "broth": { "score": 0.7, "reason": "Rich reddish-brown liquid with restrained radial sheen and a dark meniscus ring where it meets the wall, and toppings carry subtle shadows suggesting they rest in the soup. However the surface reads somewhat flat and solid, and noodle strands sit atop rather than dipping into the liquid." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the broth and differ clearly across desktop-1/2/3 and pointer frames (S-curls becoming vertical wavering plumes), indicating continuous motion; they stay thin and never obscure the food. Wisps are somewhat similar in curl motif, and reduced frames are correctly static." }, "composition": { "score": 0.85, "reason": "Centered bowl dominates the frame against a dark vignetted background with a warm glow behind the dish; steam adds vertical balance, colors are harmonious, and there is no text, UI or clutter. Mobile framing keeps the full bowl and steam readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, leafy greens, tofu cubes, nori, scallion and sesame in a plant-colored broth; no egg, meat, fish or dairy appears anywhere in the frames." } }, { "noodles": { "score": 0.6, "reason": "Noodles are recognizable thin tan strands with organic bends, but they read as scattered uniform squiggle strokes with abrupt rounded ends rather than continuous bundled strands with volume; no coherent nesting or submersion is shown." }, "toppings": { "score": 0.7, "reason": "Good variety: brown mushroom caps with highlights, veined green leaves, scallion rings, and patterned tofu cubes all read as food. The dark green strips at the back look like rigid flat panels, and leaf/cube shapes are somewhat generic and flat-shaded." }, "bowl": { "score": 0.8, "reason": "Coherent grey ceramic bowl with consistent rim band, elliptical opening, deep hemispherical body, foot ring, soft highlights and a contact shadow. Minor oddities: small tick-like tabs protruding at the left and right rim edges." }, "broth": { "score": 0.75, "reason": "Warm brown broth with darker edges suggesting depth, subtle sheen flecks and light garnish scattered on the surface; ingredients mostly sit believably on/in the liquid, though contact shadows and immersion cues are minimal." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the wisps change shape, position and spread, so steam is animated, translucent and irregular without hiding food; some stylized loop curls repeat (e.g., desktop-1 vs pointer-right), hinting at a looping pattern." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with balanced proportions, warm broth against a restrained dark vignette background, coherent soft glow and lighting, no text, UI or clutter; presentation is clean in both desktop and mobile views." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu cubes, leafy greens, scallions, seaweed-like strips and noodles in plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 100.0, "frameIntervalP95Ms": 116.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a8c9263b36fc6d2eede87b90be3d8d6c15f6112f5c785846d3bad722b22c3e7a", "gradedAt": "2026-09-30T23:30:53.408106+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.39444444444444443, "toppings": 0.5355555555555556, "bowl": 0.7122222222222222, "broth": 0.5566666666666666, "steam": 0.6933333333333334, "composition": 0.7122222222222222 }, "reward": 0.5861111111111111, "runId": "xai/grok-4.5/high", "rewardStdDev": 0.07534556193767292, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.49, "bowl": 0.76, "broth": 0.45333333333333337, "steam": 0.62, "composition": 0.7366666666666667 }, "reward": 0.5465, "sampleStdDev": 0.004444097208657767, "judgments": [ { "noodles": { "score": 0.3, "reason": "Tan strands are readable, but mostly form straight, angular sticks with abrupt bends and little supple overlap or piled volume." }, "toppings": { "score": 0.5, "reason": "Tofu, mushroom caps, a green leaf and scallion pieces provide recognizable variety. Shapes remain simplistic, with rounded rectangular nori, minimally textured cubes and schematic mushroom details." }, "bowl": { "score": 0.76, "reason": "A coherent elliptical rim, inset cavity and curved shaded walls establish convincing depth. The dark material is somewhat flat, and the base lacks a clear supporting contact shadow." }, "broth": { "score": 0.46, "reason": "Warm amber shading suggests soup, but the surface is smooth and solid-looking, with little liquid sheen or texture; ingredients mostly appear laid over it rather than immersed." }, "steam": { "score": 0.63, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the food. Broad blurry patches and separated glowing blobs are less convincing than connected, tapering wisps." }, "composition": { "score": 0.74, "reason": "The single bowl is clearly framed against a restrained dark background with harmonious warm and green accents and no clutter. Mobile presentation leaves extensive empty space and makes the food details small." }, "vegan": { "score": 1, "reason": "Visible tofu-like cubes, mushrooms, greens, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.3, "reason": "Tan strands are readable, but mostly form straight, angular overlapping sticks with little supple curvature, volume, or convincing immersion." }, "toppings": { "score": 0.48, "reason": "Tofu, mushroom caps, greens and corn are distinguishable with some small highlights. Most remain simple geometric symbols; the nori is a thick rectangular card and the single leaf has minimal organic detail." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inner cavity and rounded body have coherent perspective and useful shaded depth. Material highlights are sparse, and the dark base lacks a clear supporting surface or contact shadow." }, "broth": { "score": 0.44, "reason": "Warm amber shading separates the soup from the rim, but the surface looks flat and opaque, with little liquid sheen or texture. Ingredients appear arranged on top rather than convincingly immersed." }, "steam": { "score": 0.62, "reason": "Normal frames show changing translucent clouds above the soup without obscuring the food. The broad blurred patches read more as drifting fog than delicate rising wisps, with some disconnected upper blobs." }, "composition": { "score": 0.74, "reason": "A single centered bowl and warm restrained palette create a clear focal scene without text or clutter. The mobile dish is rather small within extensive empty space, and the very dark lower bowl loses definition." }, "vegan": { "score": 1, "reason": "Visible tofu-like cubes, mushrooms, seaweed, greens, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.3, "reason": "Tan strands are readable, but mostly form straight sticks with angular joins and shallow crossings rather than supple noodles with believable volume." }, "toppings": { "score": 0.49, "reason": "Tofu, mushroom caps, greens and corn provide recognizable variety, but remain simplified symbols; the nori is a rigid rectangular card and the leaf lacks organic texture." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, inset cavity and shaded curved walls form a coherent vessel with readable thickness. Material highlights are limited, and ground contact is indistinct." }, "broth": { "score": 0.46, "reason": "Warm amber shading suggests soup, but the surface is largely matte and flat, with little liquid texture or convincing immersion around ingredients." }, "steam": { "score": 0.61, "reason": "Normal frames show changing translucent clouds above the soup without hiding the food. The broad blurred patches look more like diffuse floating fog than fine rising wisps." }, "composition": { "score": 0.73, "reason": "A single centered bowl and restrained warm palette remain clear against the dark uncluttered background. Mobile framing leaves substantial empty space and makes the food details small." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, seaweed, greens and corn; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.48333333333333334, "bowl": 0.6, "broth": 0.55, "steam": 0.6866666666666666, "composition": 0.6 }, "reward": 0.5388333333333334, "sampleStdDev": 0.008220908303425659, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin, flat, straight strands that look like sticks or straw. They are rigid and lack bends, overlap or volume, and they sit in a narrow band on top of the broth." }, "toppings": { "score": 0.5, "reason": "The toppings include a leaf, tofu cubes, mushroom caps, scallion bits and corn. The leaf reads well, but the nori is a rigid dark green rectangle and the tofu is blocky. The toppings are small and look pasted on." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, inner wall and dark ceramic body with a subtle highlight. The body is very dark and merges into the background, and the base is pointed and unsupported, so it seems to float." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange with a glow and reads as soup. It is fairly flat and uniform, with little surface texture or sheen. Toppings sit on top rather than sinking into it." }, "steam": { "score": 0.7, "reason": "Soft translucent wisps with small sparks rise and change shape between frames. They are somewhat blobby and foggy, but they do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is a centered focal point in a dark, moody scene with warm colors. The food is small relative to the bowl and the bowl merges into the dark background, which leaves the lower half of the frame empty." }, "vegan": { "score": 0.9, "reason": "The visible toppings are tofu, mushrooms, nori, scallion, corn and a leaf, with no animal products. The appearance is plausibly vegan." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, flat, straight tan ribbons lying in a narrow cluster. They look like sticks or straw with little bend or volume, and they read as rigid and sparse." }, "toppings": { "score": 0.5, "reason": "The toppings are recognizable: a green leaf, tofu cubes, mushroom caps, scallion bits and a nori rectangle. They are small and crowded into a thin strip, and the nori is a rigid dark green block. The tofu cubes look like plain blocks, and the corn-like yellow dots are generic." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, an inner wall and a curved body with a soft highlight. The body is very dark and blends into the background. It has no visible foot or base, so it seems to float without support, and the rim perspective is slightly off against the deep body." }, "broth": { "score": 0.55, "reason": "The broth is a warm orange with a glow gradient and reads as liquid. It is fairly flat and smooth, with little sheen or texture, and the toppings sit on its surface with little sense of immersion." }, "steam": { "score": 0.68, "reason": "The steam is soft, translucent and irregular, and its shape changes from frame to frame. It drifts upward and fades, with small particles. It is somewhat blobby and foggy, but it does not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered and is the clear focal point in a dark, moody scene. The food sits in a small area and the bowl merges into the dark background. There is no distracting clutter." }, "vegan": { "score": 0.88, "reason": "Tofu, mushrooms, scallions, nori, a leaf and noodles all look plant-based. The small reddish bits are ambiguous but are probably chili." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin, flat, straight tan strips crowded in a small band. They look like sticks or straw with no supple curves, volume or overlap, so they read as rigid." }, "toppings": { "score": 0.45, "reason": "Toppings are recognizable: tofu cubes, a green leaf, mushroom caps, scallion bits and a dark nori rectangle. They are small and crude, though. The nori is a thick, boxy card, and the corn bits are plastic-looking dots." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, dark walls and a rounded base with a subtle highlight. It is dark and murky, and it seems to float without a clear support surface. The rim also looks thick." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and some glow, which is appetizing. It is fairly flat, with little surface texture. Ingredients sit on top, with weak immersion and few ripples." }, "steam": { "score": 0.68, "reason": "Soft translucent wisps with sparks rise above the bowl and change shape between frames. They are somewhat blobby and foggy, but they do not hide the food." }, "composition": { "score": 0.6, "reason": "The bowl is centered in a dark, moody scene with a warm focal glow. The food is small and dim, and there is a lot of empty space. The pointer shifts look fine." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, scallion, corn, leaf and nori are visible. No animal ingredients appear." } } ], "diagnostics": { "vegan": 0.8933333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.48333333333333334, "toppings": 0.6333333333333333, "bowl": 0.7766666666666667, "broth": 0.6666666666666666, "steam": 0.7733333333333333, "composition": 0.8 }, "reward": 0.673, "sampleStdDev": 0.023574350468252556, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are thin tan strands lying almost horizontally across the broth with only slight bends; they read as scattered, fairly rigid matchstick-like strands lying flat on the surface rather than supple bundled noodles with volume, though they are continuous and not maze-like." }, "toppings": { "score": 0.6, "reason": "Recognizable stylized items: white tofu cubes, mushroom slices with cross-sections, segmented scallion sticks, a green leaf, yellow bits, and a dark block (nori). Defects: the tofu are plain flat squares, and the dark green nori is a featureless boxy rectangle; shapes are simple geometric rather than organic." }, "bowl": { "score": 0.75, "reason": "Coherent thick rim ellipse, dark ceramic-like body with subtle gradient and a rim highlight, tapering base, consistent 3/4 perspective and clean joins. Minor shortcoming: the bowl floats in a void with no visible surface contact or shadow." }, "broth": { "score": 0.65, "reason": "Warm amber liquid surface with radial glow, darker inner shadow near the rim, and toppings sitting on the surface with reasonable contact. Slightly flat and gradient-only; immersion depth is minimal." }, "steam": { "score": 0.75, "reason": "Across desktop-1/2/3 and pointer frames the plume changes shape continuously: translucent irregular wisps rising, spreading and fading with small glowing particles, never hiding the food. Reduced-motion frames are properly static. Slightly fog-like rather than distinct wisps." }, "composition": { "score": 0.75, "reason": "Clean dark atmospheric scene with the bowl as clear focal point, warm broth glow against near-black background, no text or clutter. Slight imbalance from large empty dark space and the unsupported floating bowl; mobile frame places the dish small with much empty area above." }, "vegan": { "score": 0.6, "reason": "Visible ingredients are mostly plant-based: tofu, mushrooms, scallions, leaf, nori-like block. However, a small pinkish-red swirled piece near the center is ambiguous and could read as narutomaki fish cake, so an intermediate score is used." } }, { "noodles": { "score": 0.5, "reason": "Noodles are readable as thin tan strands fanned across the broth, but they render as mostly straight, rigid, flat sticks lying side-by-side with little supple bend, volume, or coherent overlap, giving a straw-like rather than noodle-like quality." }, "toppings": { "score": 0.65, "reason": "Good variety and readability: a dark nori sheet with fold detail, white tofu cubes, brown mushroom halves with pale centers, scallion segments, chili slivers, and a well-drawn green leaf. Shapes are somewhat simplified/geometric (blocky nori rectangle, marshmallow-like tofu), keeping it competent rather than polished." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: clean elliptical rim with an inner lip highlight, deep tapered walls, visible foot, believable contact shadow, and consistent perspective that even tilts plausibly in the pointer frames. Only minor softness in the very dark body detail." }, "broth": { "score": 0.7, "reason": "Warm orange-brown surface with restrained radial sheen and darker edges reads as appetizing broth; nori stands partially submerged and scallions/mushrooms sit at the surface. Slight flatness of the liquid and toppings reading as arranged on top rather than immersed keep it from higher." }, "steam": { "score": 0.85, "reason": "Across the time-ordered desktop frames the translucent wisps visibly change shape, rise and dissipate with drifting gold specks, never obscuring the dish; they are irregular and not mechanical plumes. Slight blobbiness in the upper haze is the only shortcoming." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric staging with the glowing bowl as clear focal point, warm rim light, no text or clutter, and responsive tilt in pointer frames. Mobile framing leaves a lot of empty space above the small bowl, a minor imbalance." }, "vegan": { "score": 1, "reason": "Visible ingredients are tofu, mushrooms, nori, scallions, chili and a leaf in plant broth; no egg, meat, seafood or dairy is depicted." } }, { "noodles": { "score": 0.4, "reason": "Noodles read as thin, straight tan rods laid flat across the broth in a crisscross fan; they lack supple bends, volume, and coherent strand overlap, reading more like rigid straw/sticks than pliable ramen strands." }, "toppings": { "score": 0.65, "reason": "Recognizable flat-style toppings: white tofu cubes, brown mushroom slices with cream undersides, a green leaf, scallion segments, yellow corn-like dots, and a dark nori-like sheet (though the nori block reads slightly book-like with a fold line). Shapes are simple but coherent for the illustration style." }, "bowl": { "score": 0.78, "reason": "Coherent dark ceramic bowl: elliptical rim with a crisp highlight arc, a visible wall thickness, cavity holding the broth, and a smoothly curved body with shading. The rim/body perspective matches; it merely floats without ground contact on the dark backdrop, a minor stylistic issue." }, "broth": { "score": 0.65, "reason": "Warm amber surface with a radial glow and subtle sheen; toppings rest on the surface with mild integration. The liquid reads somewhat flat, with little occlusion or depth where noodles and toppings meet the surface." }, "steam": { "score": 0.72, "reason": "Translucent grey-brown wisps rise above the bowl and clearly change shape across the time-ordered desktop frames (desktop-1 to desktop-3 and pointer frames), showing continuous drift with golden sparkle particles; slightly blobby/diffuse but never hides the food. Reduced-motion frames appropriately freeze it." }, "composition": { "score": 0.85, "reason": "Centered, balanced dish with the glowing amber broth as a clear focal point against a clean dark atmospheric background; no clutter, text, or competing elements, and presentation stays restrained and attractive." }, "vegan": { "score": 1.0, "reason": "Visible ingredients \u2014 tofu cubes, mushrooms, leafy green, scallions, corn-like dots, nori-like seaweed, noodles in plant broth \u2014 are all plausibly plant-based; no animal-derived toppings are visible." } } ], "diagnostics": { "vegan": 0.8666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.92, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50.099999999999454, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "abb06d6ec55b501234a4c409ef29e853a9eed3e1cd165a0d6ae76abeabd91875", "gradedAt": "2026-09-30T23:32:47.603818+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5055555555555555, "toppings": 0.5144444444444445, "bowl": 0.6677777777777778, "broth": 0.5655555555555556, "steam": 0.5688888888888889, "composition": 0.6855555555555556 }, "reward": 0.0, "runId": "xai/grok-4.5/low", "rewardStdDev": 0.066613798481607, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.4666666666666667, "toppings": 0.42666666666666664, "bowl": 0.67, "broth": 0.48, "steam": 0.5066666666666667, "composition": 0.69 }, "reward": 0.527, "sampleStdDev": 0.0034999999999999476, "judgments": [ { "noodles": { "score": 0.48, "reason": "Thin continuous curves suggest noodles, but sparse flat loops and abrupt ends lack convincing overlap, volume, and immersion." }, "toppings": { "score": 0.42, "reason": "Tofu, corn, greens, and nori are recognizable, but mostly appear as plain blocks, circles, and strokes; the nori is a rigid rectangular card with little organic texture." }, "bowl": { "score": 0.66, "reason": "The elliptical rim and curved body form a coherent cavity with readable thickness. Nearly uniform dark walls and a soft shadow provide limited ceramic material and grounding." }, "broth": { "score": 0.46, "reason": "Warm amber shading suggests soup, but the smooth opaque surface has little liquid texture or sheen, and ingredients largely appear laid over it rather than immersed." }, "steam": { "score": 0.52, "reason": "Normal frames show changing plume positions and opacity without obscuring the food. However, the steam consists of broad blurred columns and detached hazy blobs rather than irregular delicate wisps." }, "composition": { "score": 0.7, "reason": "A single centered dish and warm accents remain readable against a restrained dark setting without text or UI. Extensive empty space and visible rectangular background boundaries weaken the presentation." }, "vegan": { "score": 0, "reason": "A clearly depicted egg half with white surrounding a yellow yolk is visible among the otherwise plant-like toppings." } }, { "noodles": { "score": 0.46, "reason": "Thin continuous cream strands are readable, but sparse looping curves look like flat wire scribbles rather than a supple, overlapping noodle mass." }, "toppings": { "score": 0.43, "reason": "Corn, tofu, greens, seaweed and an egg half are recognizable, but mostly rendered as plain circles, rectangles and strokes. The nori is a rigid straight-edged card, with little organic food texture." }, "bowl": { "score": 0.67, "reason": "The elliptical rim, dark inner wall and rounded body form a coherent vessel with visible thickness. Minimal material shading and a gap above the diffuse shadow make its support feel uncertain." }, "broth": { "score": 0.5, "reason": "Warm amber shading suggests soup, but the surface is smooth and nearly solid-looking. Ingredients lack convincing immersion, contact ripples or liquid occlusion." }, "steam": { "score": 0.48, "reason": "Normal frames show changing translucent plumes above the bowl without obscuring the food. Their broad blurred columns and rounded detached patches resemble fog blobs more than irregular delicate wisps." }, "composition": { "score": 0.68, "reason": "The centered dish and restrained warm palette remain readable against the dark background, without text or UI clutter. Excess empty space, visible background bands and very dim bowl details weaken the presentation." }, "vegan": { "score": 0, "reason": "A clearly depicted egg half with white and yellow yolk is visible on the right side of the soup." } }, { "noodles": { "score": 0.46, "reason": "Thin continuous curves suggest noodles, but the sparse flat loops and abrupt upright ends resemble drawn wires rather than a supple, overlapping noodle mass." }, "toppings": { "score": 0.43, "reason": "Corn, tofu, greens, seaweed and an egg half are readable, but mostly reduced to circles, plain blocks and lines. The straight rectangular nori lacks organic edges and texture." }, "bowl": { "score": 0.68, "reason": "The elliptical rim and rounded body form a coherent cavity with visible thickness. Nearly uniform dark walls lack ceramic highlights, and the soft shadow gives weak grounding." }, "broth": { "score": 0.48, "reason": "Warm amber shading suggests soup, but the surface appears like a smooth solid fill. Ingredients have little immersion, contact shading or liquid occlusion." }, "steam": { "score": 0.52, "reason": "Normal frames show changing plume positions and opacity without obscuring the food. However, the steam consists of broad blurred columns and detached hazy blobs rather than delicate irregular wisps." }, "composition": { "score": 0.69, "reason": "The centered dish and warm-on-dark palette provide a clear, uncluttered focal point. Large empty margins, very dark bowl walls and visible background bands weaken the presentation, particularly on mobile." }, "vegan": { "score": 0, "reason": "A clearly depicted egg half with white surrounding a yellow yolk is visibly animal-derived." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5, "bowl": 0.5833333333333334, "broth": 0.6, "steam": 0.55, "composition": 0.6 }, "reward": 0.5441666666666667, "sampleStdDev": 0.003818813079129846, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin, wavy tan strands are visible and curve with some overlap, but they are very fine, small and sparse, so they read as scribbly threads rather than a supple noodle mass." }, "toppings": { "score": 0.5, "reason": "Nori, tofu cubes, corn, scallion strands and a mushroom are recognizable, but they are tiny and simple. The nori is a rigid dark rectangle and the tofu is blocky." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark rounded body and a shadow beneath. It looks like it is floating, the body is very dark and featureless, and the chopsticks are faint and ghostly in the main frames." }, "broth": { "score": 0.6, "reason": "The warm orange broth has a pleasing glow gradient and a faint ripple ring. It reads somewhat flat, and the toppings sit on the surface without much immersion." }, "steam": { "score": 0.5, "reason": "The soft translucent plumes vary across frames and drift upward. They are blurry vertical columns that look detached from the bowl and somewhat mechanical, and they fade out before reaching the broth." }, "composition": { "score": 0.6, "reason": "The bowl is a clear centered focal point on a dark atmospheric background with warm lighting. It is small in the frame, and there are odd vertical banding panels on the sides." }, "vegan": { "score": 0.3, "reason": "A halved egg with a yellow yolk is clearly visible, which is an animal-derived topping. The other ingredients are vegan-appropriate." } }, { "noodles": { "score": 0.5, "reason": "Thin tan strands curve and overlap in the broth and read as noodles, but they are small, sparse and flat, with little volume or mass." }, "toppings": { "score": 0.5, "reason": "Nori, tofu cubes, corn, scallion strands and a mushroom are recognizable, but they are tiny and simplistic. The nori is a rigid dark card and the tofu is blocky. The egg half is a visible animal ingredient." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent elliptical rim, a dark rounded body, a soft shadow and a warm rim highlight. It looks slightly floating, with a thin contact shadow and little material detail." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange with a glow gradient and a faint sheen ring, which looks appetizing. It is fairly flat, and the toppings sit on top more than in it." }, "steam": { "score": 0.55, "reason": "Soft translucent plumes change between frames and fade as they rise. They are blurry vertical columns that look somewhat mechanical and start above the bowl rather than from the soup." }, "composition": { "score": 0.6, "reason": "The bowl is centered in a dark, moody scene with warm lighting and no clutter. It is small in the frame, there are odd vertical bands in the background, and the chopsticks are faint." }, "vegan": { "score": 0.2, "reason": "A halved egg with a yellow yolk is clearly visible, which is not vegan. The other toppings are plant-based." } }, { "noodles": { "score": 0.5, "reason": "Thin tan strands curve and overlap in the broth and read as noodles, but they are small, sparse and somewhat wiry, with little volume." }, "toppings": { "score": 0.5, "reason": "Nori, tofu cubes, corn, scallion shreds and a mushroom are recognizable, but they are small and simple. The nori is a rigid dark card and the tofu is blocky." }, "bowl": { "score": 0.55, "reason": "The bowl has a coherent rim, a round body and a soft shadow. It looks like a flat dark hemisphere that floats, with minimal ceramic detail and weak contact with the ground." }, "broth": { "score": 0.6, "reason": "The warm orange broth has a pleasing gradient and a subtle ripple ring. It is fairly flat, and the toppings sit on the surface with little immersion." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes drift and change across frames. They are blurry, column-like and somewhat detached from the bowl, and they fade out before reaching the soup." }, "composition": { "score": 0.6, "reason": "The dark atmosphere and warm glow focus attention on a centered bowl. The bowl is small in the frame, there are odd vertical bands in the background, and the chopsticks are nearly invisible in the normal frames." }, "vegan": { "score": 0.2, "reason": "A halved egg with a yolk is clearly visible, which is an animal-derived ingredient. The rest of the bowl appears plant-based." } } ], "diagnostics": { "vegan": 0.23333333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.55, "toppings": 0.6166666666666667, "bowl": 0.75, "broth": 0.6166666666666667, "steam": 0.65, "composition": 0.7666666666666667 }, "reward": 0.65, "sampleStdDev": 0.02610076627227633, "judgments": [ { "noodles": { "score": 0.5, "reason": "Pale strands are continuous but read as thin, flat looping lines \u2014 repeated loop-de-loop tangles around the egg and a few crossing curves, more wire-like drawing than supple noodle volume, and the left half of the bowl has no visible noodles." }, "toppings": { "score": 0.55, "reason": "Readable variety: striped nori sheet, white tofu cubes, corn dots, scallion squiggles, a soft egg half and a dark oval, but all are flat, simplified vector shapes \u2014 the tofu are plain featureless cubes and the nori is a rigid rectangle card, lacking organic texture and depth." }, "bowl": { "score": 0.75, "reason": "Coherent stylized ceramic bowl: elliptical rim with a lighter inner edge, smooth dark hemispherical body, plausible high-angle perspective, soft contact shadow beneath. Simple and slightly plain, but no broken joins or distortion." }, "broth": { "score": 0.55, "reason": "Warm glowing amber surface reads as appetizing at first glance, but it is essentially a flat luminous gradient disc \u2014 little liquid texture, restrained sheen or depth, and toppings rest on it with minimal occlusion or contact shading." }, "steam": { "score": 0.65, "reason": "Wisps clearly change across frames: three plumes in desktop-1 rise, merge and fade by desktop-3 and pointer frames, so genuine rising/fading motion is supported. However the plumes are fairly opaque gray, somewhat blob-like and mechanically symmetric, partially reading as fog rather than translucent vapor." }, "composition": { "score": 0.8, "reason": "Balanced centered focal dish in a moody dark scene with warm under-glow, chopsticks adding interest, no text or clutter, and steam stays above without hiding the food; only weakness is the very heavy vignette that swallows most of the frame." }, "vegan": { "score": 0, "reason": "A clearly depicted soft-boiled egg half (white with round yellow yolk) sits on the broth \u2014 an unmistakable animal-derived topping, which zeroes this diagnostic." } }, { "noodles": { "score": 0.6, "reason": "Pale cream strands form readable wavy loops around the egg and across the broth, with coherent placement, but they are single-weight flat strokes with repeated similar loop shapes and no volume or depth shading." }, "toppings": { "score": 0.65, "reason": "Recognizable stylized set: soft-boiled egg half with yolk highlight, nori sheet with grain lines, corn kernels, scallion curls and chili flecks. The white tofu blocks read as flat paper-like slips and the nori is a rather rigid rectangle; overall competent but simplified." }, "bowl": { "score": 0.75, "reason": "Coherent hemisphere bowl with elliptical rim, dark inner cavity and rim thickness, smooth lacquer-like body and a contact shadow beneath. Chopsticks read well in the reduced frames but render ghostly/semi-transparent in the normal frames, a minor inconsistency." }, "broth": { "score": 0.7, "reason": "Warm orange liquid surface with restrained radial sheen; toppings and noodles appear to sit in the broth with soft occlusion rather than floating. Slightly flat and uniform in texture, lacking surface detail like oil or ripple variation." }, "steam": { "score": 0.6, "reason": "Translucent grey wisps clearly animate across the time-ordered frames: plumes rise, spread, drift and fade between desktop-1, -2 and -3 without hiding the food. The plumes are somewhat blob-stacked in shape, slightly mechanical." }, "composition": { "score": 0.75, "reason": "Centered bowl as clear focal point in a dark atmospheric scene with warm glow and soft shadow; no text, UI or clutter. Minor artifacts: a faint vertical band seam on the left background edge and the semi-transparent chopsticks in normal frames." }, "vegan": { "score": 0, "reason": "An unmistakable halved boiled egg with white and round yellow yolk sits in the bowl; egg halves are explicitly animal-derived, so vegan compliance fails." } }, { "noodles": { "score": 0.55, "reason": "Pale thin strands curl across the right side of the broth with consistent line weight, but they read as flat repeated S-loops drawn on the surface rather than strands with volume or believable immersion; no real noodle mass or thickness variation." }, "toppings": { "score": 0.65, "reason": "Recognizable variety: a nori sheet with line detail, white tofu cubes, corn kernels, scallion strips and a well-drawn egg half with yolk. Shapes are flat and simplified \u2014 tofu is featureless squares and scallions are thin scribbles \u2014 but nothing reads as generic plastic discs." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with a clear rim ellipse, cavity, rounded walls and a soft ground shadow; the viewing angle is consistent and chopsticks rest plausibly on the rim in reduced frames. Minor issue: the body is very dark and the bowl sits slightly above its shadow, giving a faint floating feel." }, "broth": { "score": 0.6, "reason": "Warm appetizing orange surface with a soft central glow and a subtle lighter swirl near the rim; toppings cast small contact shadows and sit on the liquid. The surface is fairly flat with limited texture or depth cues." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps clearly change shape, position and count across the time-ordered desktop frames (three tapered plumes in frame 1, broader cloudy shapes in frame 2, narrower fading wisps in frame 3), showing continuous animation. Wisps are translucent and never hide the food, though some mid-rise shapes read as slightly detached blobs." }, "composition": { "score": 0.75, "reason": "Centered focal bowl with warm glow against a restrained dark scene, no text, menus or clutter; mobile framing also stays balanced. Minor deduction for faint rectangular seams in the vignette/background gradient and a large empty lower half." }, "vegan": { "score": 0, "reason": "A clearly visible soft-boiled egg half with yolk sits in the bowl; eggs are unmistakable animal-derived ingredients, so the scene fails the vegan check despite otherwise plant-based toppings." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.07777777777777778, "veganFailed": true }, "performance": { "frameIntervalMedianMs": 116.70000000000073, "frameIntervalP95Ms": 133.29999999999927, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "fbd213d3eff49334ba00ece57a65ee66c40ea01237d96ccc324b9ffc1c912f66", "gradedAt": "2026-09-30T23:31:56.938649+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4633333333333334, "toppings": 0.46555555555555556, "bowl": 0.5166666666666667, "broth": 0.5222222222222223, "steam": 0.5933333333333334, "composition": 0.6755555555555556 }, "reward": 0.5211666666666667, "runId": "xai/grok-4.5/medium", "rewardStdDev": 0.07010052306034052, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.47333333333333333, "toppings": 0.44666666666666666, "bowl": 0.5, "broth": 0.48333333333333334, "steam": 0.5966666666666667, "composition": 0.7266666666666667 }, "reward": 0.5168333333333334, "sampleStdDev": 0.0046188021535169465, "judgments": [ { "noodles": { "score": 0.48, "reason": "Thin yellow strands have continuous curves and some crossings, but remain a sparse, flat arrangement of similar arcs without convincing overlap or noodle volume." }, "toppings": { "score": 0.45, "reason": "Mushroom caps and stems, pale tofu, and corn are recognizable, but largely simplified geometric symbols. Nori looks like a thick rectangular card, and greens are bare curved lines rather than leaves." }, "bowl": { "score": 0.5, "reason": "The elliptical rim and curved walls form a readable cavity with restrained highlights. However, the bowl visibly floats above its separated shadow, lacking credible support or contact." }, "broth": { "score": 0.48, "reason": "Warm orange shading suggests soup, but the surface is mostly smooth and solid-looking, with little liquid sheen or texture. Ingredients lack convincing immersion and contact effects." }, "steam": { "score": 0.6, "reason": "Normal frames show translucent plumes changing position, height, and intensity without obscuring the food. Their broad, blurred columns and detached upper patches lack delicate irregular wisp structure." }, "composition": { "score": 0.73, "reason": "A single centered dish remains readable against a restrained dark, warm background, with no text or clutter. Extensive empty space and subdued contrast make the presentation somewhat underdeveloped." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, corn, greens, seaweed, and red vegetables alongside noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Thin yellow strands have continuous curved paths, but form a sparse, flat set of similar arcs with little overlap depth or convincing noodle volume." }, "toppings": { "score": 0.45, "reason": "Mushroom caps and stems, white tofu-like pieces and corn are recognizable. However, the nori is a rigid rounded card, greens are bare curved strokes, and most ingredients lack organic texture." }, "bowl": { "score": 0.5, "reason": "Elliptical rim and curved walls form a readable cavity with restrained highlights. The conspicuous gap above the separate ground shadow makes the bowl appear unsupported and floating." }, "broth": { "score": 0.48, "reason": "Warm orange shading suggests soup, but the surface is nearly featureless and matte, with little liquid sheen or immersion detail; toppings appear placed on top rather than integrated." }, "steam": { "score": 0.6, "reason": "Normal frames show translucent plumes changing position, height and density without hiding the food. Their broad blurry columns and detached upper patches lack fine irregular wisps and clearly readable emergence." }, "composition": { "score": 0.73, "reason": "A single centered dish remains readable against a restrained dark, warm background without text or clutter. Extensive empty space and subdued contrast make the presentation somewhat underdeveloped, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible mushrooms, tofu-like white pieces, corn, greens, seaweed and noodles plausibly depict vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Thin yellow strands have gentle continuous bends, but form a sparse, flat network with little overlap depth or convincing noodle volume." }, "toppings": { "score": 0.44, "reason": "Mushroom caps and stems, pale tofu, corn and green garnish are recognizable, but minimally textured geometric forms dominate; nori resembles a rigid rounded card and greens are simple arcs." }, "bowl": { "score": 0.5, "reason": "Elliptical rim and curved walls form a readable cavity with restrained highlights, but the bowl visibly floats above a separated shadow without convincing support or contact." }, "broth": { "score": 0.49, "reason": "Warm orange shading gives the surface some depth, but it remains smooth and solid-looking, with little liquid sheen or convincing immersion around the toppings." }, "steam": { "score": 0.59, "reason": "Normal frames show translucent plumes changing position and height without obscuring the food. Their broad blurred columns and detached cloud-like upper portions lack fine irregular wisps." }, "composition": { "score": 0.72, "reason": "A single centered dish remains clear against a restrained dark, warm background, with no distracting text or UI. Generous empty space and subdued contrast make the presentation somewhat underfilled." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, corn, greens and seaweed alongside noodles; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.4, "bowl": 0.4, "broth": 0.5333333333333333, "steam": 0.5666666666666667, "composition": 0.55 }, "reward": 0.45333333333333337, "sampleStdDev": 0.011273124382057216, "judgments": [ { "noodles": { "score": 0.4, "reason": "Thin pale yellow strands form a small tangle, but they are flat, wire-like loops with little volume and sit on the surface rather than in the broth." }, "toppings": { "score": 0.4, "reason": "Shiitake-like mushrooms, scallion curls, tofu cubes and corn are recognizable but tiny and simplistic. The nori is a rigid dark green card and the tofu is a glossy block, so generic forms dominate." }, "bowl": { "score": 0.4, "reason": "The bowl has a believable rim ellipse, but the lower body is a thin wireframe-like dark arc with a transparent look and no solid wall. It floats above its shadow and the broth does not fill the cavity, so the geometry is incoherent." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and a soft edge glow that read as appetizing. It is fairly flat, with little depth, and the toppings sit on top without clear immersion." }, "steam": { "score": 0.55, "reason": "Soft translucent plumes rise and drift differently across frames, so the motion is visible. They look like blurry vertical columns, are somewhat mechanical, and sometimes detach from the bowl." }, "composition": { "score": 0.55, "reason": "The dark, moody scene centers the bowl with a warm glow and no clutter. The bowl is small in the frame and the food is tiny, so the focal dish lacks impact." }, "vegan": { "score": 0.9, "reason": "The toppings are mushrooms, tofu, nori, scallion, corn and chili, with no visible animal products." } }, { "noodles": { "score": 0.4, "reason": "Thin pale strands are visible but sparse and small. They read as flat scribbled curves on the surface, with little volume or overlap, so they look wire-like." }, "toppings": { "score": 0.4, "reason": "Small mushrooms with stems read as shiitake, and there are green scallion curves, red chili bits and corn dots. The tofu cubes are white glossy blobs, and the nori is a rigid dark green rectangle that looks like a card. The toppings are tiny and generic." }, "bowl": { "score": 0.4, "reason": "The rim ellipse is fine, but the body is a thin dark wire-like crescent with a transparent-looking cavity. The bowl looks hollow and detached, it floats over its shadow, and it has no ceramic thickness." }, "broth": { "score": 0.5, "reason": "The orange broth has a warm glow and an edge vignette that give it some depth. It reads as a fairly flat disc with little surface texture, and the toppings sit on it rather than in it." }, "steam": { "score": 0.55, "reason": "Soft translucent plumes rise and drift differently across the frames. They are blurry vertical columns that look somewhat mechanical, and they start well above the bowl rather than from the soup." }, "composition": { "score": 0.5, "reason": "The dark, moody scene has a centered focal bowl and a warm palette with no clutter. The bowl is small in the frame and the food is tiny, so there is a lot of empty space." }, "vegan": { "score": 0.9, "reason": "The toppings are mushrooms, tofu, scallions, nori, corn and chili. No animal products are visible." } }, { "noodles": { "score": 0.4, "reason": "Thin pale strands form a small cluster of loops at the center. They read as noodles but are sparse, flat and wire-like, with little volume or overlap." }, "toppings": { "score": 0.4, "reason": "Shiitake mushrooms with stems, tofu cubes, green scallion curls, chili slices and a nori rectangle are all present and recognizable. The nori is a rigid dark card, the tofu is glossy blocks, and everything is tiny and simplistic." }, "bowl": { "score": 0.4, "reason": "The elliptical rim is coherent, but the underside is a thin dark wire-like arc with a transparent cavity. The bowl floats above its shadow with no solid base or wall thickness, so it looks detached and unsupported." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and soft rim shading, which looks appetizing. The surface is fairly flat and the toppings sit on it without much immersion or ripple." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise from the bowl and change position and shape between frames, drifting and fading. They are somewhat blocky, column-like and blurry, but they do not hide the food." }, "composition": { "score": 0.6, "reason": "A centered bowl sits in a dark, moody scene with warm lighting and a vignette. The food is the clear focus with no clutter, but the bowl is small in the frame and the floating look weakens the presentation." }, "vegan": { "score": 0.9, "reason": "The visible toppings are mushrooms, tofu, scallions, nori, chili and noodles. No animal products are visible." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5166666666666667, "toppings": 0.55, "bowl": 0.65, "broth": 0.55, "steam": 0.6166666666666667, "composition": 0.75 }, "reward": 0.5933333333333334, "sampleStdDev": 0.019094065395649326, "judgments": [ { "noodles": { "score": 0.5, "reason": "Pale-yellow strands curve gently around the center, but they are thin flat lines lying on the broth surface with no volume or immersion; they read as surface squiggles rather than a supple noodle mass." }, "toppings": { "score": 0.55, "reason": "Recognizable variety: tofu cubes, two mushrooms with caps and stems, scallion arcs, red chili ovals, and nori rectangles. But forms are geometric and plastic-like: glossy white cubes, and the nori reads as a rigid dark green card standing upright." }, "bowl": { "score": 0.7, "reason": "Coherent stylized bowl: thick dark rim ring, consistent high-angle ellipse, hemispherical body with a left-side highlight and an under-shadow. Minor issues: the body is nearly silhouette-dark and the bowl appears slightly detached from its shadow ellipse." }, "broth": { "score": 0.55, "reason": "Warm orange radial-gradient surface is appetizing and toppings cast soft contact shadows suggesting immersion, but the liquid is a smooth featureless glow with no oil sheen, ripples, or depth texture, reading somewhat solid." }, "steam": { "score": 0.65, "reason": "Steam is translucent, irregular, and clearly changes shape and position across desktop-1/2/3 and pointer frames, rising without hiding food. Defects: plumes are blobby puffs, and some frames show gaps/disconnected floating masses above the bowl rim." }, "composition": { "score": 0.8, "reason": "Clean centered focal dish with warm glow against a dark vignette, subtle ground reflection, no clutter or UI, consistent lighting across desktop and mobile crops. Slightly sparse and dim overall." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, scallions, chili, and nori in plant broth; no egg, meat, fish, or dairy visible." } }, { "noodles": { "score": 0.5, "reason": "Yellow strands are continuous with plausible bends, but they form a sparse flat cluster painted on the surface with little volume or coherent overlap; they read as thin decorative squiggles rather than a satisfying noodle mass." }, "toppings": { "score": 0.55, "reason": "Good variety (mushrooms with caps and stems, white tofu cubes, green scallion curls, red chili bits), but the dark green nori reads as rigid upright cards/plates and the tofu as plain featureless squares pasted onto the surface, giving a synthetic look." }, "bowl": { "score": 0.6, "reason": "Rim ellipse, inner edge and tapering dark body are coherent and the oblique angle is consistent, but the bowl hovers detached well above its shadow, reading as unsupported, and the material reads as flat dark silhouette with minimal ceramic highlights." }, "broth": { "score": 0.55, "reason": "Warm orange surface with a soft radial glow is appetizing at a glance, but it is essentially a flat gradient disc; toppings sit on top with minimal occlusion or immersion, and there is little liquid texture or sheen." }, "steam": { "score": 0.6, "reason": "Translucent grey wisps clearly change shape and position across the time-ordered frames (plumes shift, spread and fade), so motion is genuine, but several plumes become detached blobs floating above the bowl rather than continuous risers." }, "composition": { "score": 0.7, "reason": "Centered focal dish with warm glow against a restrained dark backdrop; no clutter or competing elements. Desktop is well balanced, though the mobile frame leaves the small bowl low and off-center with excess empty space." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, nori, scallion, chili and noodles in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.55, "reason": "Yellow strands curve across the broth with consistent thickness and a few bends, but they are sparse, flat single-line squiggles with no clumping or volume; a repeated S-curve motif makes them read as decorative rather than a believable noodle mass." }, "toppings": { "score": 0.55, "reason": "Tofu cubes with simple shading, two brown mushroom caps with stems, scallion arcs, red chili slivers and yellow dots are recognizable, but the nori is a rigid upright dark-green rectangle that reads as a standing card, and all pieces are flat vector shapes pasted on the surface." }, "bowl": { "score": 0.65, "reason": "Coherent ellipse rim with thick lacquer-like ring, deep rounded body below, a soft highlight on the flank, and a grounded contact shadow; construction joins cleanly with no broken geometry, though the styling is very simple and the rim band is unusually wide." }, "broth": { "score": 0.55, "reason": "Orange surface with a radial glow and edge darkening gives some depth and toppings rest on it plausibly, but the liquid is essentially a smooth flat gradient with no texture, sheen variation, or visible immersion/occlusion of the noodles and toppings." }, "steam": { "score": 0.6, "reason": "Frames at 1031ms, 2919ms, 4822ms and pointer frames show translucent grey wisps that shift position, merge, rise and fade, so motion is supported; however the plumes are blurry disconnected blobs that sometimes detach from the rim rather than continuous spreading ribbons." }, "composition": { "score": 0.75, "reason": "Centered focal dish on a dark atmospheric backdrop with a warm vignette and glow, balanced margins, no clutter or UI; presentation is clean and attractive even though the scene is minimal and the bowl occupies a modest share of the frame." }, "vegan": { "score": 1, "reason": "Visible ingredients are tofu cubes, mushrooms, scallion, nori, chili slices and yellow vegetable bits in plant broth; no meat, egg, seafood or dairy appears." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9666666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 150.0, "frameIntervalP95Ms": 150, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "837168dca9c72ca15aa4246b838ea0edf41b9bfe75314716cfcdb13c28d27cc2", "gradedAt": "2026-09-30T23:29:40.126836+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5633333333333334, "toppings": 0.5866666666666667, "bowl": 0.7777777777777778, "broth": 0.61, "steam": 0.6933333333333334, "composition": 0.7988888888888889 }, "reward": 0.6597777777777778, "runId": "xai/grok-4.6/high", "rewardStdDev": 0.04522331421013003, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.64, "toppings": 0.5, "bowl": 0.8233333333333333, "broth": 0.6133333333333333, "steam": 0.7133333333333333, "composition": 0.85 }, "reward": 0.6709999999999999, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0.64, "reason": "Golden strands have continuous curves, highlights and readable overlaps, but form a rather flat, densely parallel tangle with limited immersed volume." }, "toppings": { "score": 0.5, "reason": "Tofu, corn, greens and seaweed provide recognizable variety, but smooth oval mushroom caps, blade-like leaves and a rigid rectangular nori sheet dominate over organic texture." }, "bowl": { "score": 0.83, "reason": "Nested elliptical rim, visible inner wall and smoothly shaded rounded body create coherent depth and thickness. The broad side highlight supports a lacquer-like finish, though the base contact is soft and slightly floating." }, "broth": { "score": 0.62, "reason": "Warm reddish broth has subtle tonal variation and scattered seasoning, but little liquid sheen or contact detail; most ingredients appear layered onto its surface rather than partially immersed." }, "steam": { "score": 0.7, "reason": "Normal frames show translucent plumes changing height and shape above the soup without obscuring it. Their narrow, mostly upright columns and surrounding detached hazy spots are less convincing than spreading irregular wisps." }, "composition": { "score": 0.85, "reason": "A centered, clearly readable dish combines warm food colors with a restrained dark background and ample negative space, without text or clutter. Mobile framing retains the whole bowl, though makes the food relatively small." }, "vegan": { "score": 1, "reason": "Visible tofu-like cubes, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } }, { "noodles": { "score": 0.64, "reason": "Golden strands have consistent thickness, curved bends and readable overlaps, but form a rather flat, densely parallel pile with limited submerged volume." }, "toppings": { "score": 0.5, "reason": "Tofu, greens, nori, corn and mushrooms are recognizable and varied, but smooth oval mushroom disks, angular featureless tofu blocks and stiff blade-like leaves dominate; organic texture is sparse." }, "bowl": { "score": 0.82, "reason": "The continuous elliptical rim, visible inner wall and rounded body create coherent depth and perspective, with a warm rim highlight and shaded lacquer-like exterior. The broad angular body highlight and soft, indistinct ground contact slightly weaken the material and support." }, "broth": { "score": 0.61, "reason": "Warm reddish broth has subtle tonal variation and scattered seasoning, but little liquid sheen or meniscus detail; ingredients largely appear laid atop a flat surface rather than immersed." }, "steam": { "score": 0.72, "reason": "Normal frames show translucent plumes changing height and curvature above the soup without obscuring toppings. They remain mostly narrow, parallel blurred columns, with limited irregular spreading and some detached haze." }, "composition": { "score": 0.85, "reason": "A single centered bowl stays clearly readable against a restrained dark background, with warm food colors, complementary greens and no text or UI clutter. Mobile preserves the full dish, though generous empty space makes it relatively small." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.64, "reason": "Golden strands have rounded thickness, continuous bends and readable overlaps, but the pile is shallow and many strands form similarly flattened zigzags with limited immersed volume." }, "toppings": { "score": 0.5, "reason": "Tofu, greens, corn, sprouts and seaweed provide recognizable variety. However, smooth oval mushroom caps lack stems or gills, leaves resemble stiff pointed cards, and the nori is an overly regular rectangular sheet; simplified geometric forms dominate." }, "bowl": { "score": 0.82, "reason": "The continuous elliptical rim, visible inner wall and smoothly shaded rounded body establish coherent depth and thickness. Warm rim highlights suggest dark ceramic, though the broad triangular body highlight feels graphic and ground contact is soft." }, "broth": { "score": 0.61, "reason": "Warm reddish broth has subtle shading and scattered seasoning, with a readable liquid boundary inside the rim. Surface reflections and immersion cues are weak, leaving several toppings looking placed on a flat surface." }, "steam": { "score": 0.72, "reason": "Normal frames show translucent plumes changing height and curvature above the soup while keeping the food visible. The wisps remain rather narrow and parallel, with limited lateral spreading and some detached hazy spots." }, "composition": { "score": 0.85, "reason": "A centered, unobstructed dish, restrained warm brown and green palette, and ample dark negative space create a clear atmospheric focal point without text or clutter. Mobile framing preserves the full bowl, though it makes the food details small." }, "vegan": { "score": 1, "reason": "Visible tofu-like cubes, greens, mushrooms, corn, seaweed, sprouts and noodles plausibly form vegan ramen; no visibly animal-derived topping is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5933333333333333, "bowl": 0.71, "broth": 0.5833333333333334, "steam": 0.6, "composition": 0.7133333333333333 }, "reward": 0.61, "sampleStdDev": 0.011135528725660053, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are a small, tangled mass of short, straight, fairly rigid golden strands. They read as noodles with a bit of overlap, but they look like short fragments rather than supple continuous strands, and the mass is small in the bowl." }, "toppings": { "score": 0.58, "reason": "There is a variety of toppings: a nori sheet, green leaves, shiitake-like mushrooms, tofu cubes, corn kernels, enoki strands and scallion rings. All are recognizable and some have shading. The nori is a stiff, rigid dark slab with a grid, the tofu cubes are blocky, and the mushrooms are glossy ovals." }, "bowl": { "score": 0.68, "reason": "Coherent elliptical rim with a bronze highlight, a rounded dark ceramic body with a curved highlight, and a warm glow beneath. Perspective is consistent. The base is a pointed bottom with no foot, so the bowl seems to float and the contact with a surface is unclear." }, "broth": { "score": 0.55, "reason": "The red-orange broth has a gradient and small fleck details, so it reads as liquid. It looks fairly flat and opaque, and the toppings sit on it without much immersion, ripple or sheen." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise from the bowl and vary between frames, and they respond slightly to the pointer. They are thin, somewhat streaky and uniform, and there are faint bokeh blobs, but they do not hide the food." }, "composition": { "score": 0.72, "reason": "The bowl is centered as a clear focal point in a dark, atmospheric scene with a warm glow and good color contrast of red, green and gold. The food is small within the bowl, and the dark, empty surroundings are a bit heavy." }, "vegan": { "score": 0.93, "reason": "Only plant-based ingredients are visible: tofu, mushrooms, nori, greens, corn, enoki and noodles. No egg, meat or fish." } }, { "noodles": { "score": 0.5, "reason": "Noodles are a small cluster of short, thin, straight-ish golden strands. They read as noodles, but they look like short cut pieces and a scribbly tangle with little volume. They are not supple long strands and are small for the bowl." }, "toppings": { "score": 0.6, "reason": "Toppings are varied and recognizable: a nori sheet with a grid texture, green leaves, shiitake-like mushrooms, tofu cubes, corn kernels, enoki strands and scallion rings. The shapes are simple and somewhat vector-like. The nori is a stiff rectangle and the tofu blocks are plain, but the variety and scale are reasonable." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim with a copper bevel and a rounded dark body. Its glossy highlight swoosh and warm glow underneath give it depth. The base tapers to a point with no foot, so it looks floating and slightly odd. The rim perspective is consistent with the body." }, "broth": { "score": 0.6, "reason": "The red-orange broth has a soft gradient and speckles of oil and pepper, and it reads as a spicy liquid. The surface is fairly flat, with little sheen or ripple. The toppings sit on it acceptably but have little immersion or contact detail." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise above the bowl and change shape between frames. They drift and fade, and the pointer frames show a response. They are somewhat thin and streaky, with faint bokeh blobs in the background. They do not obscure the food." }, "composition": { "score": 0.72, "reason": "The bowl is centered in a dark, atmospheric scene with warm under-glow, and it is the clear focal point. The palette of red, green and gold is attractive. The food is small relative to the bowl and there is a lot of empty dark space, but the scene has no clutter or text." }, "vegan": { "score": 0.93, "reason": "Visible ingredients are nori, leaves, mushrooms, tofu, corn, enoki and scallions in a red broth. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are a small cluster of short, thin golden strands that read as noodles with some overlap, but they look like short stubby bits and are tiny relative to the bowl, with no long supple bends." }, "toppings": { "score": 0.6, "reason": "Nori, bok choy leaves, shiitake, tofu cubes, corn and enoki are all recognizable and varied. The nori is a stiff dark rectangle with a grid, the tofu cubes are blocky, and everything is small and clustered on one side." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim with a copper lip, a dark glazed body with a soft highlight, and a warm glow beneath. The base is a pointed bottom with no visible foot, so it looks slightly unsupported, but there is no material distortion." }, "broth": { "score": 0.6, "reason": "The red-orange broth has a subtle gradient and speckles of oil, and the toppings sit in it reasonably. It is flat and fairly uniform, with little depth or sheen, so it reads a bit like a painted disc." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise from the bowl and change shape across frames. They are thin, somewhat mechanical vertical streaks, and the faint bokeh haze above is hazy. Motion is visible but not very organic." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with a warm glow and good color contrast. The food is small within the bowl and the lower area is empty, but there is no clutter." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, greens, nori, corn and noodles in a red broth. No visible meat, egg or fish." } } ], "diagnostics": { "vegan": 0.92 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.55, "toppings": 0.6666666666666666, "bowl": 0.8, "broth": 0.6333333333333333, "steam": 0.7666666666666667, "composition": 0.8333333333333334 }, "reward": 0.6983333333333334, "sampleStdDev": 0.035472994422987916, "judgments": [ { "noodles": { "score": 0.6, "reason": "Yellow strands form a recognizable central nest with some draping and wavy texture, but the mass is fairly flat and drawn-looking, with repeated little spiral curl tips (also on the white sprout strands) that read as a motif rather than supple individual noodles." }, "toppings": { "score": 0.7, "reason": "Good variety: upright dark nori sheet with grid texture, pointed green scallion leaves, brown shiitake domes, speckled tofu cubes, corn kernels, onion slice and scattered scallion rings. All are readable, but shapes are simple flat-vector forms; the mushrooms and tofu are somewhat stiff and the nori looks like a rigid card." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with a lighter glazed band, smooth hemispherical body, soft highlight streak, and a warm under-glow grounding it. Rim, cavity and body perspectives agree; only minor simplification of the inner wall and lack of a distinct foot." }, "broth": { "score": 0.7, "reason": "Rich red-orange liquid surface with a subtle radial gradient, seasoning specks and gentle sheen; the nori and toppings sit at the surface with plausible contact. The surface is somewhat flat with a faint ring-like edge shading rather than true liquid depth." }, "steam": { "score": 0.8, "reason": "Across desktop-1 to desktop-3 and the pointer frames, translucent grey wisps clearly change height, spread and fade while staying thin enough not to hide the food. Motion is continuous and non-static; wisps are slightly squiggle-like but irregular and believable." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a dark atmospheric field with a warm under-glow and soft bokeh; balanced proportions, no text, UI or clutter, and the food remains the focus in both desktop and mobile crops." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are nori, greens, mushrooms, tofu, corn, sprouts, scallion and a red chili/tomato-like element in spicy-looking broth; no egg, meat, fish or dairy appears." } }, { "noodles": { "score": 0.45, "reason": "Noodles read as a yellow scribbled tangle with flat overlapping loops and some broken white strands; coherent strand volume and believable immersion are only partly achieved." }, "toppings": { "score": 0.65, "reason": "Recognizable tofu cubes, mushrooms, greens, nori sheet, corn and scallion rings with decent variety, but shapes are flat and simplified; nori floats rather high and greens are stiff leaf cutouts." }, "bowl": { "score": 0.8, "reason": "Coherent rim, thick wall and rounded base with convincing ceramic/lacquer highlights and perspective; no distortion or broken joins observed, only minor stiffness in the rim highlight." }, "broth": { "score": 0.55, "reason": "Red broth surface is fairly flat with a uniform sheen and speckles; toppings appear pasted on top rather than sitting in the liquid, with limited depth or contact occlusion." }, "steam": { "score": 0.75, "reason": "Translucent irregular wisps rise above the bowl and visibly shift shape and position across the time-ordered desktop frames, without hiding the food; slightly blobby edges at top." }, "composition": { "score": 0.8, "reason": "Balanced centered dish with warm rim lighting against a restrained dark scene; the food is the clear focal point with no clutter or text, though the bowl fills a modest portion of frame." }, "vegan": { "score": 0.9, "reason": "Visible ingredients (tofu, mushrooms, greens, nori, corn, scallions, noodle broth) are plausibly plant-based; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.6, "reason": "Central nest of tan wavy strands reads as noodles with some overlap, but the mass is somewhat flat and scribbly, made of many short S-curves rather than long continuous strands with volume; the thin white arcs nearby read more like sprouts than noodles." }, "toppings": { "score": 0.65, "reason": "Good variety in a stylized flat-illustration idiom: upright nori sheet, green leaf/leek shapes, shiitake-like brown caps, tofu cubes, corn kernels, sprouts and scallion rings. Shapes are simplified\u2014mushrooms verge on glossy ellipses and tofu is clean geometry\u2014but all remain recognizable and appropriately scaled." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic hemisphere with a copper rim ellipse, consistent rim/body perspective, inner shadow giving cavity depth, and a soft highlight arc on the wall; contact with the surface is suggested by a warm under-glow. Minor stylization only, no broken joins or distortion." }, "broth": { "score": 0.65, "reason": "Red-orange miso-toned surface with radial darkening toward the rim and scattered chili specks reads appetizingly; toppings sit at the surface with small shadows. However the liquid reads fairly flat/solid with limited sheen and depth, and the noodle nest floats on top without strong immersion cues." }, "steam": { "score": 0.75, "reason": "Across desktop-1/2/3 the translucent gray wisps change shape, height and lean between frames, indicating continuous rising animation; they stay irregular, fade upward and never hide the food. They are somewhat discrete vertical plumes rather than fully spreading wisps." }, "composition": { "score": 0.85, "reason": "Centered, well-proportioned focal bowl in a dark atmospheric scene with restrained warm under-glow and faint drifting particles; no text, UI or clutter, and the mobile viewport crops cleanly without overflow." }, "vegan": { "score": 1.0, "reason": "Visible toppings\u2014nori, mushrooms, tofu, corn, sprouts, scallion, greens\u2014in plant broth are all plausibly vegan; no meat, fish or egg forms appear." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9622222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8d121951b2f61428b702ab1a9b6bbb469dea8bd27f38f2853ebd8202bd8afc05", "gradedAt": "2026-09-30T23:30:18.988893+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5433333333333333, "toppings": 0.48333333333333334, "bowl": 0.7366666666666667, "broth": 0.5511111111111111, "steam": 0.44, "composition": 0.7088888888888889 }, "reward": 0.5688333333333333, "runId": "xai/grok-4.6/low", "rewardStdDev": 0.056464590674156136, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.49666666666666665, "toppings": 0.45, "bowl": 0.7633333333333333, "broth": 0.4533333333333333, "steam": 0.42, "composition": 0.72 }, "reward": 0.5448333333333333, "sampleStdDev": 0.008621678104251716, "judgments": [ { "noodles": { "score": 0.49, "reason": "Continuous golden curves read as noodles, but mostly form flat, similarly thick crossing lines without convincing overlap, immersion, or piled volume." }, "toppings": { "score": 0.46, "reason": "Mushroom caps, faceted tofu cubes, and a dotted yellow garnish are recognizable, but remain simple geometric symbols with little organic texture; green details are sparse." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, rounded dark walls, and contact shadow form a coherent supported bowl. Rim thickness is readable, though the material shading and cavity depth are rather basic." }, "broth": { "score": 0.45, "reason": "Warm reddish color and a soft highlight suggest soup, but the broad surface looks nearly solid and ingredients appear laid on top, with little contact or immersion." }, "steam": { "score": 0.43, "reason": "Pale translucent haze changes between normal frames and leaves the food readable, but stays close to the surface as blurred patches rather than clearly rising, spreading wisps." }, "composition": { "score": 0.72, "reason": "A centered single bowl and restrained warm-on-dark palette keep the dish readable without text or UI. Detached chopsticks and extensive empty space, particularly on mobile, weaken the arrangement." }, "vegan": { "score": 0.97, "reason": "Visible mushrooms, tofu-like cubes, green garnishes, and noodles plausibly form vegan ramen; no material animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have smooth bends, but form a flat tangle of similarly weighted curves with little overlap shading, volume, or submerged depth." }, "toppings": { "score": 0.46, "reason": "Mushroom caps, tofu cubes and a dotted yellow vegetable piece are recognizable, but mostly use simple disks and featureless geometric forms. Organic texture and leafy detail are sparse." }, "bowl": { "score": 0.77, "reason": "The elliptical rim, rounded walls and grounded shadow form a coherent vessel. Rim thickness is readable, though the nearly black body has limited ceramic highlights and depth definition." }, "broth": { "score": 0.47, "reason": "Warm orange shading suggests soup, but the surface is largely a smooth solid-colored ellipse with outlined rings. Ingredients appear laid across it rather than convincingly immersed." }, "steam": { "score": 0.43, "reason": "Pale translucent haze changes position across normal frames without obscuring the food. It remains mostly low diffuse patches, lacking clearly rising, spreading wisps and readable fade trajectories." }, "composition": { "score": 0.72, "reason": "The single bowl is clearly framed against a restrained dark, warm background without text or UI. Detached chopsticks feel awkward, and the mobile dish occupies a relatively small portion of the frame." }, "vegan": { "score": 0.96, "reason": "Visible mushroom, tofu-like cubes, vegetable garnish and noodles plausibly form vegan ramen; no material animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands have curved bends, but form a flat tangle with little thickness variation, coherent overlap, or submerged volume." }, "toppings": { "score": 0.43, "reason": "Mushroom, tofu cubes and a dotted corn-like oval are recognizable, but mostly use smooth generic shapes with minimal organic texture. Green garnishes in reduced frames are sparse schematic strokes." }, "bowl": { "score": 0.76, "reason": "Elliptical rim, connected rounded walls and contact shadow establish coherent depth and support. Restrained edge highlights help, though the nearly black body has limited ceramic shading and base definition." }, "broth": { "score": 0.44, "reason": "Warm orange shading suggests soup, but the broad surface looks solid and largely textureless. Ingredients appear laid over it rather than immersed, and thin elliptical rings feel diagrammatic." }, "steam": { "score": 0.4, "reason": "Soft translucent haze visibly changes across normal frames without hiding the noodles. It remains low over the surface, with rounded patches rather than clearly rising, spreading irregular wisps." }, "composition": { "score": 0.72, "reason": "Centered dish, warm restrained palette and uncluttered dark background give a clear focal point. Floating chopsticks weaken spatial coherence, and the mobile dish is small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible noodles, mushroom, tofu-like cubes, corn-like garnish and green seaweed or vegetables plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 0.9766666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6, "toppings": 0.4, "bowl": 0.68, "broth": 0.5833333333333334, "steam": 0.3333333333333333, "composition": 0.64 }, "reward": 0.5283333333333333, "sampleStdDev": 0.010016652800877822, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles are supple, continuous, overlapping curved strands with a believable golden color and a coherent tangle. They look thin and a bit uniform, and they sit as a small flat cluster on top of the broth with little volume." }, "toppings": { "score": 0.4, "reason": "Toppings are sparse and small: a shiitake cap with a thin dark stem, two white tofu cubes, a yellow corn-like blob, and a few specks. The tofu cubes are plain blocks. The nori is hard to read, and the scallion bits appear only in the reduced frames. Variety is limited and the details are crude." }, "bowl": { "score": 0.68, "reason": "The dark bowl has a coherent elliptical rim, a tapering body, a ground shadow and a subtle rim highlight. The perspective is consistent and it looks grounded. The material is fairly flat and matte, with little ceramic or lacquer character and little visible wall thickness." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange-red gradient with a soft highlight and faint ripple rings, which reads as an appetizing liquid. Ingredients sit on the surface with limited immersion, and the dark shadow shapes drifting across it look odd." }, "steam": { "score": 0.3, "reason": "The steam is barely visible in the desktop frames, appearing as faint blurry blobs over the broth. Dark elongated shapes move across the surface, which reads more like shadows than wisps. The reduced-motion frames show a faint rectangular column. Steam rising above the bowl is weak." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark atmospheric scene with a warm glow, so the dish is the clear focus. The chopsticks float detached in the upper right, and the bowl is small, leaving a lot of empty space. Colors are cohesive." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are mushroom, tofu, corn, noodles and a plant-looking broth, with no meat, egg or fish. The appearance is plausibly vegan." } }, { "noodles": { "score": 0.6, "reason": "Golden noodles are supple, overlapping and curved, with believable thin strands. The mass is small and flat-looking, sitting on the surface like a thin tangle, and it lacks volume." }, "toppings": { "score": 0.4, "reason": "A shiitake, two tofu cubes and a yellow corn-like leaf are all tiny and sparse. The tofu cubes are plain blocks, the yellow item is ambiguous, and there is no readable nori or greens in the normal frames. The mushroom is small but recognizable." }, "bowl": { "score": 0.68, "reason": "The dark bowl has a coherent elliptical rim, inner wall, tapered body and a contact shadow. Perspective is consistent. The bowl is very dark and plain, with little material highlight or detail, and the base reads weakly." }, "broth": { "score": 0.6, "reason": "The orange-red broth has a pleasing gradient, a sheen and subtle ripple rings. It reads slightly flat and opaque, and the ingredients look somewhat pasted on the surface, with little immersion or occlusion." }, "steam": { "score": 0.4, "reason": "Faint translucent wisps and dark elongated blobs drift across the broth surface across frames. The motion is visible, but the steam looks like smudges on the soup and barely rises above the bowl. In the reduced frames a tall faint column appears." }, "composition": { "score": 0.62, "reason": "The bowl is centered in a dark, atmospheric scene with a warm glow and a chopstick accent, and the food is the focus. The bowl is small, the noodles and toppings occupy a small area, and the chopsticks float detached in the upper right." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are noodles, mushroom, tofu and a corn-like item, with no meat, egg or fish." } }, { "noodles": { "score": 0.6, "reason": "Noodles are supple, overlapping, continuous wavy strands with believable thickness. But they form a small, sparse clump in a huge broth area, and the looping is somewhat repetitive and flat." }, "toppings": { "score": 0.4, "reason": "A shiitake, two tofu cubes and a yellow corn-like leaf are recognizable, but they are tiny, simplistic and sparse. The tofu reads as blocky white cubes. A dark nori slice and scallions appear only in the reduced frames. Overall the toppings look generic and lack texture." }, "bowl": { "score": 0.68, "reason": "The rim is a coherent ellipse with a wall, a base shadow and a subtle rim highlight. The matte dark ceramic has little detail, and the body below the rim is nearly lost in the dark. The shadow is slightly detached from the base, but there is no distortion." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a warm gradient, a soft sheen and a faint reflection. It reads as a flat, smooth surface with little depth. Ingredients sit on top rather than being immersed, and the thin concentric ring looks like a target." }, "steam": { "score": 0.3, "reason": "The desktop frames show only faint, blurry blobs and dark elongated shadows drifting over the broth. There are barely any visible wisps rising above the bowl. The reduced frame shows a static, rectangular-looking haze." }, "composition": { "score": 0.62, "reason": "The bowl is a centered focal point in a dark, atmospheric scene with a warm glow and no clutter. The food is small within the bowl, the chopsticks float disconnected in the upper right, and the overall scene is sparse." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are noodles, mushroom, tofu, corn and scallions, with no meat, egg or fish." } } ], "diagnostics": { "vegan": 0.8833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5333333333333333, "toppings": 0.6, "bowl": 0.7666666666666667, "broth": 0.6166666666666667, "steam": 0.5666666666666667, "composition": 0.7666666666666667 }, "reward": 0.6333333333333333, "sampleStdDev": 0.07285830998131465, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are flat scribbled yellow strands clustered on the surface, reading as a tangled spaghetti-like mass rather than supple ramen strands emerging from broth; flat loops overlap without volume or believable immersion." }, "toppings": { "score": 0.5, "reason": "Recognizable mushroom (cap with stem) and tofu cubes; however the corn is a glossy yellow disk with dotted seeds, a rigid scallion-green streak floats oddly, and shapes sit pasted on the broth surface with little texture or variety." }, "bowl": { "score": 0.75, "reason": "Coherent black ceramic bowl with a clear rim, dark walls, elliptical opening, soft floor shadow and support; stylized and clean, though the rim/wall reads slightly flat and lacks highlights that would give material depth." }, "broth": { "score": 0.55, "reason": "Reddish-brown liquid with a glossy sheen and a faint rotating swirl shows liquid character, but it reads as a hard glossy shell; toppings and noodles appear pasted on top rather than submerged, with weak contact and occlusion." }, "steam": { "score": 0.4, "reason": "Frames show a rotating swirling band and small drifting specks rather than translucent wisps rising and fading from the soup; the bright blobs and swirl remain largely similar in position, reading more like a rotating overlay than steam; small particle movement shows some temporal change." }, "composition": { "score": 0.7, "reason": "Balanced centered bowl with dark atmospheric background, floating chopsticks, and soft glow; readable and attractive, though the mostly empty frame, static chopsticks at odd angle, and small clipped yellow sliver top-left slightly distract." }, "vegan": { "score": 0.9, "reason": "Visible toppings (mushroom, tofu, corn, greens, chili) are all plant-based; no animal-derived ingredients apparent." } }, { "bowl": { "reason": "Dark donburi reads coherently: thick elliptical rim, tapered glossy walls, believable cavity and contact shadow; no broken joins, though the foot is hidden and materials are very plain.", "score": 0.75 }, "broth": { "reason": "Warm miso-toned surface with a soft sheen and drifting dark leaf shadows adds depth, but faint concentric ring lines read as flat target circles and noodles/toppings appear pasted on top rather than immersed.", "score": 0.65 }, "composition": { "reason": "Well-balanced centered dish on an atmospheric dark backdrop with restrained amber glow and floating chopsticks; food stays the clear focal point with no clutter, though the scene is sparse and the topping cluster occupies little of the bowl.", "score": 0.8 }, "noodles": { "reason": "Yellow strands form a tangled pile with coherent bends and a darker under-layer hinting volume, but they read as flat uniform scribbly strokes lying on the surface with little thickness variation or believable immersion.", "score": 0.55 }, "steam": { "reason": "Translucent wisps shift across the desktop frames and pointer-left shows haze rising above the rim without hiding food, but the effect is faint, dotted and mostly hugs the surface rather than forming defined rising plumes.", "score": 0.65 }, "toppings": { "reason": "Mushroom cap and stem, tofu cubes, a corn slice with kernel dots and chili flecks are recognizable with organic-ish shapes, but the set is sparse, cubes are plain flat blocks, and ambiguous dark drifting leaves read more as shadows than ingredients.", "score": 0.65 }, "vegan": { "reason": "Visible ingredients (mushroom, tofu, corn, chili, noodles, scallion arcs) are all plausibly plant-based; no animal-derived toppings appear.", "score": 1 } }, { "vegan": { "score": 1, "reason": "Toppings are clearly plant-based: mushroom, white tofu cubes, corn, scallion rings, seaweed strip, chili flecks over noodles in broth; no animal-derived items visible." }, "toppings": { "score": 0.65, "reason": "Recognizable variety: a brown mushroom cap with stem, shaded tofu cubes, a dotted corn kernel, green scallion arcs, a seaweed sliver and red chili flecks. Shapes are simple flat vector forms with limited texture, and topping density is sparse relative to the bowl, but nothing reads as generic plastic blocks." }, "steam": { "score": 0.65, "reason": "Translucent wisps drift and change shape across desktop-1/2/3 (shifting lens-shaped smears) and clear plumes rise above the rim in pointer-left and mobile frames, so motion is supported and not static. Wisps are faint and sometimes read as smudgy blobs rather than defined rising vapor." }, "noodles": { "score": 0.6, "reason": "Thin yellow strands form a continuous tangled nest with smooth varied bends and coherent overlaps, sitting with a soft shadow on the broth. However strands are uniformly flat strokes with little volume or thickness variation, reading somewhat scribbly in the flat style." }, "composition": { "score": 0.8, "reason": "Bowl is the clear focal point, centered with warm under-glow on a dark atmospheric ground, balanced by floating chopsticks at upper right and sparse dust motes; no text, UI or clutter. Presentation is clean, with the food itself carrying the scene." }, "broth": { "score": 0.65, "reason": "Warm red-orange gradient broth with restrained sheen and believable contact shadows under the mushroom and noodle nest. Defects: a faint thin concentric circle ring around the toppings, plus pale horizontal smears and the drifting dark lens shapes that muddy the surface rather than enriching it." }, "bowl": { "score": 0.8, "reason": "Coherent high-angle construction: elliptical opening with a gray rim ring and inner highlight arc, smoothly curved dark ceramic body with gradient shading, and a soft grounded contact shadow beneath. Rim, body and shadow perspective agree; only minor simplicity keeps it from higher." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9422222222222222, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.35000000000036, "frameIntervalP95Ms": 83.40000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "9b2f10310e835bbbba539afb0bc7f3dd41450eba374e61b4bd23db8b93cec382", "gradedAt": "2026-09-30T23:30:04.867557+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6077777777777778, "toppings": 0.6, "bowl": 0.7977777777777778, "broth": 0.6522222222222221, "steam": 0.6611111111111111, "composition": 0.7977777777777778 }, "reward": 0.6752777777777778, "runId": "xai/grok-4.6/medium", "rewardStdDev": 0.08446865515505561, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5499999999999999, "toppings": 0.5, "bowl": 0.8633333333333333, "broth": 0.5733333333333333, "steam": 0.6, "composition": 0.85 }, "reward": 0.64, "sampleStdDev": 0.009643650760992963, "judgments": [ { "noodles": { "score": 0.57, "reason": "Continuous yellow strands have curved bends and readable overlaps, but mostly form a flat crisscrossed layer with repetitive parallel curves and little dimensional volume." }, "toppings": { "score": 0.5, "reason": "Tofu, mushrooms, greens, corn and scallions are recognizable, but simplified blocks and smooth ovals dominate. The nori resembles a thick rounded card rather than thin irregular seaweed." }, "bowl": { "score": 0.86, "reason": "The elliptical layered rim joins coherently to shaded curved walls, with convincing thickness, restrained highlights and a grounding contact shadow. The dark base is somewhat indistinct, but no material construction defect is observed." }, "broth": { "score": 0.58, "reason": "Warm reddish broth has a restrained highlight and visible surface detail, but reads fairly flat; most ingredients appear laid on top without convincing immersion or liquid contact." }, "steam": { "score": 0.6, "reason": "Normal frames show translucent vapor changing position, height and density above the soup. However, much of it consists of soft oval blobs rather than connected irregular wisps, with some haze washing across the food." }, "composition": { "score": 0.85, "reason": "A single clearly framed bowl remains the focus against a restrained dark background, with warm food colors and coherent shading. No text or UI distracts; the mobile dish is slightly small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, seaweed, scallions and corn alongside noodles; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.5, "reason": "Continuous yellow strands have readable crossings and curved ends, but mostly form flat, parallel ribbons with little supple volume or convincing submersion." }, "toppings": { "score": 0.5, "reason": "Tofu, mushrooms, greens, corn and scallions provide recognizable variety. However, smooth oval mushroom caps, spherical corn and especially the thick rounded nori card make much of the food look geometric and artificial." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, consistent wall curvature, layered lip highlights and grounded contact shadow create a convincing dark ceramic vessel. The nearly black lower wall slightly obscures its base." }, "broth": { "score": 0.57, "reason": "Warm reddish-brown color and restrained surface highlights suggest soup, but the surface reads fairly flat, with limited immersion cues and ingredients appearing laid atop it." }, "steam": { "score": 0.6, "reason": "Normal frames show translucent vapor changing height, position and opacity above the soup. The plumes remain noticeably oval and blob-like rather than irregular wisps, with some detached upper patches." }, "composition": { "score": 0.85, "reason": "A single clearly framed bowl remains the focal point against restrained warm-dark lighting, with no text or UI clutter. Desktop balance is strong; the mobile dish is somewhat small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed, corn and noodles plausibly form vegan ramen; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.58, "reason": "Continuous golden strands have readable overlaps and curved ends, but mostly form flat, similarly sweeping ribbons with little rounded volume or submerged depth." }, "toppings": { "score": 0.5, "reason": "Tofu, mushrooms, greens, corn and scallions are recognizable, but simplified forms dominate: nori resembles a rigid rounded card, mushroom caps are smooth ovals, and corn is a cluster of uniform disks." }, "bowl": { "score": 0.87, "reason": "The coherent elliptical rim, layered ceramic lip, shaded curved walls and grounded contact shadow convincingly establish a deep vessel. No material construction defect is observed, though the dark base loses some definition." }, "broth": { "score": 0.57, "reason": "Warm reddish broth has restrained highlights and subtle surface flecks, but reads rather flat; noodles and front tofu appear laid on top with little liquid contact or immersion." }, "steam": { "score": 0.6, "reason": "Normal frames show changing plume positions, height and opacity, supporting motion and fading. However, the steam consists largely of soft oval puffs rather than irregular connected wisps, with some floating detached above the bowl." }, "composition": { "score": 0.85, "reason": "A single uncluttered dish remains the clear focus against a restrained dark backdrop, with warm food colors and coherent bowl shading. Mobile framing preserves the complete bowl; the desktop placement is somewhat low and leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, corn, scallions and noodles plausibly constitute vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6066666666666667, "toppings": 0.55, "bowl": 0.68, "broth": 0.6, "steam": 0.6, "composition": 0.6933333333333334 }, "reward": 0.6141666666666667, "sampleStdDev": 0.0030550504633038962, "judgments": [ { "noodles": { "score": 0.6, "reason": "Fine golden strands overlap with some curvature and volume, and they read as noodles sitting in the broth. They are thin and somewhat scribbly, and the mass is flat and small within the wide bowl." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, enoki, shiitake, bok choy, scallion rings and corn are all recognizable and vegan-appropriate. The nori is a tall rounded gravestone-like slab standing upright, which looks rigid and unnatural. The mushrooms are smooth dark ovals with little texture." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body with highlights, and a contact shadow. The rim and body perspective match. The dark body is murky and low in detail, and the rim looks somewhat thick." }, "broth": { "score": 0.6, "reason": "The red-orange broth has a soft glow and a sheen near the rim, and the noodles sit in it plausibly. It looks fairly flat, with little surface texture or ripple, and the toppings sit on it without much immersion." }, "steam": { "score": 0.6, "reason": "Across the frames, translucent wisps rise, spread and fade, and they change between frames. Some puffs look like blurry blobs, and in places the steam is detached from the bowl." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with a warm glow, and the food is the focus. Colors are attractive. The bowl sits low, leaving a large empty area above, and the dark lower bowl is murky." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, greens, corn, nori and noodles are visible, with no meat, egg or fish." } }, { "noodles": { "score": 0.62, "reason": "Noodles are thin, wavy, overlapping golden strands with a believable tangle and volume. They look a bit flat and uniform, and they sit on the broth like a mat rather than being clearly immersed." }, "toppings": { "score": 0.55, "reason": "Toppings are varied and recognizable: tofu cubes, enoki, shiitake, bok choy, corn, scallion rings and nori. The nori is a rigid upright rounded-top slab that looks like a tombstone, and the tofu is blocky. Overall they read as symbols rather than detailed food." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a curved body with a soft highlight, and a ground shadow. The dark body is somewhat muddy and low in detail, and the base contact is unclear, but the perspective is consistent and no major distortion is visible." }, "broth": { "score": 0.6, "reason": "The broth is a rich red-orange with a glowing highlight and some sheen. It looks fairly flat and has limited ripple or depth. Toppings sit on it with minimal contact cues, though the nori has some occlusion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise and change between frames, spreading and fading. The plumes are blobby and blur-like, with some disconnected puffs, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "A centered bowl in a dark atmospheric scene with a warm glow keeps the food as the clear focus. There is no clutter or text. The bowl's lower half is very dark and the top of the frame is empty." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, vegetables, corn, nori and noodles are visible, with no animal products. The broth is red and plausibly plant-based." } }, { "noodles": { "score": 0.6, "reason": "Thin golden strands overlap with some curvature and read as noodles in broth. They are fairly uniform and flat, with limited volume, and several run in parallel." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake, bok choy leaves, enoki, scallion rings and corn are all recognizable. The nori is a rounded upright tombstone-like card, and the tofu is blocky. The bok choy is decent, but several items look generic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, curved walls, a highlight and a ground shadow. It is very dark and low in detail, but the perspective is consistent and no distortion is visible." }, "broth": { "score": 0.6, "reason": "The red-orange broth has a warm glow and a sheen near the rim, and the noodles sit plausibly in it. It is somewhat flat, with limited depth and little contact or occlusion around the toppings." }, "steam": { "score": 0.6, "reason": "Soft translucent plumes rise and fade, and they change between frames. Some are blobby ovals, and the lower steam over the rim is a bit hazy and mechanical. Overall the motion is continuous and does not hide the food." }, "composition": { "score": 0.7, "reason": "The bowl is a central focal point in a dark, moody scene with warm glow and no distracting UI. The bowl body is very dark and the upper half of the frame is empty, but the color palette is appealing." }, "vegan": { "score": 0.93, "reason": "The visible toppings are tofu, mushrooms, bok choy, enoki, corn, scallions and nori, all plant-based. No animal products are visible." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.75, "bowl": 0.85, "broth": 0.7833333333333333, "steam": 0.7833333333333333, "composition": 0.85 }, "reward": 0.7716666666666667, "sampleStdDev": 0.003818813079129846, "judgments": [ { "noodles": { "score": 0.7, "reason": "Yellow strands are continuous with plausible bends and bundle into a voluminous clump over the right rim, but they read somewhat flat and uniform, like a repetitive wavy mat laid on the surface rather than supple piled noodles." }, "toppings": { "score": 0.75, "reason": "Good variety and mostly readable: speckled nori sheet, stacked tofu blocks, enoki stalks with caps, scallion rings, dark shiitake ellipses, corn dots and leafy greens. Weaknesses: the corn is abstract dot clusters and the green leaves are somewhat blobby and amorphous." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: consistent elliptical rim with cream inner band, smooth tapering body, subtle streak highlights, and a soft grounded contact shadow. No broken joins or perspective clashes; minor shortcoming is the very plain, smooth body." }, "broth": { "score": 0.8, "reason": "Appetizing red-brown miso-toned surface with a warm central sheen and speckled seasoning; toppings overlap and sit into the liquid with believable occlusion. Slight haziness from steam veils part of the surface, and the liquid is fairly smooth-textured." }, "steam": { "score": 0.75, "reason": "Time-ordered frames show clear animated steam: a tall plume in desktop-1, lower wisps in desktop-2, a low surface haze in desktop-3, and tall plumes again in pointer frames, so it rises and dissipates. Deduction: in some frames it is a fairly dense, blobby gray cloud of overlapping round puffs rather than fine translucent wisps." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric backdrop with a warm glow, soft shadow and faint specks; no text, UI or competing elements, and the food stays the focal point. Steam haze slightly veils the bowl at times but remains restrained." }, "vegan": { "score": 0.5, "reason": "Plant-friendly toppings dominate (nori, tofu, enoki, shiitake, corn, scallion, greens, no egg or meat), but the pink spiral swirl near the center resembles a narutomaki fishcake (or possibly a shrimp curl); its appearance is ambiguous, so an intermediate score is used rather than full credit." } }, { "noodles": { "score": 0.65, "reason": "Golden strands read as noodles in broth with some bends and overlap, but they are drawn as fairly uniform flat ribbons with repetitive parallel lines and slightly rigid sweeps, especially near the right side." }, "toppings": { "score": 0.75, "reason": "Good variety: standing nori sheet, tofu cubes, enoki bundle with capped stems, dark shiitake ellipses, green leafy vegetable, corn kernels and scallion rings all read as their ingredients; the nori's glossy highlight and the simplified green blob shapes look somewhat stylized/plastic in places." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with a cream inner rim line, tapering dark ceramic body, consistent perspective, subtle side highlight and a grounding contact shadow; no visible distortion or detached parts." }, "broth": { "score": 0.75, "reason": "Red-brown liquid surface with tonal variation and a warm sheen under the steam; toppings sit in and emerge from the broth believably, though the surface is somewhat flat and simplified in the illustration style." }, "steam": { "score": 0.8, "reason": "Steam clearly animates across time-ordered frames: a tall S-wisp at 1s, a lower diffuse cloud at 2.9s, faded surface mist at 4.8s, and taller multi-lobe plumes in the pointer frames; it stays translucent and never obscures the food." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark vignette with warm backlight glow and faint dust specks; balanced proportions, no distracting UI or clutter, coherent moody lighting." }, "vegan": { "score": 0.75, "reason": "Visible toppings (tofu, enoki, shiitake, greens, corn, scallion, nori) are all plant-based; the small red swirl in the broth is ambiguous (chili oil vs. fish-cake-like spiral), so not full credit." } }, { "noodles": { "score": 0.65, "reason": "Continuous yellow strands with plausible bends and a shadowed strand layer for volume, but they read mostly as flat wavy ribbons painted on the broth surface rather than a coherent noodle mound; no rigid maze or broken strands." }, "toppings": { "score": 0.75, "reason": "Good variety: tofu cubes with chive detail, dark nori slab, enoki with dotted caps, brown mushroom caps, green florets, corn kernels and scallion rings. Shapes are recognizable though simplified; corn reads slightly as bead clusters and mushroom caps are a bit glossy and generic." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: elliptical rim with visible thickness and lighter inner band, deep rounded body, plausible base contact with a soft floor shadow, and subtle rim/body highlights. Flat-stylized but no distortion or detached parts." }, "broth": { "score": 0.8, "reason": "Rich reddish-brown liquid surface with soft sheen highlight, a chili-oil swirl, and believable occlusion shadows under the nori and tofu; toppings sit in the broth rather than floating. Surface depth is modestly flat in places." }, "steam": { "score": 0.8, "reason": "Frames show animated translucent steam: the plume rises, shifts shape, billows and fades between desktop-1, -2, -3 and pointer frames without hiding the dish. Some wisps become disconnected soft blobs near the top in a few frames." }, "composition": { "score": 0.85, "reason": "Centered focal dish in a restrained dark scene with a warm glow behind the bowl, faint dust specks, balanced proportions and no clutter, text or competing elements; food stays the focus." }, "vegan": { "score": 0.5, "reason": "Toppings appear plant-based (tofu, enoki, mushrooms, greens, corn, scallion, nori), but the small red-pink spiral swirl in the broth center is ambiguous - it could read as a narutomaki-style fish cake or simply a chili-oil drizzle, so an intermediate score is applied." } } ], "diagnostics": { "vegan": 0.5833333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.84, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 116.59999999999945, "frameIntervalP95Ms": 200, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "4cf5a66cc8396ea9e92d6d92a9b11e4ddecb61e0dd0ae6ef07eaaf9910ae1576", "gradedAt": "2026-09-30T23:28:11.856106+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5344444444444445, "toppings": 0.6144444444444445, "bowl": 0.7144444444444444, "broth": 0.5688888888888889, "steam": 0.4911111111111111, "composition": 0.7488888888888888 }, "reward": 0.6088333333333333, "runId": "xai/grok-4.6/xhigh", "rewardStdDev": 0.08925431828955577, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48666666666666664, "toppings": 0.5766666666666667, "bowl": 0.76, "broth": 0.48333333333333334, "steam": 0.44, "composition": 0.8066666666666666 }, "reward": 0.5885, "sampleStdDev": 0.007858116822750838, "judgments": [ { "noodles": { "score": 0.48, "reason": "Readable golden strands have consistent thickness, but tightly parallel curved bundles and aligned blunt ends resemble flat combs rather than supple, interwoven noodles." }, "toppings": { "score": 0.58, "reason": "Mushroom caps, tiny enoki, scallion rings and veined leaves provide readable variety. Corn is bead-like, tofu lacks texture, and the broad rounded nori and orange slabs look rigid and schematic." }, "bowl": { "score": 0.76, "reason": "The continuous cream rim, dark outer wall and lower-edge shading form a coherent near-overhead vessel. Material highlights and cavity depth are somewhat subdued, but no material construction defect is observed." }, "broth": { "score": 0.49, "reason": "Warm brown gradients suggest soup, but the surface lacks convincing liquid sheen or contact menisci; ingredients largely appear laid over a flat colored field rather than immersed." }, "steam": { "score": 0.42, "reason": "Translucent haze changes position and density across normal frames without fully obscuring the food. It reads more as broad glowing patches than distinct rising, spreading wisps; the mobile plume is especially bright." }, "composition": { "score": 0.82, "reason": "A single centered dish, warm food colors against black, generous margins and absence of text or clutter create a clear focal presentation. Mobile framing makes the dish relatively small and the upper glow competes slightly with it." }, "vegan": { "score": 0.65, "reason": "Tofu, mushrooms, greens, corn and seaweed plausibly support vegan ramen, but the three orange-brown slabs resemble meat-like slices and remain visually ambiguous." } }, { "noodles": { "score": 0.48, "reason": "Clearly separated golden strands have consistent thickness, but mostly form parallel flat bundles with abrupt exposed ends and little tangled volume." }, "toppings": { "score": 0.57, "reason": "Mushroom caps, small enoki, leafy greens, tofu and scallion rings provide readable variety. Corn looks like shiny beads, nori resembles a thick rounded panel, and several orange pieces are generic slabs." }, "bowl": { "score": 0.76, "reason": "The near-overhead rim and inner cavity are coherent, with consistent thickness and a shaded outer wall. Material highlights and interior depth remain somewhat flat, but no material construction defect is observed." }, "broth": { "score": 0.48, "reason": "Warm brown color and a soft highlight suggest soup, but the surface is largely featureless and ingredients appear layered on top without convincing wet contact or partial immersion." }, "steam": { "score": 0.42, "reason": "Translucent haze changes subtly across normal desktop frames and leaves most food readable. It reads more as diffuse luminous fog than distinct rising, spreading wisps; the mobile plume is excessively bright." }, "composition": { "score": 0.8, "reason": "A single centered dish, restrained dark surroundings and complementary warm soup and green garnish create a clear focal point without clutter. Mobile framing leaves substantial empty space and a bright plume competes with the food." }, "vegan": { "score": 0.72, "reason": "Most visible toppings plausibly represent vegetables, mushrooms, tofu and seaweed. The orange-brown lined slices are ambiguous and somewhat meat-like, preventing confident full vegan compliance." } }, { "noodles": { "score": 0.5, "reason": "Individual golden strands are readable, but tightly parallel curved bundles and aligned blunt ends look like flat combs rather than supple, interwoven noodles." }, "toppings": { "score": 0.58, "reason": "Mushroom caps, tiny enoki, veined leaves and scallion rings provide variety. Corn looks like shiny beads, tofu lacks texture, and nori resembles a thick rounded strip." }, "bowl": { "score": 0.76, "reason": "The continuous cream rim and dark outer wall share a coherent near-overhead perspective, with credible thickness and shading. Cavity depth and ceramic surface character are limited." }, "broth": { "score": 0.48, "reason": "Warm brown shading suggests soup, but the surface is almost featureless, with little liquid sheen or contact detail; ingredients appear layered atop it rather than immersed." }, "steam": { "score": 0.48, "reason": "Translucent haze changes subtly across normal frames and leaves most food readable, but distinct rising, spreading wisps are poorly defined. The mobile plume is conspicuously bright and dense." }, "composition": { "score": 0.8, "reason": "A centered single bowl, restrained dark surroundings and warm-green color contrast create a clear, uncluttered focus. Mobile presentation leaves substantial empty space and gives the bright plume excessive emphasis." }, "vegan": { "score": 0.65, "reason": "Tofu, greens, mushrooms, corn and seaweed appear plant-based, but the three pink-brown rounded slices at the top resemble meat and remain visually ambiguous." } } ], "diagnostics": { "vegan": 0.6733333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5166666666666667, "toppings": 0.55, "bowl": 0.5833333333333334, "broth": 0.5166666666666667, "steam": 0.4, "composition": 0.6233333333333333 }, "reward": 0.5315000000000001, "sampleStdDev": 0.013527749258468637, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles are a fairly readable tan tangle with overlapping strands, but they look stiff and rigid with repeated parallel lines. They form a small block in a large bowl, and the edges are cut off abruptly on the left and right." }, "toppings": { "score": 0.55, "reason": "Variety is good: nori sheet, shiitake, enoki, corn, tofu cubes, leaves, scallion. Nori is a thick dark rounded slab. The salmon-pink slices with striations look like meat or fish (possibly fish cake), and the tofu cubes are smooth blocks. Corn kernels are glossy. The toppings are clustered and pasted on." }, "bowl": { "score": 0.6, "reason": "A top-down circular bowl with a tan rim, dark outer ring and soft shadow is coherent and symmetrical. It is flat, with little wall depth, thickness or material highlights, and there is no visible base contact." }, "broth": { "score": 0.5, "reason": "The orange-red broth has a pleasant gradient and vignette, but it reads flat and opaque. Ingredients sit on top with little immersion or contact ripples, and there is little surface sheen." }, "steam": { "score": 0.4, "reason": "Faint hazy wisps over the food change slightly between frames. They are subtle and fairly static on desktop. The mobile frame shows a large glowing plume, but it looks like a blob of light." }, "composition": { "score": 0.65, "reason": "The bowl is centered and is the clear focus in a dark, restrained scene with warm colors. The toppings are lopsided and clustered, the bowl is small relative to the frame on mobile, and there is lots of empty broth." }, "vegan": { "score": 0.5, "reason": "Most ingredients are plant-based: tofu, mushrooms, nori, corn and vegetables. The pink striated slices look like meat or fish cake, which makes the appearance ambiguous." } }, { "noodles": { "score": 0.5, "reason": "The noodles read as thin parallel strands with a crossing pattern, and they have some volume. They look stiff and rigid, like a woven mat, with a hard cut edge on the left side. They sit on top of the broth rather than in it." }, "toppings": { "score": 0.55, "reason": "The toppings are varied and recognizable: shiitake, nori, corn, tofu cubes, leaves, scallion and enoki. The pink-orange slices look like meat or fish cake, which is ambiguous. The nori is a thick dark slab, the corn is a flat cluster, and the tofu cubes are plain and rounded." }, "bowl": { "score": 0.55, "reason": "The bowl is a top-down concentric ring with a tan rim and a dark outer shadow. It is clean and coherent, but flat, with little wall depth, base or material detail." }, "broth": { "score": 0.5, "reason": "The broth is a warm orange-red with a vignette and some depth. It looks like a flat gradient disc, and the ingredients are pasted on top with little immersion or contact." }, "steam": { "score": 0.4, "reason": "The steam is faint and hazy on desktop, with slight variation between frames. It is barely visible and subtle. On mobile it forms a bright glowing plume." }, "composition": { "score": 0.6, "reason": "The bowl is centered on a dark background, with the food as the focus and warm colors. The toppings are clustered off-center and the lighting is fairly flat." }, "vegan": { "score": 0.5, "reason": "Most ingredients are plant-based: tofu, mushrooms, corn, nori and greens. The pink-orange slices resemble meat or fish cake, so the appearance is ambiguous." } }, { "noodles": { "score": 0.55, "reason": "Noodles are a fairly readable tangled bundle of tan strands with some overlap, but they look like rigid, repeated parallel lines in a cropped, flat-edged mass with hard cut edges on the right and left." }, "toppings": { "score": 0.55, "reason": "Variety is good: nori, shiitake, enoki, tofu cubes, corn, bok choy leaves, scallions. Several pieces look generic though: the pinkish slices read as meat or fish cake, the tofu is blocky, and the corn is a plain grid of beads." }, "bowl": { "score": 0.6, "reason": "Top-down bowl with a tan rim, darker outer ring and soft shadow reads as a coherent ceramic bowl. It is fairly flat, with little wall depth, a weak base or contact, and modest highlights." }, "broth": { "score": 0.55, "reason": "The orange-red broth has a pleasant vignette gradient and depth, but it is fairly flat and uniform. Ingredients sit on top with little immersion, occlusion or ripple." }, "steam": { "score": 0.4, "reason": "Faint hazy steam shifts slightly between desktop frames and is translucent, but it is subtle and blotchy, barely visible above the food. The mobile frame shows a bright glowing plume." }, "composition": { "score": 0.62, "reason": "A centered bowl on a dark background gives a clear focal point with warm colors. The ingredients are clustered off-center, leaving the broth's right side empty, and the bowl is fairly small." }, "vegan": { "score": 0.45, "reason": "Most toppings are plant-based, but the pink slices resemble ham or fish cake and look animal-derived, which makes compliance ambiguous." } } ], "diagnostics": { "vegan": 0.48333333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.7166666666666667, "bowl": 0.8, "broth": 0.7066666666666667, "steam": 0.6333333333333333, "composition": 0.8166666666666667 }, "reward": 0.7065, "sampleStdDev": 0.041824036151476326, "judgments": [ { "noodles": { "score": 0.55, "reason": "Central noodle bundle fans out with consistent strand thickness and sesame/chili details, but strands read as flat rigid lines with abrupt ends and the mass sits on the broth surface rather than dipping into it." }, "toppings": { "score": 0.7, "reason": "Good variety and readability: shiitake caps, enoki stalks, tofu cubes, spinach leaves, corn, scallion rings, bamboo shoots and nori strips all read as their ingredients; minor flatness and the pink striped slices look slightly like generic stacked discs." }, "bowl": { "score": 0.75, "reason": "Coherent top-down construction: dark wood-grain outer ring, tan ceramic rim and consistent circular geometry with a soft under-shadow; only a thin rim and limited depth evidence keep it from higher marks." }, "broth": { "score": 0.72, "reason": "Rich reddish-brown liquid with restrained glossy highlight and gradient depth; toppings cast soft shadows and appear seated in the broth, though some items (noodles, tofu) read as pasted above the surface." }, "steam": { "score": 0.5, "reason": "Across desktop frames the bright haze above the bowl shifts shape and position, showing some ambient motion, but it reads as a diffuse foggy glow rather than defined translucent wisps, and in the mobile frame it is an opaque fog mass." }, "composition": { "score": 0.8, "reason": "Centered bowl as clear focal dish against a dark atmospheric background with balanced negative space and appetizing warm palette; no text or clutter, though the mobile crop leaves the bowl low with steam dominating." }, "vegan": { "score": 0.4, "reason": "Mushrooms, tofu, corn, spinach, nori, bamboo and scallions are plant-appropriate, but the pink-brown striped slices strongly resemble chashu pork or fish cake, making the appearance ambiguously animal-derived." } }, { "noodles": { "score": 0.65, "reason": "Continuous wheat-toned strands fanning from the bowl center with sesame speckle and coherent bundling; however the parallel repeating wave pattern reads slightly flat and mechanical, more a stylized fan than a voluminous nest." }, "toppings": { "score": 0.75, "reason": "Good variety and legibility: shiitake caps, enoki cluster, tofu cubes, leafy greens, scallion rings, bamboo strips, seaweed sheet and chili threads all read as food; corn is rendered as uniform glossy yellow spheres, a slightly plastic touch, and tofu/naruto-like pieces are smooth blocks." }, "bowl": { "score": 0.8, "reason": "Coherent top-down donabe/lacquer bowl with cream inner rim, textured dark body and rim highlight crescent, consistent circular perspective and grounded shadow; wall depth is inherently limited by the flat overhead angle but nothing appears broken or detached." }, "broth": { "score": 0.7, "reason": "Dark reddish-brown liquid with restrained sheen and a soft top glow; most toppings cast subtle shadows suggesting immersion, though the surface is fairly flat and empty in the lower-right, and items like corn and bamboo sit with minimal occlusion." }, "steam": { "score": 0.75, "reason": "Translucent pale wisps drift and change shape across the desktop and pointer frames (visible shifts between ~1s, ~3s, ~5s and pointer captures), rising above the bowl without hiding the food; the effect is subtle and somewhat blobby, and nearly suppressed in reduced-motion frames as expected." }, "composition": { "score": 0.85, "reason": "Centered bowl as clear focal point against a dark moody background, warm broth glow, no text/UI/clutter or competing elements; minor dead space in the bowl's lower broth area, but presentation is balanced and atmospheric." }, "vegan": { "score": 0.5, "reason": "Ingredients are otherwise plant-appropriate (tofu, mushrooms, vegetables, seaweed, noodles), but the stacked salmon-toned slices with layered curves strongly resemble chashu pork or fish cake; appearance is ambiguous enough for an intermediate score rather than 0." } }, { "noodles": { "score": 0.6, "reason": "Noodles form a recognizable nest of continuous thin strands with gentle waves and a few visible strand ends near the corn, but they read as a flat fanned sheet with etched striation lines rather than supple strands with volume; shading is uniform and immersion depth is absent." }, "toppings": { "score": 0.7, "reason": "Good variety: dark nori slabs, speckled shiitake caps, an enoki cluster, white tofu cubes, corn kernels, scallion rings, leafy greens and chili threads. Weaknesses: corn and tofu are glossy and slightly plastic-looking, the mushrooms are a bit blobby, and the ridged tan/pink logs look artificial." }, "bowl": { "score": 0.85, "reason": "Coherent overhead bowl: clean cream rim with a consistent highlight, dark speckled ceramic outer ring, soft contact shadow, and a smooth gradient into the broth cavity. Top-down angle is intentional and internally consistent; only minor flatness of the rim." }, "broth": { "score": 0.7, "reason": "Rich miso-toned surface with a convincing radial depth gradient and a faint steam sheen near the rim. However, several toppings (tofu, corn, greens) show little occlusion and appear pasted onto the surface rather than sitting in liquid." }, "steam": { "score": 0.65, "reason": "Desktop frames show translucent wisps that shift shape and position across the time-ordered frames, so steam is animated. Defects: the wisps are fog-like and diffuse rather than defined plumes, and in mobile-1 the steam becomes a bright, semi-opaque column that washes out the toppings." }, "composition": { "score": 0.8, "reason": "Centered bowl with warm top lighting, dark atmospheric vignette, harmonious palette and no text or clutter; the dish stays the focal point. Slightly generous empty margins on desktop and the heavy mobile steam plume distract mildly." }, "vegan": { "score": 0.45, "reason": "Most toppings are plausibly vegan (tofu, corn, mushrooms, nori, scallion, greens), but the three ridged tan/pink ovals strongly resemble chikuwa/kamaboko fish cakes; appearance is ambiguous, so an intermediate score applies." } } ], "diagnostics": { "vegan": 0.45 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.5355555555555556, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 91.65000000000055, "frameIntervalP95Ms": 116.60000000000036, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "7a74299237df32235a7ad44c18889d9c10ffe85ce0dfe670128af66558c6e298", "gradedAt": "2026-09-30T23:27:44.867479+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.55, "toppings": 0.6155555555555556, "bowl": 0.7155555555555555, "broth": 0.5844444444444444, "steam": 0.61, "composition": 0.7511111111111111 }, "reward": 0.6320555555555556, "runId": "xai/grok-4.7/high", "rewardStdDev": 0.055516847626377104, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6566666666666667, "bowl": 0.73, "broth": 0.56, "steam": 0.53, "composition": 0.7933333333333333 }, "reward": 0.6383333333333334, "sampleStdDev": 0.0037859388972001857, "judgments": [ { "noodles": { "score": 0.55, "reason": "Cream-colored strands have rounded bends and some readable overlap, but much of the center merges into broad flat ribbons with little convincing strand volume." }, "toppings": { "score": 0.65, "reason": "Veined greens, mushroom caps and slices, tofu pieces, scallion rings and corn provide readable variety. Shapes are overly smooth and simplified, especially the glossy mushroom cap and flat greens." }, "bowl": { "score": 0.73, "reason": "The elliptical opening, shaded inner wall and continuous rounded exterior form a coherent vessel with a restrained rim highlight. The base and surface contact remain indistinct, giving the bowl a slightly hovering appearance." }, "broth": { "score": 0.56, "reason": "Warm orange liquid surrounds the food with small highlights, but its layered elliptical bands look graphic and flat. Ingredients show limited immersion and contact with the liquid." }, "steam": { "score": 0.53, "reason": "Normal frames show changing translucent puffs above the soup without concealing the food. However, the persistent broad central column and disconnected round blobs resemble blurred fog more than irregular rising wisps." }, "composition": { "score": 0.8, "reason": "A single uncluttered dish remains the clear focus against a dark background, with harmonious warm browns and contrasting green garnishes. The mobile dish is rather small within the extensive empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn and other plant garnishes alongside noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.58, "reason": "Cream strands have curved bends and some readable overlap, but broad, flattened strokes merge into a pale central mass with limited strand separation and volume." }, "toppings": { "score": 0.66, "reason": "Veined greens, mushroom caps and slices, tofu pieces, scallion rings, and corn provide readable variety. Several shapes remain overly smooth and emblematic, with sparse organic texture." }, "bowl": { "score": 0.73, "reason": "The elliptical opening, shaded inner cavity, continuous outer wall, and rim highlight form a coherent vessel. The broad front wall looks somewhat flattened, and the base/contact is weakly established." }, "broth": { "score": 0.57, "reason": "Warm orange liquid is visible around the ingredients with a few highlights, but its layered elliptical bands look graphic rather than liquid; immersion and contact around toppings are shallow." }, "steam": { "score": 0.52, "reason": "Normal frames show changing translucent vapor and particle positions above the soup. The main plume remains a dense blurred column accompanied by disconnected glowing blobs rather than fine, irregular wisps." }, "composition": { "score": 0.79, "reason": "A single centered dish, restrained dark background, and warm-versus-green palette create a clear focal point without text or clutter. The mobile dish is relatively small, and the tall bright steam competes slightly with the food." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, scallions, corn, and red vegetables; no clearly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.57, "reason": "Cream-colored strands have curved bends and some readable overlap, but broad merged ribbons dominate the center, limiting strand separation and convincing volume." }, "toppings": { "score": 0.66, "reason": "Veined greens, mushroom caps and slices, tofu, scallion rings, and corn provide readable variety. Several leaves are geometric, and glossy smooth caps and kernels lack organic texture." }, "bowl": { "score": 0.73, "reason": "The elliptical rim, shaded interior, and curved outer wall form a coherent vessel with a restrained highlight. The very broad front wall and indistinct base/contact make its support less convincing." }, "broth": { "score": 0.55, "reason": "Warm orange broth is visible around the ingredients, with small highlights suggesting liquid. Its smooth banded surface has little depth, and several rear toppings appear perched above rather than immersed." }, "steam": { "score": 0.54, "reason": "Normal frames show changing positions of translucent vapor patches above the soup, supporting motion. The persistent dense central column and detached round blobs resemble blurred fog more than irregular spreading wisps." }, "composition": { "score": 0.79, "reason": "A single unobstructed dish remains the focus against a restrained dark background, with warm food colors and balanced green accents. The mobile dish is small within extensive empty space, and the tall vapor column competes slightly with the food." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, scallions, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.54, "bowl": 0.6333333333333333, "broth": 0.5433333333333333, "steam": 0.6, "composition": 0.6766666666666666 }, "reward": 0.5736666666666667, "sampleStdDev": 0.017580766005306284, "judgments": [ { "noodles": { "score": 0.5, "reason": "Pale noodle strands are visible and softly curved, but they are small, clumped and hard to read within the small food area. No individual strand volume is convincing." }, "toppings": { "score": 0.55, "reason": "There are recognizable bok choy leaves, a dark green leaf, shiitake, tofu cubes, corn, scallion and chili. They are tiny and stylized, and the tofu cubes look like small rigid blocks. The variety is decent but the detail is limited." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent rim, wall and base with a glossy highlight and warm lighting. It is very wide and shallow, so the food is cramped against the back, and the bottom looks slightly floaty with little contact shadow." }, "broth": { "score": 0.55, "reason": "The orange broth is visible at the edges and has a warm sheen. It is a thin sliver mostly covered by ingredients, with little depth or texture." }, "steam": { "score": 0.6, "reason": "Soft translucent steam rises and drifts, with changing shapes across frames. It includes blurry bokeh-like blobs and forms a fairly uniform column, but it does not hide the food." }, "composition": { "score": 0.65, "reason": "The bowl is centered in a dark, atmospheric scene with warm glow and no clutter. The food is small relative to the bowl and the bowl's body is a large empty area." }, "vegan": { "score": 0.92, "reason": "Toppings are tofu, mushrooms, vegetables, corn and chili. No meat or egg is visible." } }, { "noodles": { "score": 0.5, "reason": "Pale noodle strands are visible and curve softly, but they are small, tangled and hard to read within the crowded bowl. No distinct supple volume is visible." }, "toppings": { "score": 0.55, "reason": "Bok choy leaves, dark leaf, shiitake, tofu cubes, corn, scallions, chili and tomato are all recognizable. They are small and somewhat clip-art-like, and the tofu cubes look like flat cards with odd dots." }, "bowl": { "score": 0.68, "reason": "The wide, shallow bowl has a coherent rim, inner wall, tapered base and warm highlights, and it reads as ceramic. The base is vague and the bowl appears to float without clear support or contact. The interior cavity is also shallow." }, "broth": { "score": 0.58, "reason": "The orange-red broth has a warm glow and some sheen, and the ingredients sit partly in it. The visible surface is small and mostly covered, and its depth is limited." }, "steam": { "score": 0.6, "reason": "Translucent steam rises from the bowl and changes shape across frames with soft bokeh blobs. It looks somewhat blobby and column-like rather than wispy, and it is slightly mechanical." }, "composition": { "score": 0.72, "reason": "The bowl is a centered focal point on a dark background with a warm glow and restrained atmosphere. The food is small relative to the bowl and the lower half is empty, but there is no distracting clutter." }, "vegan": { "score": 0.88, "reason": "Visible ingredients are tofu, mushrooms, vegetables, corn, chili and noodles. No meat, egg or fish is evident." } }, { "noodles": { "score": 0.5, "reason": "Pale noodle strands are visible as soft curved loops, but they are small, tangled and partly hidden by toppings, so they read as a muddy mass with little volume." }, "toppings": { "score": 0.52, "reason": "Toppings include bok choy leaves, a dark leaf, mushrooms, tofu cubes, corn, a chili ring and a tomato. They are recognizable and varied, but flat and small, and the tofu cubes and tomato look like generic vector shapes." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, inner wall, rim highlight and a warm glow on the body. The base is not clearly defined and the bowl seems to float with no contact shadow, and the shape is slightly lopsided." }, "broth": { "score": 0.5, "reason": "The orange broth is visible only as a thin band around the ingredients. It has little surface texture or depth, and the items look placed on it rather than immersed." }, "steam": { "score": 0.6, "reason": "The steam is translucent and drifts differently across frames, with soft blobs that rise and spread. It looks somewhat blurry and blobby, and the column stays fairly uniform." }, "composition": { "score": 0.66, "reason": "The bowl is a clear focal point in a dark, warm atmosphere with good color contrast. The food is small within the bowl and the bowl sits with no grounding." }, "vegan": { "score": 0.9, "reason": "The toppings are tofu, mushrooms, greens, corn and chili, with no visible meat, fish or egg." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5833333333333334, "toppings": 0.65, "bowl": 0.7833333333333333, "broth": 0.65, "steam": 0.7, "composition": 0.7833333333333333 }, "reward": 0.6841666666666667, "sampleStdDev": 0.06027713773341713, "judgments": [ { "noodles": { "score": 0.55, "reason": "A small nest of pale wavy strands is visible in the center, with organic S-bends rather than rigid ribbons, but the noodle mass is sparse and thin relative to the bowl, strands end abruptly, and they read as flat light strokes pasted on the surface rather than a coherent, voluminous bundle sitting in broth." }, "toppings": { "score": 0.6, "reason": "Good variety and readable garnishes: green leaves, mushroom slices with cream cross-sections, white tofu cubes, corn kernels, scallion rings, chili ring and strips. However the two large glossy brown ellipses and the red blob read as shiny plastic stones rather than convincing food, and the tofu cubes look like flat white cards." }, "bowl": { "score": 0.7, "reason": "A coherent shallow donburi-style bowl: smooth elliptical rim with an inner lip highlight, consistent wall gradient from warm tan to dark brown, and a rounded base. Perspective is internally consistent. Minor shortcomings: the bowl appears to hover, with a diffuse glow beneath that reads more like emitted light than a contact shadow." }, "broth": { "score": 0.6, "reason": "The liquid surface has a warm orange-red gradient with a pale creamy swirl in the middle and a sheen along the inner rim, giving some depth. But several toppings (corn cluster, red blob, chili strips) sit on top with little occlusion, feeling pasted onto a fairly flat surface rather than immersed." }, "steam": { "score": 0.65, "reason": "Across the time-ordered normal frames the plume clearly changes shape (broad diffuse column, then denser, then structured with distinct puffs), confirming animated rising steam. It is translucent and does not hide the food. Weaknesses: the scattered round specks read as disconnected floating bubbles and the plume is somewhat fog-like and mechanical near the top." }, "composition": { "score": 0.7, "reason": "Clean dark atmospheric presentation: centered dish, warm rim glow, vertical steam, no text, UI or competing elements, and it scales acceptably on mobile. The bowl is somewhat small with large empty dark areas above and below, and the empty bottom shelf reads slightly vacant, but the dish remains the clear focal point." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: leafy greens, mushroom slices and caps, tofu cubes, corn, scallions, chili, and what appear to be herbs or an edible flower; no meat, egg, fish or dairy shapes are present, and the broth reads as plant-based." } }, { "noodles": { "score": 0.55, "reason": "Pale creamy noodle strands are visible in the center of the bowl with some curved, supple bends, but they read as soft squiggles under a glaze rather than distinct continuous strands with volume; strand definition is weak and partly lost among toppings." }, "toppings": { "score": 0.65, "reason": "Good variety and recognizability: tofu cubes with speckles, scallion rings, red chili strips, corn kernels, mushroom slices, dark leafy greens and a sprout garnish. Defects: the large brown mushroom cap and the red oval read as glossy plastic-like blobs, and the leaves are simple flat shapes." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with believable wall thickness, smooth tapering ceramic body, warm gradient material, rim highlight, and a soft contact reflection beneath. Slightly shallow/wide proportions and the rear inner wall fades into steam, but no broken geometry." }, "broth": { "score": 0.6, "reason": "Orange-red broth with lighter creamy swirls reads as liquid and ingredients mostly sit in it, but the surface is fairly flat with limited depth and some toppings appear pasted onto the surface rather than immersed." }, "steam": { "score": 0.7, "reason": "Across desktop-1/2/3 and pointer frames the plume shape changes clearly, showing animated, translucent, irregular steam that rises and spreads without hiding the food. The scattered small circular puffs look somewhat mechanical and blob-like rather than continuous wisps." }, "composition": { "score": 0.8, "reason": "Centered, readable focal dish against a dark atmospheric backdrop with subtle warm glow; balanced framing, coherent lighting, no text or clutter. The bowl occupies a modest portion of the frame but remains the clear focus." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, scallions, chili, corn, leafy greens and plant broth; no egg, meat, fish or dairy is depicted." } }, { "noodles": { "score": 0.65, "reason": "Pale cream strands with gentle curves are visible in the broth center and read as noodles, but they are somewhat pale and indistinct, blending into cream-colored garnish clumps; no rigid or maze-like defects, though strand definition is limited." }, "toppings": { "score": 0.7, "reason": "Good variety: a shiitake cap with visible gills, a speckled tofu cube, chili rings, corn kernels, scallion slivers, garlic cloves, sliced mushrooms and stylized leaves; the glossy red tomato-like sphere reads slightly plastic, but overall ingredients are recognizable with appropriate shapes and scale." }, "bowl": { "score": 0.85, "reason": "Coherent wide ramen bowl with a clean elliptical rim, inner cavity shading, consistent rim/body perspective, glossy ceramic gradient highlights, and a soft contact glow/shadow beneath; no broken joins or distortion observed." }, "broth": { "score": 0.75, "reason": "Warm orange liquid surface with a lighter center suggesting depth and a restrained sheen; ingredients mostly sit within the surface, though the broth reads slightly flat with limited texture around the toppings." }, "steam": { "score": 0.75, "reason": "Across the time-ordered desktop frames the translucent plume changes silhouette and its wisps/particles shift position, indicating genuine rising motion; some detached round particles read as floating blobs, and the plume is somewhat diffuse, but food remains visible." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric scene with the bowl as a clear focal point, warm broth glow against near-black background, soft under-bowl light, and no distracting text, UI or clutter; the presentation is attractive and restrained." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, corn, chili, leaves, garlic, scallions and plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9666666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999982, "frameIntervalP95Ms": 83.30000000000109, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5837cf58a1e8ee3212c7c33224b1c55a99920d7907a054631260d7b7624bfb95", "gradedAt": "2026-09-30T23:28:11.744801+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5577777777777778, "toppings": 0.5433333333333333, "bowl": 0.7766666666666666, "broth": 0.6322222222222222, "steam": 0.55, "composition": 0.7522222222222222 }, "reward": 0.6236666666666667, "runId": "xai/grok-4.7/low", "rewardStdDev": 0.05112321716532847, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.54, "toppings": 0.49666666666666665, "bowl": 0.8533333333333333, "broth": 0.5733333333333333, "steam": 0.4666666666666667, "composition": 0.7766666666666667 }, "reward": 0.6078333333333333, "sampleStdDev": 0.0057518113089125, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous pale strands have smooth bends and readable overlap, but mostly form parallel flat waves with little tangled volume or convincing submersion." }, "toppings": { "score": 0.5, "reason": "Greens, tofu, corn and brown mushroom-like slices provide variety, but details are sparse; mushrooms read largely as disks and the dark leaf-like piece lacks thin, irregular seaweed texture." }, "bowl": { "score": 0.85, "reason": "Nested elliptical rim edges, consistent curved walls, warm highlights and a grounding shadow create coherent thickness and depth. No material construction defect is observed, though the surface shading is simple." }, "broth": { "score": 0.57, "reason": "Warm amber shading and soft surface highlights suggest soup, but the liquid is very smooth and ingredients largely appear laid on top rather than partially immersed." }, "steam": { "score": 0.48, "reason": "Normal frames show translucent puffs changing position and fading above the soup without hiding most food. However, separated round blurred blobs dominate instead of connected irregular rising wisps." }, "composition": { "score": 0.78, "reason": "A single readable dish, restrained warm palette and uncluttered dark setting establish a clear focus. The bowl sits somewhat low amid extensive empty space, and mobile steam is disproportionately large relative to the dish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, leafy greens, corn, mushrooms and seaweed; no clearly animal-derived ingredient is visible." } }, { "noodles": { "score": 0.55, "reason": "Continuous pale strands have smooth bends and readable crossings, but form a mostly flat, parallel wavy band with little piled volume or convincing submersion." }, "toppings": { "score": 0.5, "reason": "Greens, tofu, corn and mushroom-like slices provide variety, but the mushrooms are simple disks, tofu lacks texture, and the dark seaweed resembles a smooth pointed leaf rather than irregular nori." }, "bowl": { "score": 0.85, "reason": "Nested elliptical rim edges, consistent rounded walls, warm highlights and a contact shadow establish coherent depth and support. No material construction defect is observed; the exterior shading is somewhat simplified." }, "broth": { "score": 0.57, "reason": "Warm brown-orange shading and soft highlights suggest soup, but the surface is largely smooth and opaque, with ingredients appearing laid over it rather than visibly immersed." }, "steam": { "score": 0.46, "reason": "Normal frames show translucent patches changing position and fading above the soup without substantially obscuring it. They read as separated blurry blobs rather than connected, irregular rising wisps." }, "composition": { "score": 0.77, "reason": "A single readable dish, restrained warm palette and uncluttered dark background create a clear focus. The bowl sits low with substantial empty space, and the mobile dish is relatively small." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, greens, mushrooms, corn and seaweed; no clearly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.57, "reason": "Continuous pale strands have smooth bends and readable overlaps, but lie as a mostly flat set of similar horizontal waves with little submerged volume." }, "toppings": { "score": 0.49, "reason": "Greens, tofu-like pieces and corn are recognizable, but mushroom-like brown disks lack structure and the dark seaweed resembles a smooth pointed leaf. Most ingredients have sparse texture and schematic shapes." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, visible inner wall, rounded body and contact shadow establish convincing depth. Layered golden rim highlights and dark ceramic shading are controlled; the broad oval body highlight is somewhat graphic, but no material construction defect is observed." }, "broth": { "score": 0.58, "reason": "Warm brown gradients and restrained highlights suggest soup, but the surface is very smooth and ingredients appear laid over it with limited immersion or contact detail." }, "steam": { "score": 0.46, "reason": "Normal frames show translucent patches changing position and fading above the soup without substantially hiding it. However, these are predominantly disconnected blurred blobs rather than connected, irregular rising wisps." }, "composition": { "score": 0.78, "reason": "The single dish remains clear against a restrained dark background, with harmonious warm colors and no distracting UI. The bowl sits rather low amid extensive empty space, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, corn and seaweed with noodles; no clearly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5666666666666667, "toppings": 0.5166666666666667, "bowl": 0.6933333333333334, "broth": 0.64, "steam": 0.48333333333333334, "composition": 0.6466666666666667 }, "reward": 0.5823333333333334, "sampleStdDev": 0.032316146634976936, "judgments": [ { "noodles": { "score": 0.55, "reason": "Thin tan strands wave and overlap with some continuity, but they are sparse, flat and ribbon-like, sit in a small cluster, and read as thin lines rather than a voluminous noodle mass." }, "toppings": { "score": 0.5, "reason": "Shiitake-like mushrooms, a green leaf, scallion curls, tofu cubes and a dark nori-like leaf are recognizable. They are small, simple and vector-flat, and the tofu cubes look like rounded blocks." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim with a copper sheen, a dark body, a base shadow and a highlight. Perspective is consistent. It looks fairly simple, and the curved line on the left wall is a bit odd." }, "broth": { "score": 0.62, "reason": "Warm reddish-brown broth with a gradient and soft highlights gives depth, and the ingredients sit in it. It is a bit flat and uniform, and the red blobs drifting across the surface look artificial." }, "steam": { "score": 0.5, "reason": "Steam changes between frames and drifts upward, but in the normal frames it is made of blurry round blobs that look like disconnected puffs. It is more convincing in the pointer frames, where it forms softer wisps." }, "composition": { "score": 0.66, "reason": "The bowl is centered in a dark, moody scene with a warm glow and clean colors, and there is no clutter. The bowl is small in the frame and leaves a lot of empty dark space." }, "vegan": { "score": 0.9, "reason": "Only noodles, mushrooms, tofu, greens and seaweed are visible, with no meat or egg. The red spots are ambiguous but read as chili." } }, { "noodles": { "score": 0.6, "reason": "Thin wavy golden strands overlap and read as noodles with some organic bends, but they are sparse, flat and sit in a narrow band with little volume." }, "toppings": { "score": 0.55, "reason": "Shiitake-like mushrooms, a green leaf, scallion curls, tofu cubes and a dark nori-like leaf are recognizable and varied, but they are small and simplistic, and the tofu cubes look like plain blocks." }, "bowl": { "score": 0.72, "reason": "Coherent elliptical rim with a copper highlight, a dark glossy body, a base shadow and a reflection accent. The perspective is consistent. The body is a bit dark and plain, and the highlight arc looks slightly odd." }, "broth": { "score": 0.68, "reason": "Warm reddish-brown broth with a gradient, a soft surface sheen and some floating chili oil spots. It reads as appetizing, though the ingredients sit somewhat on the surface and the depth is limited." }, "steam": { "score": 0.55, "reason": "Steam moves and changes across frames, from blobby puffs to softer wisps, and it stays translucent. Early frames look like disconnected round blurry blobs rather than flowing wisps." }, "composition": { "score": 0.66, "reason": "Centered bowl in a dark atmospheric scene with a warm palette and no clutter. The bowl is small in the frame with a lot of empty dark space, especially on mobile." }, "vegan": { "score": 0.93, "reason": "Only mushrooms, tofu, greens, seaweed and noodles are visible. No animal ingredients are apparent." } }, { "noodles": { "score": 0.55, "reason": "Thin, wavy, cream-colored strands overlap with some continuity and sit in the broth. They are sparse, flat and somewhat schematic, with little volume, and they read as a thin bundle rather than a full nest." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, shiitake-like caps, a green leaf, scallion curls and a dark nori/leaf shape are all recognizable. They are small and simplified, and the tofu blocks look like plain rounded rectangles. Variety is decent but the detail is limited." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim with a copper sheen, a dark body, a base highlight and a contact shadow. The perspective is consistent. The body is very dark and low in detail, and the odd curved line on its left side is a minor flaw." }, "broth": { "score": 0.62, "reason": "Warm reddish-brown broth with a gradient, highlights and some chili oil patches reads as appetizing. It is fairly flat, and the ingredients sit on it with only limited immersion or occlusion." }, "steam": { "score": 0.4, "reason": "Steam moves and changes across frames, but early frames show blobby, bokeh-like round puffs that look disconnected. Later frames become wispier. The reduced-motion frames show mechanical vertical pills." }, "composition": { "score": 0.62, "reason": "The bowl is a clear focal point in a dark, moody scene with a warm palette and no clutter. It is small in the frame with a lot of empty space, and the steam blobs are somewhat distracting." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, greens, noodles and seaweed are visible, with no meat or egg. The red chili patches are plausibly plant-based." } } ], "diagnostics": { "vegan": 0.91 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.6166666666666667, "bowl": 0.7833333333333333, "broth": 0.6833333333333333, "steam": 0.7, "composition": 0.8333333333333334 }, "reward": 0.6808333333333334, "sampleStdDev": 0.05149838185160122, "judgments": [ { "noodles": { "score": 0.5, "reason": "Noodles read as flat, repeated cream squiggles spread across the surface with little overlap, volume or nest-like gathering; they are recognizable strands but stylized flat loops rather than supple voluminous noodles." }, "toppings": { "score": 0.55, "reason": "Decent variety: mushroom caps, tofu cubes, green vegetables, corn kernels, scallion slivers and a dark seaweed-like leaf are identifiable, but shapes are simplified flat vector forms \u2014 tofu is plain featureless rectangles and mushrooms are simple glossy ovals." }, "bowl": { "score": 0.75, "reason": "Coherent high-angle ellipse construction: continuous gold rim band, dark inner wall ring, tapered lacquer-like body with a curved highlight, visible foot and soft contact shadow; only minor stylized flatness in the body highlight." }, "broth": { "score": 0.65, "reason": "Warm reddish-brown surface with radial depth, a soft central glow that intensifies in pointer frames, and a subtle oil-sheen ellipse; ingredients mostly sit in the liquid, though some toppings look slightly pasted onto the flat surface." }, "steam": { "score": 0.6, "reason": "Translucent gray puffs clearly change shape and position across the time-ordered frames (rising, spreading, fading) and brighten on hover, so motion is supported; however they read as somewhat disconnected rounded blobs rather than continuous flowing wisps." }, "composition": { "score": 0.8, "reason": "Centered focal bowl in a restrained dark atmospheric scene with warm coherent lighting and no clutter, text or competing elements; presentation is clean, if intentionally sparse." }, "vegan": { "score": 1.0, "reason": "Visible ingredients \u2014 mushrooms, tofu, greens, corn, scallions, seaweed-like leaf and noodle broth \u2014 are all plant-based; no egg, meat, fish or dairy shapes appear." } }, { "noodles": { "score": 0.6, "reason": "Continuous cream wavy strands with plausible thickness and no broken segments, but they lie as flat, quite regular repeated S-curves pasted on the broth surface with little volume or immersion." }, "toppings": { "score": 0.7, "reason": "Readable variety: two mushroom caps with highlight crescents, two shaded tofu cubes, scallion curls, sesame dots, chili and a leafy sprig. Shapes are simple and flat; the dark leaf and orange shape are slightly ambiguous." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with bronze ring, visible interior thickness, rounded ceramic body with a highlight arc, base ellipse and grounded shadow; perspective between rim, cavity and body is consistent." }, "broth": { "score": 0.75, "reason": "Warm amber gradient surface with restrained sheen, floating specks and a soft highlight; ingredients sit at the surface with believable contact, though depth cues are minimal in the flat style." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps evolve clearly across desktop-1/2/3 and pointer frames, rising and fading without hiding the food; puffs read somewhat as discrete blurred blobs rather than unified plumes." }, "composition": { "score": 0.85, "reason": "Balanced centered focal dish on a dark atmospheric background with a warm glow, attractive warm-cool contrast, no clutter or UI; bowl could be slightly larger but presentation is clean." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, scallions, chili, sesame, leafy greens and noodles in plant broth; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.6, "reason": "Cream wavy strands form a coherent nest on the right with plausible bends and consistent thickness, but they read as thin flat squiggles with a repeated-loop feel and only sparse coverage, leaving large bare broth areas." }, "toppings": { "score": 0.6, "reason": "Recognizable symbols: dark mushroom caps with highlight, small tofu cubes, scallion strokes, corn dots, a chili-red bit and a dark leaf. Variety is present, but shapes are flat cutouts with little organic texture or detail." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a bronze band, dark ceramic body, believable perspective, soft base contact and ground shadow. Polished stylized construction; the body reads slightly shallow but nothing is broken or detached." }, "broth": { "score": 0.65, "reason": "Warm reddish-brown liquid with soft lighter sheen swirls and a gentle gradient; ingredients sit within the rim at the surface. Limited depth and ingredients read as laid on top rather than submerged." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the translucent blurred wisps shift, thin out and dissipate higher over time, hugging the bowl without hiding food. Slightly blobby discrete puffs rather than continuous ribbons." }, "composition": { "score": 0.85, "reason": "Centered focal bowl in a restrained dark scene with a warm glow, coherent lighting and no clutter or text; mobile viewport scales the same dish cleanly." }, "vegan": { "score": 0.9, "reason": "Tofu cubes, mushrooms, scallions, corn dots and a leaf are plausibly plant-based; the cream swirl with green accent is mildly ambiguous but reads as a vegetable/leaf rather than fish cake, and no meat, egg or dairy is visible." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9588888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.19999999999982, "frameIntervalP95Ms": 116.60000000000036, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "6fa345a9739a01306aa8772a1aae1d131926576c9e718eaae682822971730d9f", "gradedAt": "2026-09-30T23:27:08.535771+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5688888888888889, "toppings": 0.62, "bowl": 0.7655555555555555, "broth": 0.6388888888888888, "steam": 0.7044444444444444, "composition": 0.8166666666666667 }, "reward": 0.6731111111111111, "runId": "xai/grok-4.7/xhigh", "rewardStdDev": 0.05676103253830758, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.54, "toppings": 0.5866666666666667, "bowl": 0.8166666666666667, "broth": 0.5833333333333333, "steam": 0.7133333333333333, "composition": 0.86 }, "reward": 0.6693333333333333, "sampleStdDev": 0.014433756729740593, "judgments": [ { "noodles": { "score": 0.5, "reason": "Cream-colored strands are continuous and curved, but mostly form a flat tangle of thin loops with little thickness, layering, or submerged volume." }, "toppings": { "score": 0.58, "reason": "Veined greens, scallion rings, corn, and speckled tofu provide readable variety. Mushrooms look like polished rounded buttons, and the tall seaweed resembles a thick folded card rather than thin irregular nori." }, "bowl": { "score": 0.84, "reason": "A coherent elliptical rim, shaded inner cavity, rounded ceramic walls, and attached foot establish convincing construction. Cream highlights suggest glaze; the broad exterior shading and lateral shadow are somewhat artificial." }, "broth": { "score": 0.57, "reason": "Warm amber shading and restrained highlights identify soup, but the surface is unusually smooth and disk-like, with most ingredients appearing laid on top rather than immersed." }, "steam": { "score": 0.7, "reason": "Normal desktop frames show translucent wisps changing shape and density above the soup without obscuring it. Several tall plumes retain similar parallel contours, limiting the impression of irregular spreading and dissipation." }, "composition": { "score": 0.86, "reason": "The centered dish is clearly framed against a restrained dark brown background, with warm ceramic tones and green accents providing a readable focal point on desktop and mobile. No distracting text or UI is present; mobile leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn, scallions, and seaweed with noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Cream-colored strands have consistent thickness and curved overlaps, but mostly form a flat web of elongated loops with little convincing submerged volume." }, "toppings": { "score": 0.6, "reason": "Veined greens, scallion rings, speckled tofu and mushroom caps provide readable variety. Mushrooms are overly smooth and disk-like, and the tall nori resembles a thick rigid folded card." }, "bowl": { "score": 0.79, "reason": "Coherent elliptical rim, shaded cavity and rounded ceramic walls establish convincing depth. Broad highlights are somewhat patchy, and the small foot has weak contact with the dark supporting surface." }, "broth": { "score": 0.59, "reason": "Warm amber shading and restrained highlights suggest soup, but the surface is very smooth and ingredients largely appear laid on top rather than partially immersed." }, "steam": { "score": 0.72, "reason": "Translucent wisps visibly change shape and density across normal frames and fade upward without obscuring the food. Several tall plumes retain similar parallel contours, limiting natural irregularity." }, "composition": { "score": 0.86, "reason": "The single dish is clearly isolated against a restrained dark background, with warm ceramic and broth contrasting green garnishes. No distracting text or UI is present; mobile framing leaves somewhat excessive empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, scallions, corn and seaweed, with no discernible animal-derived ingredients." } }, { "noodles": { "score": 0.62, "reason": "Thin cream strands have continuous curved paths and readable crossings, but form a mostly flat, loosely scribbled layer with little submerged volume." }, "toppings": { "score": 0.58, "reason": "Leaf veins, scallion rings, speckled tofu and mushroom caps provide recognizable variety. Smooth rounded mushrooms and the tall, thick-looking seaweed strip remain overly geometric and lack organic texture." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, shaded inner cavity and rounded ceramic walls form coherent depth, with convincing rim thickness and warm highlights. The small foot and diffuse ground shadow make contact slightly weak." }, "broth": { "score": 0.59, "reason": "Warm amber shading and restrained highlights suggest soup, but the surface is unusually smooth and several toppings appear laid on top rather than partially immersed." }, "steam": { "score": 0.72, "reason": "Normal frames show changes in translucent wisps above the soup without obscuring the food. The plumes retain similar upright, parallel shapes, limiting the impression of irregular spreading and dissipation." }, "composition": { "score": 0.86, "reason": "A centered, clearly readable dish, warm cream-and-green palette and restrained dark backdrop create a strong focal hierarchy without text or clutter. Mobile framing preserves the dish, though leaves considerable empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, scallions, mushrooms, corn and seaweed plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.6066666666666667, "bowl": 0.68, "broth": 0.6, "steam": 0.6, "composition": 0.74 }, "reward": 0.6183333333333334, "sampleStdDev": 0.009291573243177578, "judgments": [ { "noodles": { "score": 0.5, "reason": "Fine, thin cream strands overlap with some natural curves, but they read as a small, flat scribbled tangle. They sit on the broth surface with little volume or depth, and some are hard to tell apart." }, "toppings": { "score": 0.6, "reason": "Bok choy leaves with veins, tofu cubes with speckles, shiitake caps, a tall nori sheet, scallion rings, corn kernels and ginger slices give good variety. The mushrooms look like glossy smooth ovals, and the nori is a stiff slab standing upright." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a believable cream ceramic body with soft highlights, and a base with a foot ring. The translucent rim overlay looks slightly odd, the shadow blobs on the side are confusing, and the bowl appears to float." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a darker inner rim and ingredients that sit in it with some occlusion. It is fairly flat and smooth, with a slightly target-like ring and little surface texture." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape between frames, and they don't hide the food. They are faint and somewhat regular, like vertical streaks, and the first frame shows less spread." }, "composition": { "score": 0.72, "reason": "The bowl is centered on a dark, atmospheric background with a warm glow, so the food is the clear focus and the colors are appealing. The nori is slightly distracting and the bowl is a bit small in the frame." }, "vegan": { "score": 0.93, "reason": "All the visible ingredients are plant-based: tofu, mushrooms, bok choy, nori, corn, scallion and noodles. There are no animal products." } }, { "noodles": { "score": 0.55, "reason": "Thin cream strands overlap with some organic bends and read as noodles, but they form a small, flat, scribbly tangle. There is little volume and they are partly lost under the scallion bits." }, "toppings": { "score": 0.6, "reason": "The toppings are varied and readable: bok choy leaves, mushrooms, tofu cubes with speckles, a tall nori sheet, corn kernels, chili and sliced ginger. The mushrooms look like glossy ovals, and the nori is a rigid, smooth card. The tofu is a bit blocky." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, inner wall, rounded body and foot, with soft ceramic shading and highlights. The translucent rim band looks a little glassy, and the floor shadows are odd gray ovals that don't match the base contact." }, "broth": { "score": 0.6, "reason": "The broth is warm orange-brown with a darker edge and ingredients sitting in it. It reads as a fairly flat disc with little surface sheen or ripple, and immersion is modest." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise above the bowl and change shape across frames, spreading and fading. They are fairly thin and uniform, and they originate above the rim rather than from the soup surface." }, "composition": { "score": 0.75, "reason": "The bowl is centered on a dark, warm atmospheric background and is clearly the focal point, with no text or clutter. Colors are harmonious. The food is small within the bowl, and the nori dominates the top." }, "vegan": { "score": 0.93, "reason": "The visible ingredients are tofu, mushrooms, bok choy, nori, corn, scallions, ginger and noodles. No animal products are visible." } }, { "noodles": { "score": 0.55, "reason": "Fine cream strands overlap in a believable tangle with some curve, but they are thin, small and sit like a flat mat; they read as scribbly and lack volume." }, "toppings": { "score": 0.62, "reason": "Bok choy leaves, shiitake, tofu cubes with pores, scallion rings, corn kernels and a tall nori sheet are all recognizable and varied. The mushrooms look like smooth glossy ovals, the nori is a stiff rectangular card, and the items are small." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent rim, inner wall, rounded body and foot, with soft highlights and a cast shadow. The translucent rim overlay looks slightly odd, and the grey shadow blobs at the sides look detached." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a darker edge and a rim ring that give it depth. It reads fairly flat and the ingredients sit on top more than in it." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise and vary between frames, shifting and fading without hiding the food. They are fairly faint and somewhat uniform in shape." }, "composition": { "score": 0.75, "reason": "The bowl is centered as a clear focal point against a dark, warm vignette, with pleasing colors and no clutter. The food is small within the bowl." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, scallions, corn, nori and noodles all look plant-based, with no visible animal ingredients." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6333333333333333, "toppings": 0.6666666666666666, "bowl": 0.8, "broth": 0.7333333333333333, "steam": 0.8, "composition": 0.85 }, "reward": 0.7316666666666667, "sampleStdDev": 0.025166114784235794, "judgments": [ { "noodles": { "score": 0.7, "reason": "Continuous pale wavy strands spread across the broth with believable placement under the greens and scallions, but they read as flat drawn squiggles with little volume; the small scribbled clump near the tofu looks loose rather than a coherent nest." }, "toppings": { "score": 0.7, "reason": "Good variety: readable bok choy with pale stems, glossy shiitake caps, speckled tofu cubes, scallion rings, corn and chili. The nori is a very tall, rigid dark slab towering like a card, and its teal edge stripe looks synthetic, which reduces realism." }, "bowl": { "score": 0.8, "reason": "Coherent donburi-style vessel: clean rim ellipse, dark interior wall, rounded body with soft ceramic highlights and a small foot with contact shadow. Perspective is consistent; only minor soft smudges on the lower body." }, "broth": { "score": 0.75, "reason": "Appetizing orange-brown surface with restrained sheen and darker shading toward the far edge; ingredients mostly sit convincingly in the liquid, though tofu cubes and squiggle strands look slightly pasted at the rim boundary and depth is shallow." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise above the soup and clearly change shape and spread across desktop-1 through the pointer frames, fading without obscuring the food; wisps are somewhat parallel/mechanical in the first frame." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric background with a warm side glow and subtle specks; no text, UI or clutter, and the food stays the clear focal point." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are bok choy, mushrooms, nori, tofu, scallions, corn and chili in plant broth; no animal-derived toppings are apparent." } }, { "noodles": { "score": 0.6, "reason": "Thin pale strands form a readable nest in the broth with some coherent bends, but they read as flat wire-like scribbles with repeated wavy loops, abrupt breaks, and little volume; the white squiggle cluster at the front edge looks disconnected." }, "toppings": { "score": 0.6, "reason": "Recognizable bok choy, tofu cubes with speckles, scallion rings, corn, chili and bean sprouts, but the dominant toppings are flawed: the large shiitake is an overly glossy plastic-looking dome and the nori is an oversized rigid dark green slab standing like a thick card rather than thin seaweed." }, "bowl": { "score": 0.8, "reason": "Coherent ceramic bowl with a believable rim, brown interior wall, curved body, visible foot and grounding shadow; gradient shading reads as glazed stoneware, with only minor banding oddities at the inner rim join." }, "broth": { "score": 0.7, "reason": "Appetizing brown-toned liquid surface with restrained sheen, edge ring and soft occlusion shadows where tofu, sprouts and greens sit in the soup; slightly flat in the mid-region and a faint light patch under the mushroom, but ingredients read as immersed rather than pasted on." }, "steam": { "score": 0.8, "reason": "Translucent irregular grey wisps rise from the soup and visibly change shape and position across the time-ordered desktop frames (1013ms/2709ms/4559ms), spreading and fading without obscuring the food; somewhat cloudy/blobby in the upper region but clearly animated." }, "composition": { "score": 0.85, "reason": "Balanced centered focal dish on a dark atmospheric background with warm bokeh and rim lighting, restrained palette of browns/greens, no text, UI or competing elements; food remains the focus throughout." }, "vegan": { "score": 1.0, "reason": "Visible toppings are nori, shiitake mushrooms, bok choy, tofu, scallions, corn, chili and sprouts in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.6, "reason": "Noodles are present as pale cream strands with gentle curves across the broth, plus white squiggle strands near the front. They read as noodles but lie entirely flat on the surface with no lifted volume, coherent bundling or overlap depth; the front squiggles look scribble-like. Acceptable for the flat illustration style but sparse and thin." }, "toppings": { "score": 0.7, "reason": "Good variety and mostly readable: bok choy leaves with pale stalks, a large shiitake-style cap plus two smaller mushrooms, speckled tofu cubes, scallion rings, corn kernels, a red chili dab and pale bamboo-shoot slices. Defects: the large dark green nori is a thick rigid slab resembling a rounded card, and the big mushroom is overly glossy/plastic in finish." }, "bowl": { "score": 0.8, "reason": "Coherent ceramic bowl: clean elliptical rim with visible interior wall and cavity depth, smooth gradient body with a sheen band, small foot and grounded contact shadow. Perspective between rim and body is consistent. Minor shortcomings: very bulbous deep silhouette and the body highlight reads slightly like a seam." }, "broth": { "score": 0.75, "reason": "Rich miso-brown liquid surface with soft gradient, restrained sheen near the left, and a darker meniscus ring where it meets the inner wall. Ingredients sit in the broth with small contact shadows under the mushroom, though the surface is somewhat flat and a few toppings (tofu, nori base) appear pasted onto the plane rather than submerged." }, "steam": { "score": 0.8, "reason": "Translucent grey wisps rise from the soup/nori area and clearly change shape across desktop-1/2/3 and pointer frames, showing continuous rise, spread and fade without hiding the food. Wisps occasionally break into disconnected curls, but no opaque fog and no static behavior in normal frames." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric background with a warm ambient glow and faint star specks; coherent warm lighting, clean focal read, no text, UI or competing elements, and it scales sensibly in the mobile frame. Minor: large empty dark areas above and below, and the dark background does not compensate for the flat noodle work." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: mushrooms, bok choy, tofu, scallions, corn, chili, bamboo shoots, nori and noodles in a brown broth. No eggs, meat, fish or dairy visible." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9788888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 83.40000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "802871c37930fae9fbee50b7e3c91d5c0dc7ea930fc8bfe7b166bdd4ca5a0063", "gradedAt": "2026-09-30T23:33:01.133119+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.4688888888888889, "toppings": 0.5744444444444444, "bowl": 0.7388888888888889, "broth": 0.5588888888888889, "steam": 0.66, "composition": 0.7388888888888889 }, "reward": 0.6139444444444445, "runId": "z.ai/glm-5.3/high", "rewardStdDev": 0.07900093764094292, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.47333333333333333, "toppings": 0.49, "bowl": 0.7666666666666667, "broth": 0.44333333333333336, "steam": 0.6133333333333333, "composition": 0.7366666666666667 }, "reward": 0.5805, "sampleStdDev": 0.007365459931328155, "judgments": [ { "noodles": { "score": 0.48, "reason": "Cream strands are readable beneath the toppings, but mostly form sparse, flat parallel bends with little interwoven volume or convincing immersion." }, "toppings": { "score": 0.49, "reason": "Tofu, mushrooms, scallion rings and veined greens provide variety, but smooth mushroom disks, spherical corn and a rigid rectangular nori card make the food predominantly schematic." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls and grounded base form a coherent vessel with readable thickness. Broad angular highlights and a dark interior-like oval on the front weaken the ceramic finish." }, "broth": { "score": 0.43, "reason": "The warm orange surface clearly fills the bowl, but looks like a flat solid plane with minimal liquid sheen, depth or contact effects around ingredients." }, "steam": { "score": 0.61, "reason": "Normal desktop frames show pale vapor rising and expanding above the soup while leaving food readable. Rounded cloudy patches dominate over delicate wisps, and the mobile plume is excessively bright." }, "composition": { "score": 0.73, "reason": "Warm food colors stand out against a restrained dark setting without text or clutter. Oversized foreground chopsticks compete with the food, and the mobile dish is small within substantial empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn, nori and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Cream-colored strands are readable, but mostly form sparse, flat parallel bends and short rigid strips with little tangled volume or convincing immersion." }, "toppings": { "score": 0.48, "reason": "Tofu, mushroom caps, greens and scallion rings provide variety, but smooth brown disks, spherical corn, circular leaves and a rigid rectangular nori card dominate; organic texture is minimal." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, rounded walls and grounded base form a coherent vessel with readable thickness and highlights. Broad faceted shading and an interior-looking oval on the front weaken the material depiction." }, "broth": { "score": 0.44, "reason": "The warm orange surface is clear and contained within the rim, but looks like a nearly solid flat fill, with little liquid sheen, depth or contact around the toppings." }, "steam": { "score": 0.61, "reason": "Normal desktop frames show pale vapor expanding upward from the soup and changing distribution. It remains clear of most food, but resembles clusters of blurred puffs rather than delicate irregular wisps; mobile vapor is excessively bright." }, "composition": { "score": 0.73, "reason": "A restrained dark palette and uncluttered setting make the bowl the focal point. Oversized diagonal chopsticks partially obstruct the dish, and the mobile bowl is small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, scallions, seaweed and noodles plausibly depict vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Cream strands are readable and overlap beneath toppings, but mostly form sparse, flat parallel curves and a rigid folded segment rather than supple noodle volume." }, "toppings": { "score": 0.5, "reason": "Mushroom caps, tofu, scallion rings and veined greens provide variety, but spherical corn, disk-like leaves and a rigid rounded rectangular nori card make the food predominantly geometric and minimally textured." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, rounded walls and grounded base form a coherent vessel with clear thickness. Broad angular highlights and a dark oval across the lower front weaken the ceramic material and curvature." }, "broth": { "score": 0.46, "reason": "The warm orange surface clearly fills the bowl, but it is nearly uniform and matte, with little liquid sheen or depth; ingredients look placed on top rather than convincingly immersed." }, "steam": { "score": 0.62, "reason": "Normal desktop frames show translucent vapor rising and spreading above the soup without obscuring food. The vapor consists mainly of blurred cloud blobs rather than fine irregular wisps, and becomes an overly bright patch on mobile." }, "composition": { "score": 0.75, "reason": "A single centered dish, warm restrained palette and uncluttered dark setting establish a clear focus. Long foreground chopsticks dominate the silhouette, while the mobile dish is small within extensive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived topping is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.55, "bowl": 0.6333333333333333, "broth": 0.55, "steam": 0.6, "composition": 0.68 }, "reward": 0.5571666666666667, "sampleStdDev": 0.008326663997864539, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles show only as a few thin pale ribbons on the broth surface, partly hidden by toppings. They are flat and sparse with little volume, and read more as streaks than supple strands." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake caps, leaves with veins, a nori sheet, scallion rings and corn-like dots are all recognizable and vegan-appropriate. The nori is a rigid rectangle, the scallions are glossy ring discs, and the corn dots look like plastic beads." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent rim, tapered body, highlight and ground shadow, and it sits on the table plausibly. The body is dark and flat, the rim perspective is slightly odd, and the chopsticks lie across it a little unnaturally." }, "broth": { "score": 0.55, "reason": "The orange broth is warm and inviting with a slight sheen, but it is a flat ellipse. Ingredients sit on top with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Across the frames, steam grows and rises as translucent, soft wisps that drift and spread. It turns a bit blobby and puffy, and it partly covers the nori. The mobile capture shows a bright glowing blob instead." }, "composition": { "score": 0.68, "reason": "The bowl is a centered focal point in a dark, moody scene with warm lighting and no clutter. The lower half is empty, and the bowl is fairly small in the mobile view." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, vegetables, seaweed and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are only a few thin pale strands at the broth surface, partly hidden by toppings. They read as flat ribbons with little volume or bends, and the corn-like golden beads add to the muddle." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake caps, scallion rings, spinach leaves with veins and a nori sheet are all recognizable. The nori is a rigid, flat rectangle standing upright, the scallions look like glossy rings, and the golden beads are ambiguous." }, "bowl": { "score": 0.6, "reason": "The bowl has a coherent rim, tapered body, base, contact shadow and highlight. The surface is dark and flat, and the rim perspective is a little off relative to the body. The chopsticks lie across the bowl believably." }, "broth": { "score": 0.55, "reason": "The orange broth is warm and appetizing with slight sheen and some carrot streaks. It is fairly flat, and the toppings sit on top more than they are immersed in it." }, "steam": { "score": 0.6, "reason": "Across frames the steam grows and rises as translucent, soft, irregular puffs, which reads as real ambient motion. It turns somewhat blobby and cloud-like and is fairly dense, and in the mobile frame it becomes a bright blown-out glow." }, "composition": { "score": 0.68, "reason": "The bowl is a clear focal point in a dark, warm, moody scene with a tabletop and a soft glow. The framing is balanced with no distracting clutter, though the lower half of the canvas is empty." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, greens, nori and scallions, with no visible animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are only a few thin cream ribbons in the broth. They are flat, short and mostly hidden under toppings and corn. There is little volume or supple overlap." }, "toppings": { "score": 0.55, "reason": "Tofu cubes, shiitake caps, spinach leaves with veins, a nori sheet, scallion rings and corn are all readable. Nori is a rigid rectangle and the scallions and corn are glossy bead-like dots. Variety is good but the execution is simplistic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent rim, tapered walls, a highlight band, a base and a contact shadow. It reads as dark ceramic and the perspective is consistent. The body is rather flat and the chopsticks overlap it awkwardly." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm glow and a thin inner rim that gives some depth. The surface is fairly flat and lacks texture. Toppings sit on it acceptably, but the nori appears pasted in." }, "steam": { "score": 0.6, "reason": "Steam grows and rises across the frames as soft translucent puffs that drift and spread. It reads as blobby clusters and becomes somewhat dense. It does not hide the food. On mobile it is a bright white blob." }, "composition": { "score": 0.68, "reason": "A centered bowl sits in a dark warm scene with a clear focal point and a pleasing orange palette. The lower half is empty. The chopsticks cut across the bowl, and the mobile bowl is small." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, corn, scallions, nori and noodles are visible. No animal products appear." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5333333333333333, "toppings": 0.6833333333333333, "bowl": 0.8166666666666667, "broth": 0.6833333333333333, "steam": 0.7666666666666667, "composition": 0.8 }, "reward": 0.7041666666666667, "sampleStdDev": 0.044884110031650756, "judgments": [ { "noodles": { "score": 0.5, "reason": "Cream wavy strands sit plausibly in the broth on the left and right, but they are flat single-weight squiggles with uniform repeated waves and no strand volume, thickness variation, or layered overlap, reading as decorative strokes rather than supple noodles." }, "toppings": { "score": 0.65, "reason": "Good variety in a clean flat style: brown mushroom slices, white tofu cubes, scallion rings, corn dots, chili bits, leafy greens, and a nori sheet with streak texture. Defects: tofu cubes are plain featureless blocks, leaf greens are odd perfect circles, and the nori is a fairly rigid rectangular card; corn/chili are generic dots." }, "bowl": { "score": 0.8, "reason": "Coherent stylized bowl: elliptical rim with visible thickness and a back-rim highlight, dark tapered walls, small foot with grounding shadow, and chopsticks resting plausibly across the near rim. Simple but well-joined with no broken geometry; body shading is modest." }, "broth": { "score": 0.6, "reason": "Warm orange liquid surface with lighter swirl strokes suggesting oil and a dark edge where it meets the inner wall; toppings rest on and in the surface believably. However the surface is quite flat with little depth, sheen variation, or immersion cues beyond the swirls." }, "steam": { "score": 0.75, "reason": "Across time-ordered frames the steam evolves continuously: faint wisps at ~1s grow into taller, shifting translucent puffs by ~2.5s and ~4s, rising and fading above the bowl without obscuring the food. Notably absent in reduced-motion frames, confirming it is animated rather than static. Puffs are slightly blobby/cloud-like rather than fine wisps." }, "composition": { "score": 0.75, "reason": "Desktop framing is balanced with the bowl as clear focal point, warm orange against a dark atmospheric background, subtle star specks and table reflection, no text or clutter. The mobile layout leaves the upper two-thirds nearly empty and the glow reads as a bright blob, slightly weakening presentation." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, scallion, corn, chili, leafy greens, and nori in an orange broth; no egg, meat, fish, or dairy-like elements appear." } }, { "noodles": { "score": 0.65, "reason": "Pale cream strands visible on the left of the broth with wavy bends, and thin lines extending under the toppings, sitting plausibly in the soup. However, they read as flat stylized ribbons with a repeated squiggle pattern and no strong volume; largely hidden under toppings." }, "toppings": { "score": 0.7, "reason": "Good variety and recognizable items: scallion rings with white centers, cross-hatched mushroom slices, tofu cubes with subtle shading, corn kernels, veined leafy greens, and a dark nori sheet. Defects: the nori reads as a thick rigid upright slab/card, and the mushroom slices are fairly flat disks, but the set overall stays food-like and varied." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim with inner wall, cavity holding broth, tapering body, visible foot with a grounded contact shadow on the table, plus a tasteful rim highlight arc. Perspective is consistent across desktop and mobile frames; only minor stiffness in the fully symmetric silhouette." }, "broth": { "score": 0.75, "reason": "Appetizing tan-orange surface with a darker edge ring near the rim giving depth, and a restrained sheen. Ingredients sit on/in the liquid rather than floating detached, though the corn and scallion dots read slightly pasted on the flat surface." }, "steam": { "score": 0.8, "reason": "Time-ordered desktop frames show steam evolving: small wisps at 1008ms, a clustered plume at 2471ms, a taller diffuse plume by 3934ms, and continued shape change in pointer frames. It is translucent, irregular, rises and fades without hiding the food; slightly blob-clustered rather than finely wispy." }, "composition": { "score": 0.85, "reason": "Balanced centered dish on a dark atmospheric backdrop with faint star specks, warm under-glow, a clean table plane and shadow; chopsticks add diagonal interest without clutter. Minor mobile defect: a bright glow blob washes over the nori/topping area in the small viewport." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, scallions, corn, leafy greens and nori in a plant-looking broth; no meat, fish, egg or dairy visible." } }, { "noodles": { "score": 0.45, "reason": "Noodles appear mainly as flat cream ribbon shapes and a swirl-like bundle at the left of the broth, with pale wavy strands partly buried under toppings; they read as rigid flat tapes rather than supple continuous strands, and the noodle mass is indistinct." }, "toppings": { "score": 0.7, "reason": "Stylized but readable variety: brown shiitake-style slices, white tofu cubes, green scallion rings, corn and chili dots, and small leafy greens all read as food. The large standing nori sheet is somewhat card-like with uniform diagonal streaks, and the cubes/leaves are simple, keeping this below highly polished." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with visible inner cavity and wall thickness, smooth gradient shading on the body, a subtle foot, and a grounded contact shadow; chopsticks rest convincingly across the rim. Only minor shortcomings: the rim highlight stroke is thin and the right rim ellipse flattens slightly." }, "broth": { "score": 0.7, "reason": "Warm orange surface with restrained sheen and faint darker swirls, and most ingredients sit in the liquid with plausible contact. The nori and tofu lack any visible waterline or immersion shading, so they look slightly pasted on, and the surface is a touch flat." }, "steam": { "score": 0.75, "reason": "Across the time-ordered desktop frames the plume clearly evolves: small wisps at ~1s grow into a taller, spreading, fading cloud by ~4-7.6s, showing continuous rise and dispersion. It is translucent and never obscures the food, though it is somewhat puffy/cloud-blobby in texture." }, "composition": { "score": 0.8, "reason": "Dark atmospheric scene with the glowing bowl as a clear focal point, warm/cool contrast, balanced chopstick diagonal, and no text or clutter. Minor deduction for the mobile frame, where the bowl shrinks with large empty space above and below and a bright halo glow behind it." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, tofu, scallions, corn, chili, seaweed and greens in an orange plant-based broth; no egg, meat, fish or dairy imagery is present." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 49.899999999999636, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "f690e543f090af5818105549f2ed7b15c072f5ff0d52babc08e3f2ded5162711", "gradedAt": "2026-09-30T23:32:48.744441+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.3977777777777778, "toppings": 0.5288888888888889, "bowl": 0.7622222222222222, "broth": 0.5844444444444444, "steam": 0.21555555555555556, "composition": 0.76 }, "reward": 0.531, "runId": "z.ai/glm-5.3/max", "rewardStdDev": 0.046146806799363456, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.3433333333333333, "toppings": 0.48, "bowl": 0.7833333333333333, "broth": 0.52, "steam": 0.18, "composition": 0.77 }, "reward": 0.5013333333333334, "sampleStdDev": 0.0011547005383792525, "judgments": [ { "noodles": { "score": 0.35, "reason": "Golden strands are visible, but sharp zigzag bends and flat crisscrossing ribbons dominate rather than supple noodles with coherent volume." }, "toppings": { "score": 0.48, "reason": "Scallion rings, tofu pieces and scored mushroom slices provide readable variety, but remain schematic; the large smooth rectangular nori looks like a rigid card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, rounded walls, decorative bands and attached foot form a coherent bowl. Restrained highlights suggest lacquer, though the cavity and support contact lack convincing depth." }, "broth": { "score": 0.52, "reason": "Warm orange shading and a subtle edge sheen suggest soup, but the surface is mostly flat and ingredients appear laid over it with little immersion or contact detail." }, "steam": { "score": 0.18, "reason": "Normal frames show faint changing haze far toward the upper left, without a readable connection to the soup or rising wisps. Reduced-motion frames show a connected but broad, fog-like plume." }, "composition": { "score": 0.77, "reason": "A single readable dish, diagonal chopsticks and restrained warm colors create a clean dark presentation without UI or clutter. Extensive empty space and small mobile food details weaken the focal impact." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, scallions, corn-like kernels and seaweed plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.33, "reason": "Golden strands are visible, but predominantly form flat angular zigzags with sharp bends and little coherent overlap or supple volume." }, "toppings": { "score": 0.48, "reason": "Tofu, scallion rings, corn and gilled mushroom slices are recognizable. Shapes remain simplistic, especially the smooth rigid rectangular nori card and disk-like mushrooms." }, "bowl": { "score": 0.79, "reason": "Consistent elliptical rim, curved walls, decorative bands and attached foot establish coherent depth. Broad flat shading and segmented rim highlights limit material richness." }, "broth": { "score": 0.52, "reason": "Warm orange coloring and edge shading read as soup, but the surface is largely flat with minimal sheen, immersion or contact cues around ingredients." }, "steam": { "score": 0.18, "reason": "Normal frames show faint changing hazy blobs far to the upper left, without a readable plume emerging from the soup. Reduced frames show a connected but dense static fog column." }, "composition": { "score": 0.77, "reason": "The isolated dish, restrained warm palette and diagonal chopsticks create a clear focal arrangement without clutter. Excess upper negative space and small mobile food details weaken balance." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, scallions, corn and seaweed plausibly form vegan ramen; no visibly animal-derived ingredients are observed." } }, { "noodles": { "score": 0.35, "reason": "Golden strands are visible, but sharp zigzag bends and flat intersecting ribbons dominate, with little supple curvature or convincing volume." }, "toppings": { "score": 0.48, "reason": "Tofu, scallion rings, corn and lined mushroom slices are recognizable. However, the mushrooms resemble flat disks and the large smooth, rounded rectangular nori reads as a rigid card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, recessed cavity, curved walls and attached foot form a coherent vessel. Red and cream bands reinforce its shape, though highlights and material shading are rather flat." }, "broth": { "score": 0.52, "reason": "Warm orange shading and a subtle surface highlight suggest soup, but the surface remains largely solid-looking, with limited immersion and contact around the toppings." }, "steam": { "score": 0.18, "reason": "Normal frames show faint drifting blurred spots far to the upper left, without a readable plume emerging from the soup. Reduced-motion frames show a connected but broad foggy column; convincing rising wisps are not established." }, "composition": { "score": 0.77, "reason": "The single dish remains clearly framed against a restrained dark brown setting, with diagonal chopsticks providing balance and no distracting UI. Large empty upper space and subdued food detail weaken the focal presentation." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, scallions, corn, nori and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.44, "bowl": 0.6866666666666666, "broth": 0.5666666666666667, "steam": 0.38333333333333336, "composition": 0.66 }, "reward": 0.5075000000000001, "sampleStdDev": 0.003122498999199211, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin zigzag line scribbles that read as wire-like and rigid, with little volume or overlap; they sit flat on the broth surface." }, "toppings": { "score": 0.45, "reason": "Tofu cubes, mushrooms, scallion rings and a nori sheet are recognizable, but they are small and simplistic. The nori is a rigid, flat green card standing upright, and the scallions are plain dots." }, "bowl": { "score": 0.7, "reason": "The dark bowl has a coherent elliptical rim, red band, light stripe, foot and soft highlights. The perspective is consistent and the bowl reads as lacquer. The stripe is slightly flat and the foot is a bit detached." }, "broth": { "score": 0.6, "reason": "The orange-brown broth has a gradient, a subtle sheen and chili oil flecks, so it reads as liquid. It is fairly flat, and the toppings sit on the surface without much immersion." }, "steam": { "score": 0.35, "reason": "In the normal frames the steam is faint and drifts to the upper left, away from the bowl, with disconnected blobs. The reduced-motion frame shows a nicer wisp, but the normal frames are weak and sit off the soup." }, "composition": { "score": 0.65, "reason": "A centered bowl in a dark, warm atmosphere keeps the food as the focus with no clutter. The bowl is small in the frame and the background is fairly empty." }, "vegan": { "score": 0.92, "reason": "Only tofu, mushrooms, nori, scallions and noodles are visible, with no meat or egg." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin zigzag line scribbles on the surface, rigid and wire-like with little volume or overlap; they read as a maze of lines rather than supple strands." }, "toppings": { "score": 0.42, "reason": "Tofu cubes, two mushrooms, a nori rectangle and scallion rings are recognizable, but they are small and simple. The nori is a flat, rigid dark card and the scallions are tiny ring dots." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent rim, dark lacquer body, red band, highlight stripe and foot with consistent perspective. The grey ring on the wall looks somewhat flat, and the contact shadow is weak." }, "broth": { "score": 0.55, "reason": "The orange broth has warm tones and a soft gradient, but it looks fairly flat and opaque. Ingredients sit on top with little immersion, and the chili specks do little to add depth." }, "steam": { "score": 0.4, "reason": "In the desktop frames the steam appears as faint blobs drifting at the top left, disconnected from the bowl. The reduced-motion frame shows a soft plume rising from the bowl, but it is fairly thick and static." }, "composition": { "score": 0.68, "reason": "The bowl is centered as a clear focal point on a dark, moody, warm background with no clutter. The food is small within the bowl and the lower part of the frame is empty." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, nori, scallions and noodles are visible, with no animal products." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin zigzag line scribbles that read as wire-like and rigid, with little volume or overlap, and they sit flat on the broth surface." }, "toppings": { "score": 0.45, "reason": "Tofu cubes, mushrooms, scallion rings and a nori sheet are recognizable but small and simplistic. The nori is a rigid dark green rectangle, and the scallions are tiny ring dots." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, red band, highlight stripe and foot, and the perspective is consistent. The shading is a little flat and the gray outline band looks slightly detached." }, "broth": { "score": 0.55, "reason": "The orange broth has a gradient and some sheen with chili oil flecks. It reads fairly flat, and the ingredients sit on top more than they are immersed." }, "steam": { "score": 0.4, "reason": "Steam is absent above the bowl in the normal frames, and only faint wisps drift in the top-left corner, detached from the soup. The reduced-motion frames show a soft plume that is static and somewhat opaque." }, "composition": { "score": 0.65, "reason": "The bowl is centered on a dark, moody background with warm colors and the food as the focus. The bowl is somewhat small in the frame, and the chopsticks and nori feel a little stiff." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, scallions, nori and noodles are visible, with no animal products." } } ], "diagnostics": { "vegan": 0.9066666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.45, "toppings": 0.6666666666666666, "bowl": 0.8166666666666667, "broth": 0.6666666666666666, "steam": 0.08333333333333333, "composition": 0.85 }, "reward": 0.5841666666666667, "sampleStdDev": 0.009464847243000485, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles read as light tan strands clustered on the right of the broth, but they are built from rigid angular zigzag polylines with repeated flat loops and no supple bends or volume; the rigid zigzag tangle dominates, capping this at .50." }, "toppings": { "score": 0.7, "reason": "Good variety and readability: two gill-textured mushroom slices, tofu blocks, a leaning dark-green nori sheet, scallion rings with white centers, corn kernels and chili flecks. Shapes are simple flat vector forms (tofu blocks are featureless rectangles, nori a plain rectangle) but nothing reads as plastic spheres or generic disks." }, "bowl": { "score": 0.8, "reason": "Coherent footed bowl: elliptical rim with dark inner wall, navy body with red and cream bands, small foot with contact shadow. Rim, body and base perspective agree across frames, and chopsticks rest convincingly on the rim. Clean ceramic/lacquer look with only minor plainness; no distortion or detached parts." }, "broth": { "score": 0.7, "reason": "Appetizing orange-red surface with a soft radial gradient, darker edge vignette and restrained sheen arcs; no concentric target discs. Toppings and scallions sit on the surface plane, though flat styling gives limited occlusion/immersion of the noodle mass." }, "steam": { "score": 0, "reason": "All five normal desktop frames (1.0s to 10.7s) show no steam above the bowl. A plume appears only in the reduced-motion captures, so in the evaluated normal sequence steam is missing, which earns 0; no fluidity can be inferred from the identical static reduced frames." }, "composition": { "score": 0.85, "reason": "Bowl is centered as a clear focal dish against a warm glowing dark backdrop with drifting bokeh motes; chopsticks add a dynamic diagonal, mobile framing also crops cleanly, and there is no text, UI or clutter. Background glow is tasteful without replacing food quality." }, "vegan": { "score": 1, "reason": "Visible toppings are tofu, mushrooms, nori, scallions, corn and chili in plant-looking broth; no meat, fish, egg or dairy forms appear." } }, { "noodles": { "score": 0.45, "reason": "Noodles are recognizable but drawn as uniform-width angular zigzag polylines bunched on the right half of the broth; the rigid, maze-like bends and flat placement with no immersion shading dominate, capping quality." }, "toppings": { "score": 0.65, "reason": "Readable variety: shiitake slices with gill lines, scallion rings with white centers, tofu blocks, corn, chili flecks and a gradient nori sheet. Shapes are simple and slightly flat, tofu is a plain block, and items sit on the surface rather than in the soup." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: thick elliptical rim with lighter lip highlight, visible interior wall, red and cream bands that follow the body curvature, footed base with grounded contact shadow, and chopsticks resting plausibly on the rim. Minor flatness/stylization in the rim thickness." }, "broth": { "score": 0.65, "reason": "Appetizing orange surface with a soft radial gradient, left-side sheen highlight and darker edge, no fake rings or solid look; however ingredients lack occlusion/immersion integration and appear pasted onto a flat liquid plane." }, "steam": { "score": 0.15, "reason": "No steam is visible in any of the five normal desktop frames spanning ~10s, including pointer frames. Steam appears only as a fixed soft plume in the reduced-motion frames, which are identical to each other; no rising/fading behavior is supported by the evidence." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with warm broth against a dark atmospheric backdrop, subtle corner bokeh, restrained glow, clear focal point, no text or clutter, and it fits mobile cleanly. Slight emptiness on the left of the broth and missing ambient steam keep it from higher." }, "vegan": { "score": 1.0, "reason": "Visible toppings are nori, mushroom slices, tofu, scallion, corn and chili in plant broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.45, "reason": "Noodles are tan angular zigzag polyline strokes clustered on the right side of the broth; they read as rigid wire squiggles rather than supple strands, and they float on the surface without coherent volume or drape into the broth." }, "toppings": { "score": 0.65, "reason": "Recognizable variety: crosshatched shiitake slices, cream tofu cubes, scallion rings, corn and chili dots. The nori is a plain rigid dark-green rectangle standing upright like a card, and shapes are very simple flat geometry with little texture." }, "bowl": { "score": 0.8, "reason": "Coherent lacquer-style bowl: clean elliptical rim with inner ring, smooth tapered body with red and cream bands, footed base with contact shadow, consistent perspective and soft highlights. Minor: highlight arcs are generic and the rim reflection is simplistic." }, "broth": { "score": 0.65, "reason": "Orange broth reads as a liquid ellipse with a soft radial gradient, a restrained sheen arc on the right, and rim-edge occlusion; toppings sit on the surface. However it is largely a flat uniform fill lacking texture, oil sheen variation, or depth." }, "steam": { "score": 0.1, "reason": "No steam is visible in any normal desktop, pointer, or mobile frame across ~10s of captures. The reduced-motion frames show a translucent plume frozen pixel-identical between two timestamps, so the steam is effectively absent or fully static in normal playback." }, "composition": { "score": 0.85, "reason": "Single focal dish on a dark warm backdrop with a tasteful glow and faint bokeh, diagonal chopsticks adding interest, no clutter or UI. Slight weakness: the lower third is mostly empty dark space and the bowl sits a touch low." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are mushrooms, tofu, scallion, corn, chili flakes, nori and noodles in plant broth; no meat, egg, fish or dairy appears." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9688888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.30000000000109, "frameIntervalP95Ms": 99.90000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "bbd2b58e5115f21af9a7d6b299673025d8ab95b17354dfca5c6e1c2f5c18f788", "gradedAt": "2026-10-01T00:05:19.506849+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6433333333333333, "toppings": 0.6811111111111111, "bowl": 0.7977777777777778, "broth": 0.6433333333333333, "steam": 0.6977777777777778, "composition": 0.7488888888888889 }, "reward": 0.702388888888889, "runId": "openai/gpt-6.1-sol/medium", "rewardStdDev": 0.04560102744131655, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.65, "bowl": 0.8566666666666667, "broth": 0.5933333333333333, "steam": 0.68, "composition": 0.6933333333333334 }, "reward": 0.6645, "sampleStdDev": 0.004924428900898034, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin continuous golden strands are readable, but densely parallel waves and repeated end loops create a flat, comb-like mat rather than supple overlapping noodle volume." }, "toppings": { "score": 0.64, "reason": "Scored mushroom caps, grilled tofu, leafy greens, corn and scallion rings provide recognizable variety. Shapes remain overly geometric, especially the angular rear green sheet, smooth mushroom caps and minimally textured tofu." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, shaded curved walls, fine radial grooves and attached foot establish convincing depth and support. Rim highlights follow the form well; the dark finish has slightly limited material richness." }, "broth": { "score": 0.59, "reason": "Warm brown shading and restrained curved highlights suggest soup, but the surface is unusually smooth and opaque, with little visible immersion or liquid contact around the ingredient arrangement." }, "steam": { "score": 0.68, "reason": "Normal frames show translucent wavering columns changing height and fading without obscuring the food. The plumes are rather uniform narrow ribbons, with conspicuous straight lower sections and little lateral dispersal." }, "composition": { "score": 0.7, "reason": "The desktop dish is clearly isolated against a restrained dark background with harmonious warm lighting and no distracting UI. The mobile framing shifts the bowl rightward and crops a substantial portion of the dish." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin continuous golden strands are readable, but closely repeated parallel waves and nested end loops form a flat, mechanically patterned mat rather than a supple tangled volume." }, "toppings": { "score": 0.66, "reason": "Scored mushroom caps, grilled tofu, corn and leafy greens provide readable variety. Tofu lacks porous texture, mushroom details repeat, and the rear green sheet is rigid and geometric." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, shaded curved walls, fine radial surface lines and an attached foot establish convincing depth and support. Rim highlights are slightly abrupt at the sides, but no material construction defect is observed." }, "broth": { "score": 0.61, "reason": "Warm brown shading and restrained curved highlights suggest liquid, but the surface is very smooth and ingredients appear mostly perched above it with little visible immersion." }, "steam": { "score": 0.68, "reason": "Translucent wisps change height and visibility across normal frames and leave the food readable. Their similarly shaped narrow columns and conspicuously straight lower sections make the motion look schematic." }, "composition": { "score": 0.7, "reason": "The desktop dish is clearly isolated against a restrained dark background with harmonious warm colors and no clutter. Mobile framing shifts the bowl far right and clips a substantial portion of the dish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, corn and sliced vegetable garnishes; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Fine continuous golden strands have readable bends, but closely repeated parallel waves and nested loops dominate, forming a flat patterned mat rather than tangled noodle volume." }, "toppings": { "score": 0.65, "reason": "Scored mushroom caps, grilled tofu, leafy greens, corn and scallion rings provide readable variety. Shapes remain simplified, with smooth mushrooms, rigid leaves and a rectangular green sheet lacking convincing seaweed texture." }, "bowl": { "score": 0.85, "reason": "A coherent elliptical rim, shaded curved walls, fine exterior fluting and attached foot establish convincing depth and support. Rim highlights are somewhat segmented and the dark material has limited ceramic sheen." }, "broth": { "score": 0.58, "reason": "Warm brown shading and restrained curved highlights suggest soup, but the surface is very smooth and opaque, with little visible immersion or liquid contact around the noodle mat and toppings." }, "steam": { "score": 0.68, "reason": "Normal frames show translucent wisps changing height and visibility above the soup without obscuring it. The plumes remain similarly narrow and sinusoidal, with conspicuously straight lower sections and limited spreading." }, "composition": { "score": 0.68, "reason": "The desktop dish is clearly isolated against a restrained dark background with cohesive warm lighting and no clutter. The mobile framing shifts the bowl far right and clips a substantial part of the dish." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.68, "toppings": 0.66, "bowl": 0.7366666666666667, "broth": 0.6133333333333333, "steam": 0.6866666666666666, "composition": 0.77 }, "reward": 0.6896666666666667, "sampleStdDev": 0.0015275252316519481, "judgments": [ { "noodles": { "score": 0.68, "reason": "Fine, many parallel strands with gentle wavy bends and a coherent nest. Strands are a bit uniform and combed, with little overlap variation, and the nest sits slightly flat on the broth." }, "toppings": { "score": 0.66, "reason": "Readable shiitake with cross scoring, bok choy with veins, tofu blocks, scallion bits, chili rings and corn. Tofu looks like blocky grooved slabs, and the scallions and chili are tiny. Variety is good, but the shapes are stylized." }, "bowl": { "score": 0.75, "reason": "Coherent elliptical rim, deep curved wall with fine vertical ridges, small foot and a contact shadow. Highlights are subtle. The rim has a slight inner-edge kink on the left, but no major distortion." }, "broth": { "score": 0.62, "reason": "Warm amber broth with a soft gradient and faint ripple lines. It reads as liquid, but it is fairly flat and matte, with limited depth. Ingredient immersion is modest." }, "steam": { "score": 0.68, "reason": "Translucent wavy wisps rise and change shape and position across frames, and they fade at the top without hiding the food. The wisps are fairly similar, thin and smooth-edged, like ribbons." }, "composition": { "score": 0.75, "reason": "Centered bowl as clear focal point on a dark vignette with warm glow, no clutter or text. The bowl is small and the large dark surround is empty. Mobile view crops the bowl at the right edge." }, "vegan": { "score": 0.96, "reason": "Tofu, mushrooms, bok choy, scallions, chili, corn and noodles. No visible animal products." } }, { "noodles": { "score": 0.68, "reason": "Dense bundle of fine golden strands with wavy bends and overlap reads as ramen noodles. The strands are fairly uniform and the mass looks like a flat mat resting on the broth, so volume is limited." }, "toppings": { "score": 0.66, "reason": "Recognizable shiitake with scored caps, a bok choy leaf with veining, tofu blocks, corn kernels, scallion bits and chili rings give good variety. The tofu looks blocky and the scallion bits are tiny and sparse, and some items look stylized and flat." }, "bowl": { "score": 0.72, "reason": "Coherent elliptical rim, deep curved body with subtle vertical ribbing, foot ring and grounded contact shadow. The rim highlight and ceramic feel are convincing. The rim and body join look slightly stylized and the lower wall is dark and flat." }, "broth": { "score": 0.6, "reason": "Warm amber broth with a soft glow and a subtle highlight line. It reads as a fairly flat gradient with little surface texture or sheen, and the ingredients sit on top without much immersion or ripple." }, "steam": { "score": 0.7, "reason": "Translucent wavy wisps rise from the bowl and drift, fade and change shape across frames. They are somewhat similar S-curve plumes and a bit mechanical, but they do not hide the food." }, "composition": { "score": 0.78, "reason": "Centered bowl with a dark vignette, warm glow and a food-focused palette. There is no clutter or text. The bowl is slightly small in the frame, and on mobile it is cropped off to the right." }, "vegan": { "score": 0.95, "reason": "Tofu, shiitake, bok choy, corn, scallion, chili and noodles are all plant-based. No visible animal ingredients." } }, { "noodles": { "score": 0.68, "reason": "Fine, many thin strands form a coherent wavy mass with overlap and highlights, but they read as a flat, somewhat uniform patch with repeated parallel lines and little volume." }, "toppings": { "score": 0.66, "reason": "Shiitake with scored caps, bok choy with veins, grilled-looking tofu blocks, scallions, chili rings and corn kernels give good variety. Tofu is blocky, the scallions are tiny and scattered, and the corn is generic." }, "bowl": { "score": 0.74, "reason": "Coherent elliptical rim, curved walls with vertical ribbing, a footed base and a contact shadow. The dark ceramic looks convincing. The rim highlight is slightly uneven and the material is a bit flat." }, "broth": { "score": 0.62, "reason": "Warm amber broth with a soft gradient and faint ripple lines. It reads as liquid, but looks fairly flat and opaque, and the ingredients sit on it with little immersion." }, "steam": { "score": 0.68, "reason": "Translucent wavy wisps change shape and position between frames, rise and fade, and do not hide the food. They are smooth and somewhat similar to each other, like mechanical ribbons." }, "composition": { "score": 0.78, "reason": "Centered bowl in a dark, moody scene with a warm glow, a clear focal point and no clutter. On mobile the bowl is cropped at the right edge." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, corn, scallions, chili and noodles. No visible animal ingredients." } } ], "diagnostics": { "vegan": 0.9533333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.75, "toppings": 0.7333333333333333, "bowl": 0.8, "broth": 0.7233333333333333, "steam": 0.7266666666666667, "composition": 0.7833333333333333 }, "reward": 0.753, "sampleStdDev": 0.03843175770115131, "judgments": [ { "noodles": { "score": 0.78, "reason": "Noodle nest reads clearly as wavy, coherent strands with believable bends and overlap in broth; slightly repetitive flat wave pattern and uniform stroke width keep it from top marks, but no maze scribbles or rigid ribbons." }, "toppings": { "score": 0.8, "reason": "Varied, recognizable toppings: two grilled tofu blocks, two shiitake caps with crosshatch gills and stems, a bok choy leaf, chili rings, scallion bits and corn/garlic chips, sensibly arranged around the noodles; shiitake cross pattern is a bit schematic and the yellow kernels slightly blobby." }, "bowl": { "score": 0.8, "reason": "Dark fluted ceramic/lacquer bowl with a coherent elliptical rim, wall thickness, foot and contact shadow; consistent slight top-down perspective and tasteful rim highlight, only minor line artifacts where the rim highlight breaks on the right." }, "broth": { "score": 0.75, "reason": "Warm brown broth with gentle gradient, left-side sheen, darker pooling under the noodles and soft shadows seating the tofu and mushrooms; surface reads a touch flat and immersion cues are subtle, but no target discs or pasted-on look." }, "steam": { "score": 0.78, "reason": "Across desktop-1/2/3 and pointer frames the wisps clearly change shape, height and count over time, rising translucent gray plumes from the broth without fogging the food; the smooth S-curve style is slightly mechanical and one segment appears briefly detached." }, "composition": { "score": 0.72, "reason": "Desktop framing is excellent: centered dish as focal point, restrained dark vignette with warm glow, no clutter or text; however the mobile frame crops the bowl against the right edge with large empty space, hurting balance." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, shiitake mushrooms, bok choy, corn/garlic, chili and scallions over noodle and plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.72, "reason": "Curvy, supple strand bundles with some overlap and chili/scallion speckles, though strands cluster into a slightly flat, dense mass rather than reading as individual lifted noodles." }, "toppings": { "score": 0.7, "reason": "Grilled tofu slabs with score marks, two shiitake caps with cross hatching, a bok choy leaf, garlic bits and chili rings \u2014 readable variety, but forms are flat vector cutouts with little texture depth and the tofu is blocky." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with rim highlights, ribbed body, visible base ring and grounded contact shadow; perspective between the elliptical rim and the tall body is slightly inconsistent, and the foot looks short." }, "broth": { "score": 0.72, "reason": "Warm amber broth surface with subtle sheen streaks; toppings sit on the surface reasonably, but edges are pasted-looking and there is no liquid depth or immersion shading." }, "steam": { "score": 0.6, "reason": "Wisps rise and sway across frames and respond to pointer tilt (shape and number change), but they are S-curved mechanical plumes with hard edges and read as extruded ribbons rather than translucent vapor; reduced frames show static steam." }, "composition": { "score": 0.78, "reason": "Well-centered dish on a dark warm vignette with a glow behind the bowl; restrained, no clutter or UI, though the mobile crop cuts the bowl awkwardly off the right edge." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, chili, garlic and noodles \u2014 all plausibly plant-based with no animal-derived toppings visible." } }, { "noodles": { "score": 0.75, "reason": "Curly noodle bundle reads clearly with consistent strand lines and scallion garnish; edges are slightly flat and bundle is a bit uniform, but no maze scribbles or broken strands." }, "toppings": { "score": 0.7, "reason": "Recognizable tofu slabs with grill marks, two shiitake caps with cross cuts, bok choy leaves, corn kernels, chili slices; shapes are slightly stylized/flat but varied and appropriate in scale." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: smooth rim highlight, ribbed fluted walls, footed base with contact shadow; perspective of rim ellipse and body is consistent with no joins or distortion issues." }, "broth": { "score": 0.7, "reason": "Warm tonal broth with soft sheen and a few droplet/speck accents; toppings sit within the surface rather than floating, though the liquid looks somewhat flat with limited depth and immersion cues." }, "steam": { "score": 0.8, "reason": "Translucent wavy wisps rise and disperse; comparing desktop frames the plumes shift shape and fade between 1s/2.5s/4s and pointer frames, indicating animated steam that does not obscure the food." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a dark atmospheric background with warm vignette glow, balanced proportions, no text, UI or clutter; mobile view crops the bowl toward the right edge, a minor framing shortcoming." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, leafy greens, corn, chili and scallions in plant-style broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9788888888888888, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 33.400000000000546, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "ade9b1a14a658cbd637f9b7273004e4880fb3ac847faea98c646884cea94d852", "gradedAt": "2026-09-30T23:59:36.057123+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5422222222222222, "toppings": 0.6422222222222222, "bowl": 0.7933333333333333, "broth": 0.6055555555555555, "steam": 0.6822222222222222, "composition": 0.7833333333333333 }, "reward": 0.6688888888888889, "runId": "openai/gpt-6.1-sol/low", "rewardStdDev": 0.040721729571068815, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.49333333333333335, "toppings": 0.6566666666666667, "bowl": 0.86, "broth": 0.5666666666666667, "steam": 0.6633333333333333, "composition": 0.8033333333333333 }, "reward": 0.6713333333333333, "sampleStdDev": 0.004252450274057684, "judgments": [ { "noodles": { "score": 0.5, "reason": "Readable golden strands overlap in a central mound, but many are angular, stiff and abruptly ended, producing a flat scribbled tangle rather than supple noodle volume." }, "toppings": { "score": 0.65, "reason": "Mushrooms have shaded caps, scored tops and stems; greens have readable veins, and tofu has distinct faces. However, leaves are flat and geometric, tofu lacks porous texture, and the arrangement is conspicuously paired." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, consistent wall curvature, dark inner lip and restrained highlights create convincing ceramic depth. A soft shadow supports the rounded base; no material construction defect is observed." }, "broth": { "score": 0.56, "reason": "Warm brown shading and scattered tiny surface rings suggest soup, but sheen and liquid depth are limited. Ingredients mostly appear laid on the surface with little immersion or contact integration." }, "steam": { "score": 0.68, "reason": "Translucent wisps emerge near the soup without obscuring food, and their brightness changes across normal frames. Their largely fixed, parallel S-shaped paths limit the impression of irregular rising and spreading steam." }, "composition": { "score": 0.82, "reason": "A single clearly framed bowl, restrained dark green background and warm food colors provide a clean atmospheric focus without text or clutter. The mobile dish is rather small, leaving excessive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushrooms, scallion-like rings and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Numerous golden strands have readable overlaps and some curved bends, but the dominant angular, thin ribbon-like tangle looks stiff and flat rather than supple." }, "toppings": { "score": 0.65, "reason": "Veined greens, scored mushroom caps with stems, tofu faces and scallion rings provide recognizable variety. Leaves are rather planar and stiff, and tofu lacks convincing surface texture." }, "bowl": { "score": 0.86, "reason": "A consistent elliptical rim, visible inner lip, rounded shaded walls and soft contact shadow create coherent depth and support. Restrained edge highlights suggest glazed ceramic; the material finish is somewhat plain." }, "broth": { "score": 0.58, "reason": "Warm brown shading and small surface bubbles suggest soup, but the surface has little liquid sheen and ingredients appear laid on top with limited immersion or contact effects." }, "steam": { "score": 0.67, "reason": "Translucent wisps emerge near the soup and vary in opacity across normal frames without obscuring the food. Their largely persistent parallel S-shaped paths feel mechanical, with limited visible spreading." }, "composition": { "score": 0.8, "reason": "The single dish remains clearly readable against a restrained dark green background, with harmonious warm food colors and no distracting UI. Mobile presentation leaves the bowl quite small amid excessive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, scallions and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.48, "reason": "Golden strands are individually readable with some curved overlaps, but numerous angular, rigid segments form a flat scribbled pile rather than supple noodle volume." }, "toppings": { "score": 0.67, "reason": "Mushroom caps and stems, veined greens, scallion rings and scored tofu provide readable variety. Leaves look flat and stiff, and tofu lacks porous texture." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, aligned cavity and curved shaded walls create coherent depth, with restrained highlights and a grounding shadow. The material is slightly matte and the base indistinct, but no material construction defect is observed." }, "broth": { "score": 0.56, "reason": "Warm brown shading and small oil-like rings suggest soup, but the surface has little liquid sheen and ingredients largely appear laid on top rather than immersed." }, "steam": { "score": 0.64, "reason": "Translucent wisps emerge above the soup without obscuring food, and their contours and intensity change across normal frames. They remain conspicuously similar blurred vertical waves with limited visible spreading." }, "composition": { "score": 0.79, "reason": "A single readable dish, restrained green-and-gold palette and uncluttered dark setting establish a clear focus. Mobile presentation makes the bowl rather small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, scallions and noodles plausibly form vegan ramen; no animal-derived ingredient is apparent." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.62, "bowl": 0.72, "broth": 0.55, "steam": 0.6333333333333333, "composition": 0.7133333333333333 }, "reward": 0.627, "sampleStdDev": 0.0036055512754639926, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are a small, short-stranded tangled pile with thin curved strands and some overlap. They read more like shredded or cut bits than long supple noodles. The mass is small relative to the bowl and looks flat and scribbly." }, "toppings": { "score": 0.62, "reason": "Bok choy leaves with veins, shiitake caps with scoring, tofu cubes, scallion rings and chili threads give good variety and recognizable forms. The tofu is blocky and a bit flat, and there is no nori. The mushrooms look slightly pasted on." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a dark inner lip, a rounded body with a gradient and a highlight, and a soft contact shadow. The perspective is consistent. The bowl looks somewhat smooth and vector-like, with little ceramic detail and a slightly floating base." }, "broth": { "score": 0.55, "reason": "The broth is a flat brown-orange with a few faint bubbles. It lacks sheen, depth and immersion cues, and the ingredients sit on the surface without ripples or occlusion." }, "steam": { "score": 0.6, "reason": "Translucent wavy wisps rise above the bowl and shift shape between frames, so there is visible motion. The wisps are thin, look similar to each other and read as stylized ribbons. They fade upward without hiding the food." }, "composition": { "score": 0.72, "reason": "The bowl is centered on a dark vignette with a warm glow, so it is a clear focal point with no clutter. The bowl is somewhat small in the frame, and the mobile view makes it tiny." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, bok choy, scallions and noodles. There are no visible animal products." } }, { "noodles": { "score": 0.5, "reason": "Noodles are short, thin, shredded-looking strands piled in a small central heap. They overlap with some volume but look like scribbled bits or bean sprouts, not supple continuous ramen noodles." }, "toppings": { "score": 0.62, "reason": "Bok choy leaves with veins, shiitake with scored crosses, tofu cubes with grooves, scallion rings and bamboo-like shoots give good variety and readable shapes. The tofu cubes look a bit blocky and the leaves are fairly flat." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a dark inner lip, a rounded body with a soft highlight, and a ground shadow. The perspective is consistent. The material is somewhat flat and the base is not visible." }, "broth": { "score": 0.55, "reason": "The broth is an even brown with faint bubble rings. It reads as liquid but is flat and opaque, with little depth. Ingredients sit on it with minimal immersion or reflection." }, "steam": { "score": 0.7, "reason": "Several translucent wavy wisps rise and change shape between frames, and they fade out near the top without covering the food. They are a little stringy and uniform, but they look soft and fairly natural." }, "composition": { "score": 0.72, "reason": "The bowl is centered and is the clear focal point against a dark vignette with a warm glow. Color is harmonious and there is no clutter. The bowl is smallish in the frame, and especially so on mobile." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, scallions and noodles are visible. No animal products are seen." } }, { "noodles": { "score": 0.55, "reason": "Many short, thin strands that read as noodles with some curls and overlap. They look like short shredded bits piled in a small clump rather than long supple strands. The mass is small relative to the bowl, and it sits on top of the broth instead of being immersed in it." }, "toppings": { "score": 0.62, "reason": "Bok choy leaves with veins, shiitake mushrooms with scored crosses, tofu cubes, scallion rings and bamboo-like shoots give good variety and are recognizable. The shapes are somewhat flat and vector-like. The tofu cubes are plain, and there is no nori." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a dark inner lip, a rounded body with a soft highlight, and a ground shadow. The perspective is consistent. The finish is somewhat flat, and the base is not visible." }, "broth": { "score": 0.55, "reason": "The broth is a warm brown with tiny bubble rings and a subtle gradient. It reads as a flat, matte surface with little sheen or depth. Toppings sit on it with minimal immersion cues." }, "steam": { "score": 0.6, "reason": "Several translucent wavy wisps rise from the bowl and shift shape between frames, so there is visible motion. They are thin, similar S-curves and look somewhat mechanical, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "The bowl is centered on a dark vignette with subtle warm light, and the food is the clear focus with no clutter. The bowl is a little small in the frame, and the colors are muted." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, greens, scallions and noodles. No animal products are visible." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6, "toppings": 0.65, "bowl": 0.8, "broth": 0.7, "steam": 0.75, "composition": 0.8333333333333334 }, "reward": 0.7083333333333334, "sampleStdDev": 0.008036375634160792, "judgments": [ { "noodles": { "score": 0.6, "reason": "Thin tan strands form a readable central nest with scallion rings and chili threads, but the pile is flat and scribbly with limited volume and no strands visibly dipping into the broth." }, "toppings": { "score": 0.6, "reason": "Recognizable shiitake-style mushrooms with score marks, scored tofu cubes, leafy greens, scallions and chili threads; shapes are readable but quite flat and simplified, with decorative plus-sign caps and geometric cubes." }, "bowl": { "score": 0.8, "reason": "Coherent teal ceramic bowl with a clean elliptical rim, visible rim thickness, tapered walls, subtle body highlight and grounded contact shadow; stylized but internally consistent, slightly shallow proportions." }, "broth": { "score": 0.75, "reason": "Warm orange-brown liquid surface with scattered bubbles, edge darkening and restrained sheen; toppings sit in contact with the surface, though depth cues are modest." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps change shape and position across desktop-1/2/3 and pointer frames, indicating genuine rising motion; wisps are somewhat thin and curly-line-like rather than volumetric vapor." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with warm glow against a dark vignette, clear focal bowl, no clutter or text; the pointer-responsive tilt adds polish without distraction." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, leafy greens, scallions and chili threads in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.6, "reason": "Thin yellow strands form a plausible tangled pile with varied bends and some strands lifting from the surface, but the mass is fairly flat and indistinct \u2014 individual strands read as dense scribbly tangles rather than clearly traced supple noodles, with little broth immersion." }, "toppings": { "score": 0.7, "reason": "Good variety: two green leafy vegetables, striped tofu cubes, brown mushroom slices with stems, scallion rings and thin red chili threads. Leaves and mushrooms are recognizable, but forms are simplified flat cutouts \u2014 tofu reads as striped boxes and mushroom caps as circles with a '+' motif \u2014 giving a slightly schematic feel." }, "bowl": { "score": 0.8, "reason": "Coherent teal bowl with a clean rim ellipse, dark interior band, smoothly tapering wall, rounded base and soft ground shadow; highlights and a consistent slightly-elevated viewing angle. Minor shortcoming: shallow interior depth and a flat uniform rim band." }, "broth": { "score": 0.65, "reason": "Brown miso-toned surface with restrained sheen, subtle gradient toward the rim and scattered oil/seasoning specks; toppings rest on it credibly. The surface is a bit flat and uniform, and the noodle pile sits atop rather than emerging from the liquid." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps rise from the soup and clearly morph across the desktop frames (each frame shows different wavy plume shapes), fading near the top without hiding the food; reduced frames are correctly static. Plumes are thin sinuous line-squiggles with similar S-templates, reading slightly schematic rather than soft volumetric wisps." }, "composition": { "score": 0.8, "reason": "Centered dish with a soft radial glow, clean dark atmosphere, ground shadow and no text or clutter; the bowl remains the clear focal point. Presentation is tasteful but sparse, with the dish occupying a modest portion of the frame." }, "vegan": { "score": 1.0, "reason": "Visible toppings are leaves, tofu, mushrooms, scallions and chili threads in plant broth \u2014 no animal-derived ingredients apparent." } }, { "noodles": { "score": 0.6, "reason": "Noodles form a piled mass of thin wavy tan strands with scattered scallion rings, but many strands read as tight repetitive squiggles and the tangle is somewhat muddled rather than supple, coherent strands." }, "toppings": { "score": 0.65, "reason": "Two scored shiitake caps with stems, green leafy pieces, scallion rings and red chili threads are recognizable, but the two tan tofu cubes look like patterned boxes and the leaves are simple flat shapes, keeping it stylized rather than rich." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with an inner wall band, smoothly tapering teal body, subtle side highlight and a soft contact shadow; consistent perspective and no broken joins." }, "broth": { "score": 0.7, "reason": "Warm orange-brown surface with gradient depth, small bubble specks and a gentle center glow; ingredients mostly sit on the surface, though the noodle pile appears slightly pasted above it." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps change shape continuously across the normal frames (distinct curls at 1s, crossed plumes at 2.5-4s, new arrangements in pointer frames), rising from the soup without hiding it; reduced frames hold static, correctly respecting reduced motion." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark atmospheric backdrop with a subtle warm glow and balanced negative space; no clutter, text or competing elements." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu cubes, mushrooms, greens, scallions and noodles in plant broth; no animal-derived ingredients are apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.699999999999818, "frameIntervalP95Ms": 16.800000000000182, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "bc8e4fc70e55f8e0e099eb4ce3017f7af957cbc6503d98e117f0f65d20b40d73", "gradedAt": "2026-10-01T00:23:08.200255+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6677777777777778, "toppings": 0.7777777777777778, "bowl": 0.8311111111111111, "broth": 0.6988888888888889, "steam": 0.5722222222222222, "composition": 0.8255555555555555 }, "reward": 0.7325, "runId": "openai/gpt-6.1-sol/high", "rewardStdDev": 0.060698023032055966, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6333333333333333, "toppings": 0.8433333333333333, "bowl": 0.9166666666666667, "broth": 0.7333333333333333, "steam": 0.38333333333333336, "composition": 0.8933333333333333 }, "reward": 0.741, "sampleStdDev": 0.012010412149464333, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin golden strands have readable bends and overlaps, but repeated elliptical loops dominate a flattened nest, with limited convincing volume or immersion." }, "toppings": { "score": 0.84, "reason": "Veined leafy greens, speckled tofu, textured mushroom caps with cut markings, and seeded chili slices are clearly depicted. Some mushroom contours and garnish shapes remain overly regular." }, "bowl": { "score": 0.92, "reason": "The continuous elliptical rim, rounded walls, attached foot, fine ceramic texture, and restrained highlights form a coherent supported vessel. No material construction defect is observed." }, "broth": { "score": 0.73, "reason": "Warm amber shading, small oil-like flecks, and subtle surface arcs suggest appetizing broth. Ingredient contact is somewhat shallow, and the surface lacks convincing translucent depth." }, "steam": { "score": 0.65, "reason": "Soft translucent wisps emerge above the soup without obscuring it, with subtle contour changes across normal frames. The plumes remain similarly arranged, limiting evidence of continuous rising, spreading, and fading." }, "composition": { "score": 0.9, "reason": "A single readable dish anchors the restrained dark scene, with warm food colors contrasting the charcoal bowl and ample space for steam. Desktop and mobile framing remain uncluttered; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, chili, scallion-like garnish, and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.7, "reason": "Thin golden strands have curved bends, varied crossings and readable highlights. The bundle is somewhat flattened, with many similar oval loops and limited submerged depth." }, "toppings": { "score": 0.85, "reason": "Veined, scalloped greens, speckled tofu, textured mushroom caps with cut markings, and seeded chili slices provide convincing variety. Mushroom cuts are slightly geometric and the leaves somewhat stiff." }, "bowl": { "score": 0.92, "reason": "A coherent elliptical rim, rounded tapering walls, subtle ceramic grain, curved decorative bands and attached foot establish solid depth and support. No material construction defect is observed." }, "broth": { "score": 0.74, "reason": "Warm amber shading, scattered oil-like flecks and delicate surface arcs read as soup. Limited meniscus and immersion cues make the central noodle pile and rear toppings look partly laid atop the surface." }, "steam": { "score": 0.25, "reason": "Soft translucent wisps rise from behind the ingredients without obscuring them, but their contours remain essentially unchanged across the three normal desktop frames; continuous rising and fading is not demonstrated." }, "composition": { "score": 0.89, "reason": "The single dish remains clearly readable against a restrained dark background, with balanced green, gold and reddish accents and consistent warm lighting. Mobile framing preserves the entire bowl; desktop placement is slightly bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible noodles, leafy greens, tofu-like pieces, mushrooms, scallions and chili plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.7, "reason": "Thin golden strands have curved bends, shading and readable overlaps, but the dense central nest feels flattened and somewhat repetitive rather than fully supple and volumetric." }, "toppings": { "score": 0.84, "reason": "Veined, irregular leafy greens, speckled tofu, textured mushroom caps with scored stars, and seeded chili slices provide convincing variety. Mushroom cuts and small garnishes are slightly uniform and diagrammatic." }, "bowl": { "score": 0.91, "reason": "The continuous elliptical rim, rounded tapering walls, subtle ceramic grain, restrained highlights and attached foot form a coherent supported vessel. No material construction or perspective defect is observed." }, "broth": { "score": 0.73, "reason": "Warm amber shading, scattered oil-like flecks and delicate surface marks suggest appetizing broth. Limited meniscus and immersion cues make the central ingredients appear somewhat laid over the surface." }, "steam": { "score": 0.25, "reason": "Soft translucent wisps emerge above the soup without obscuring it, but their shapes remain visibly unchanged across the three normal desktop frames; rising and fading motion is not demonstrated." }, "composition": { "score": 0.89, "reason": "A single readable dish anchors the restrained dark scene, with warm food colors balanced by green leaves and a charcoal bowl. Desktop and mobile framing remain uncluttered, with no material presentation defect observed." }, "vegan": { "score": 1, "reason": "Visible noodles, tofu, mushrooms, leafy greens and vegetable garnishes plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.62, "toppings": 0.7066666666666667, "bowl": 0.7366666666666667, "broth": 0.6066666666666667, "steam": 0.5733333333333334, "composition": 0.7333333333333333 }, "reward": 0.668, "sampleStdDev": 0.007566372975210763, "judgments": [ { "noodles": { "score": 0.62, "reason": "Fine, dense strands with swirl and overlap read as noodles, but they sit as a flat nest on the broth, with a slightly repetitive, scribbly pattern and little volume or immersion." }, "toppings": { "score": 0.72, "reason": "Bok choy leaves with veins, shiitake with star scoring, tofu cubes with a porous texture, and chili slices and scallion bits are all recognizable and varied. The tofu looks a bit blocky and the garnish is small and sticker-like, and there is no nori." }, "bowl": { "score": 0.75, "reason": "The dark ceramic bowl has a coherent elliptical rim with a gold edge, a curved body with subtle glaze lines, a foot ring, and a soft contact shadow. Perspective is consistent. Detail is slightly flat and the foot is small." }, "broth": { "score": 0.6, "reason": "The warm brown broth has a gradient and small oil flecks, but it looks fairly flat and opaque. Toppings sit on top more than in it, with little ripple or contact." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise above the bowl and shift subtly between frames. They are low in contrast, somewhat repetitive and symmetric, and do not clearly emerge from the soup surface." }, "composition": { "score": 0.74, "reason": "A centered bowl on a dark, restrained background keeps the food as the focus, with warm colors and no clutter. The bowl sits somewhat low, with empty space above." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, bok choy, chili, scallion and noodles are visible. No egg, meat or fish." } }, { "noodles": { "score": 0.62, "reason": "Fine, dense golden strands read as a tangled nest with overlap and some bends, but they look like thin repeated loops. The nest is small and sits on top of the broth rather than being immersed in it." }, "toppings": { "score": 0.7, "reason": "Bok choy leaves with veins, shiitake with star scoring, tofu cubes with a porous texture, and chili and scallion bits are all recognizable and varied. They are somewhat stylized and the tofu is blocky. No nori is visible." }, "bowl": { "score": 0.74, "reason": "The bowl has a coherent elliptical rim, a curved body with glaze lines, a foot ring and a soft shadow. The perspective is consistent. The material is a bit flat and the foot is small." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a gradient and faint ripples with specks, and it reads as soup. It looks fairly flat and opaque, and the toppings partly look placed on top rather than sitting in the liquid." }, "steam": { "score": 0.62, "reason": "Translucent wisps rise above the bowl and shift shape between frames, and they do not hide the food. They are faint and fairly uniform, and they start somewhat above the broth surface." }, "composition": { "score": 0.74, "reason": "The bowl is centered as a clear focal point against a dark background with warm, harmonious colors and no clutter. The framing is somewhat plain, with a lot of empty space below." }, "vegan": { "score": 0.93, "reason": "Tofu, mushrooms, bok choy, chili and noodles are all plant-based. No egg, meat or fish is visible." } }, { "noodles": { "score": 0.62, "reason": "Fine, dense golden strands read as a tangled nest with some overlap, but they look like thin repeated scribbled loops and sit as a flat mat on the broth, with little thickness variation." }, "toppings": { "score": 0.7, "reason": "Shiitake with star-cut caps, bok choy leaves with veins, tofu cubes with a porous texture, and chili and scallion bits are all recognizable and varied. They look a bit vector-like and the nori is missing, and the tofu cubes are small." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, curved walls with subtle line details, a foot ring and a soft contact shadow. The perspective is consistent, though the rim is thin and the material is somewhat flat and matte." }, "broth": { "score": 0.62, "reason": "The warm amber broth has a gradient and faint ripples and specks that give some depth. It is fairly flat and opaque, and the noodles and toppings sit on it with limited immersion or occlusion." }, "steam": { "score": 0.55, "reason": "Translucent wispy plumes rise above the bowl and shift subtly between frames, without hiding the food. They are faint, fairly uniform and repetitive, and the motion is subtle." }, "composition": { "score": 0.72, "reason": "A centered bowl on a dark, restrained background keeps the food as the focus. The color palette is warm and harmonious, with no clutter or text. The bowl is slightly low in the frame." }, "vegan": { "score": 0.95, "reason": "Only mushrooms, bok choy, tofu, chili, scallion and noodles are visible, with no egg, meat or fish." } } ], "diagnostics": { "vegan": 0.9433333333333334 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.75, "toppings": 0.7833333333333333, "bowl": 0.84, "broth": 0.7566666666666667, "steam": 0.76, "composition": 0.85 }, "reward": 0.7885, "sampleStdDev": 0.018754999333511095, "judgments": [ { "noodles": { "score": 0.75, "reason": "Noodles form a coherent tangled nest of continuous yellow strands with plausible bends and thickness; slight uniformity and repeated loops keep it from being fully natural, but no maze-like or broken strands." }, "toppings": { "score": 0.8, "reason": "Good variety: bok choy leaves with veins, speckled tofu cubes, shiitake caps with a stylized star-gill pattern plus a cut piece showing gills, chili rings with seeds, and scallion/edamame bits. Recognizable and detailed, though the green cluster sits a bit like a cutout sticker." }, "bowl": { "score": 0.82, "reason": "Dark ceramic bowl with a coherent elliptical rim, visible inner wall, subtle rim highlight, textured walls and a supported foot; perspective between rim, body and base is consistent with no visible distortion." }, "broth": { "score": 0.72, "reason": "Warm brown broth with tonal gradient, chili-flake speckles and restrained sheen; ingredients mostly read as sitting in the liquid, but immersion/contact shading is subtle and some toppings appear slightly pasted on the surface." }, "steam": { "score": 0.78, "reason": "Translucent irregular wisps rise above the bowl without hiding food; comparing desktop-1/2/3 and pointer frames the curl shapes and spread visibly change over time, confirming gentle continuous motion rather than static plumes." }, "composition": { "score": 0.85, "reason": "Centered focal dish with warm broth contrasted against a clean dark atmospheric background, balanced negative space, no text, UI or clutter; presentation is restrained and readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, mushrooms, leafy greens, chili, scallions and noodles in plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.8, "reason": "Central nest of fine yellow strands with plausible wave, thickness variation and overlap, garnished with sesame dots; slightly uniform/combed nest edges but clearly readable as noodles, not scribbles." }, "toppings": { "score": 0.8, "reason": "Good variety: bok choy leaves with veins, two speckled tofu cubes, sliced shiitake with scored cut, chili rings with seeds and sprouts. Shapes are stylized (star scoring reads decorative, tofu slightly flat) but all ingredients are recognizable and organic." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: elliptical rim with visible thickness, interior wall crescent behind the broth, smooth body with highlight, footed base with contact shadow; consistent perspective, with a tasteful tilt on pointer frames rather than distortion." }, "broth": { "score": 0.8, "reason": "Appetizing brown broth with soft gradient, subtle swirls near the rim and small bubble specks; ingredients nestle into the surface with believable contact, though the lighter halo around the noodle nest is slightly pasted-on." }, "steam": { "score": 0.75, "reason": "Translucent, irregular wisps rising above the broth that fade without obscuring food; comparing desktop-1/2/3 the plume shapes shift between frames, indicating real animation. Wisps are a bit thin and squiggle-like with similar S-curves repeated." }, "composition": { "score": 0.85, "reason": "Centered focal dish on a dark atmospheric background with restrained vignette and warm broth contrast; balanced proportions, no clutter or UI, and a clean mobile scaling." }, "vegan": { "score": 1.0, "reason": "Visible toppings are bok choy, tofu, mushrooms, chili, sprouts and noodles in plant broth; no animal-derived ingredients appear." } }, { "noodles": { "score": 0.7, "reason": "Noodle nest is drawn as many fine golden strands with varied bends and coherent overlap, reading clearly as ramen in broth. Minor flatness: the patch sits as a uniform mound with some repetitive looping and a faint glow halo, and a few strands seem to hover at the surface rather than submerge." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: veined bok choy leaves, sesame-speckled tofu cubes, sliced chilies with seeds, scattered edamame, and two scored shiitake caps. Defects: the mushroom star-cut pattern is decorative and the caps look slightly flat/glossy, and small green bits are scattered a bit uniformly." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with visible thickness, tapered dark ceramic body, footed base, subtle horizontal highlight bands, and a grounded contact shadow. Perspective between rim, body and foot is consistent; no distortion or detached parts observed." }, "broth": { "score": 0.75, "reason": "Appetizing orange-brown liquid with tonal swirls, speckled texture and restrained sheen; toppings sit at the surface with soft contact shading. Slight flatness: the surface is fairly smooth, immersion depth is shallow-looking, and the lighter halo around the noodle patch reads a bit artificial." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps rise from the bowl as irregular curving strands without hiding the food. Comparing desktop-1/2/3 and pointer frames, the wisp shapes and crossings shift over time, indicating animated rise and spread; wisps are somewhat thin and repetitive in character." }, "composition": { "score": 0.85, "reason": "Bowl is the clear focal point, centered on a clean dark atmospheric background with warm broth and toppings contrasting well. Lighting and soft shadow are coherent, no text, UI or clutter; only minor critique is the slightly low, static framing." }, "vegan": { "score": 1, "reason": "Visible ingredients are tofu, shiitake mushrooms, leafy greens, chili, edamame and noodles in plant broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9811111111111112, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "44ec364d3d1afe653950a12a257aa99c9709a8a6f42fd4610cffdfd73b3397f8", "gradedAt": "2026-10-01T00:41:44.539203+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.7333333333333333, "toppings": 0.7988888888888889, "bowl": 0.8566666666666667, "broth": 0.7011111111111111, "steam": 0.5622222222222222, "composition": 0.8644444444444445 }, "reward": 0.7586111111111111, "runId": "openai/gpt-6.1-sol/xhigh", "rewardStdDev": 0.044761818220112264, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.7133333333333333, "toppings": 0.8566666666666667, "bowl": 0.9266666666666667, "broth": 0.6933333333333334, "steam": 0.25, "composition": 0.92 }, "reward": 0.741, "sampleStdDev": 0.0032787192621510286, "judgments": [ { "noodles": { "score": 0.7, "reason": "Fine continuous strands have rounded bends, highlights and readable overlaps, but densely parallel waves make the central mound look overly combed and patterned." }, "toppings": { "score": 0.85, "reason": "Veined, folded greens, porous tofu with shaded sides, and mushrooms with scored caps and visible gills provide convincing variety. Some mushroom slices and leaf edges are slightly stiff and schematic." }, "bowl": { "score": 0.93, "reason": "A coherent elliptical rim, shaded curved walls, subtle ceramic bands and attached foot establish convincing depth and support. Restrained highlights and a contact shadow complete the material; no material construction defect is observed." }, "broth": { "score": 0.7, "reason": "Warm amber shading and scattered oil-like flecks suggest appetizing soup, but limited meniscus detail and immersion make several toppings appear placed on top of a fairly flat surface." }, "steam": { "score": 0.25, "reason": "Soft translucent wisps emerge above the soup without obscuring it, but their shapes remain effectively unchanged across the three normal desktop frames, reading as static steam." }, "composition": { "score": 0.92, "reason": "The single centered bowl remains clearly readable on desktop and mobile, with balanced green and amber accents, coherent directional lighting and restrained dark surroundings. No distracting text, clutter or material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushrooms, scallion-like rings and noodles plausibly form vegan ramen; no animal-derived topping is visible." } }, { "noodles": { "score": 0.72, "reason": "Fine continuous golden strands have curved ends, shading and readable overlaps, but the central bundles are overly parallel and regularly folded, reducing natural looseness." }, "toppings": { "score": 0.86, "reason": "Veined, scalloped greens, porous shaded tofu, and mushrooms with scored caps and readable gills provide convincing variety. Mushroom slices and scallion rings are slightly geometric and repetitive." }, "bowl": { "score": 0.93, "reason": "The elliptical rim, curved shaded walls, subtle horizontal ceramic grooves and attached foot form a coherent supported vessel. Rim highlights establish thickness; no material construction defect is observed." }, "broth": { "score": 0.7, "reason": "Warm amber shading and scattered oil-like flecks suggest appetizing broth, but the surface has limited liquid sheen and ingredients appear mostly perched above it with little visible immersion." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps emerge above the soup without obscuring food, but their contours remain effectively unchanged across the three normal desktop frames, reading as static smoke." }, "composition": { "score": 0.92, "reason": "A single clearly readable dish balances green, gold and brown against a restrained dark background. Soft directional lighting and contact shadow establish depth, and mobile framing preserves the whole bowl without clutter; no material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible tofu, leafy greens, mushrooms, scallions and noodles plausibly form vegan ramen, with no visibly animal-derived topping." } }, { "noodles": { "score": 0.72, "reason": "Fine continuous golden strands have curved ends, shading and readable overlap, but the dense parallel bundles look overly arranged and somewhat flattened rather than loosely piled." }, "toppings": { "score": 0.86, "reason": "Veined, folded greens, porous tofu with shaded sides, and mushrooms with scored caps and readable gills provide convincing variety. Mushroom slices are slightly stiff and geometric." }, "bowl": { "score": 0.92, "reason": "The elliptical rim, rounded tapering walls and attached foot form coherent depth and perspective. Subtle horizontal ceramic lines, directional highlights and a contact shadow support the material; no material construction defect is observed." }, "broth": { "score": 0.68, "reason": "Warm amber shading and scattered oil-like flecks suggest appetizing soup, but the surface is relatively matte and ingredients have limited visible immersion or liquid contact." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps emerge above the soup without obscuring food, but their shapes appear unchanged across the three normal desktop frames, indicating static steam." }, "composition": { "score": 0.92, "reason": "A single clearly readable bowl anchors the dark scene, with balanced green, gold and brown ingredient groups and restrained warm lighting. Desktop and mobile framing remain uncluttered; no material presentation defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings read as leafy vegetables, tofu, mushrooms and sliced green garnish alongside noodles; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.7266666666666667, "toppings": 0.74, "bowl": 0.7933333333333333, "broth": 0.6266666666666667, "steam": 0.62, "composition": 0.8066666666666666 }, "reward": 0.7253333333333334, "sampleStdDev": 0.0061101009266077925, "judgments": [ { "noodles": { "score": 0.74, "reason": "Fine, supple golden strands overlap with natural bends and a believable volume. Some strands look slightly combed and uniform, and the mass sits flat on the broth with limited immersion." }, "toppings": { "score": 0.74, "reason": "Bok choy leaves with veins, shiitake caps with star cuts and gills, and tofu cubes with speckled texture are all readable and varied. Tofu cubes are fairly blocky, scallion rings are tiny and uniform, and there is no nori." }, "bowl": { "score": 0.8, "reason": "The bowl has a coherent elliptical rim, curved walls with subtle ring lines, a foot ring, and a contact shadow. Perspective is consistent. The material reads somewhat matte and the base is a bit small." }, "broth": { "score": 0.66, "reason": "The warm brown broth has speckle texture and a rim glow, and it reads as miso-like. The surface is quite flat and matte, and toppings sit on it with little ripple or immersion." }, "steam": { "score": 0.62, "reason": "Faint translucent wisps rise above the bowl and change shape between frames without hiding the food. They are subtle and thin, and the plumes look somewhat similar to each other." }, "composition": { "score": 0.8, "reason": "A centered bowl on a dark vignette with warm broth against cool green ceramic gives a clear focal point. Colors are balanced and there is no clutter. The scene is slightly static and the bottom area is empty." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, scallions, chili and noodles are all plant-based, with no visible animal products." } }, { "noodles": { "score": 0.72, "reason": "Fine golden strands overlap in a soft, flowing bundle with natural bends and some thin loose strands. The mass is a bit uniform and flat, and it sits somewhat on top of the broth, with limited immersion." }, "toppings": { "score": 0.74, "reason": "Bok choy leaves have veins and shading, shiitake caps show star-cut gills, and tofu cubes are textured with visible thickness. Scallion rings and chili slices add variety. The tofu is slightly blocky and the scallion rings are somewhat uniform. No nori." }, "bowl": { "score": 0.78, "reason": "Coherent elliptical rim with a thick lip, a rounded body with subtle horizontal throwing lines, a foot ring, and a contact shadow. The glaze highlights are soft and the perspective is consistent. The foot looks slightly small and the bowl floats a bit above its shadow." }, "broth": { "score": 0.62, "reason": "Warm brown broth with speckles and a few ripple lines reads as miso or tonkotsu-style soup. The surface is fairly flat and matte, with little depth or sheen, and the toppings sit more on top than in the liquid." }, "steam": { "score": 0.62, "reason": "Thin translucent wisps rise above the bowl and change shape between frames without hiding the food. They are faint and sparse, so the effect is subtle." }, "composition": { "score": 0.8, "reason": "A centered bowl on a dark vignette makes the dish the clear focal point. The colors are harmonious, with green, tan and orange against a dark background. There is no clutter or text. The empty space around the bowl is large but acceptable." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, greens, scallions, chili and noodles are visible. No animal ingredients are apparent." } }, { "noodles": { "score": 0.72, "reason": "The noodle mass has many fine, flowing strands that overlap and curve with believable volume and a warm golden tone. It sits as a single neat bundle rather than being immersed in the broth, and the strands are slightly uniform and combed-looking." }, "toppings": { "score": 0.74, "reason": "The greens have veined leaves, the shiitake caps show gills and star-shaped scoring, and the tofu cubes are textured and shaded. Scallion rings and chili slices add variety. Some elements, like the tofu cubes and mushrooms, look slightly pasted on top, and there is no nori." }, "bowl": { "score": 0.8, "reason": "The bowl has a coherent elliptical rim, a thick lip, a curved body with subtle ceramic ridges, and a footed base with a contact shadow. The glaze highlights are restrained. The foot is a little small and the rim has a mildly flat feel, but no material distortion is visible." }, "broth": { "score": 0.6, "reason": "The broth is a warm brown with speckled oil texture and a plausible surface. It reads somewhat flat and matte, with little sheen or depth, and the ingredients sit on it with minimal immersion or ripples." }, "steam": { "score": 0.62, "reason": "The steam is thin, translucent wisps that change shape between frames and fade as they rise. It is faint and sparse, and the plumes are somewhat similar to each other." }, "composition": { "score": 0.82, "reason": "The bowl is centered as a clear focal point against a dark vignette, with a warm and green palette that stays balanced. There is no clutter or text, and the lighting is coherent." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, greens, scallions, chili and noodles in a brown broth. No animal products are apparent." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.76, "toppings": 0.8, "bowl": 0.85, "broth": 0.7833333333333333, "steam": 0.8166666666666667, "composition": 0.8666666666666667 }, "reward": 0.8095000000000001, "sampleStdDev": 0.007794228634059987, "judgments": [ { "noodles": { "score": 0.75, "reason": "Noodles form a coherent wavy yellow bundle with consistent strand thickness and readable overlaps, plus a few loose strands trailing toward the mushrooms; however, the mass is a fairly flat, neatly arranged pile with somewhat repetitive wave strokes and limited volume." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: bok choy with vein detail, speckled tofu cubes with plausible 3D faces, shiitake halves showing gills, scallion rings and chili slices. The cream mushroom cross-sections read slightly ambiguously, but shapes are mostly organic, not plastic blocks." }, "bowl": { "score": 0.85, "reason": "Coherent dark-green ceramic bowl: consistent elliptical rim with visible wall thickness, smoothly tapering body with glossy sheen bands, small foot with grounding shadow. Minor nit: the foot reads a touch small relative to the deep body." }, "broth": { "score": 0.75, "reason": "Appetizing orange-brown surface with speckled oil/chili particles, soft sheen, and darker edges suggesting curvature; toppings have contact shadows. Somewhat flat overall, and tofu cubes sit slightly pasted-on rather than partially submerged." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise from the soup and clearly change shape between desktop-1/2/3 and pointer frames, indicating genuine animation; they stay subtle and never obscure the food. Wisps are faint, so presence is more than visual noise but not dramatic." }, "composition": { "score": 0.9, "reason": "Centered, well-proportioned bowl as clear focal point against a restrained dark background with a warm glow; attractive color contrast between greens, tofu, and broth; no text, clutter, or competing elements." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are bok choy, tofu cubes, mushrooms, scallions, chili, noodles and plant broth \u2014 no animal-derived toppings present." } }, { "bowl": { "reason": "Dark ceramic bowl with coherent elliptical rim, thick inner wall, tapered body and small foot with a contact shadow; highlights and banding read as glazed stoneware. No broken joins; perspective is consistent across all frames.", "score": 0.85 }, "broth": { "reason": "Rich brown broth with speckled texture and restrained sheen; greens, noodles and mushrooms show contact shadows and partial immersion, so toppings sit in the liquid rather than floating above it. Surface could show a bit more depth near the far rim.", "score": 0.8 }, "composition": { "reason": "Warm-lit focal bowl centered on a clean dark vignette with no text, UI or clutter; colors are balanced and the dish stays the focus in both desktop and mobile framings. Background glow is tasteful, not compensating for weak food.", "score": 0.85 }, "noodles": { "reason": "Central nest of pale strands with plausible curvature, varied bends and coherent overlaps, garnished with scallion rings and chili threads. Slightly repetitive looping and a mildly flat, clumped mass at the edges keep it from the top tier, but it reads as noodles rather than wire scribbles.", "score": 0.75 }, "steam": { "reason": "Translucent irregular wisps rise, curl and spread above the bowl and their shapes clearly change across desktop-1/2/3 while staying identical in the reduced-motion frames, confirming genuine animation. Wisps are slightly faint and sparse but never hide the food.", "score": 0.8 }, "toppings": { "reason": "Good variety: leafy bok-choy greens with vein detail, speckled tofu cubes, shiitake-style mushrooms with gills and carved caps, scallion rings and chili slices. Minor deductions for the uniform cube repetition and the slightly decorative star-cut mushroom caps.", "score": 0.8 }, "vegan": { "reason": "Visible ingredients are tofu, mushrooms, leafy greens, noodles, scallions and chili in plant broth; no meat, fish, egg or dairy present.", "score": 1.0 } }, { "bowl": { "reason": "Coherent dark-teal ceramic bowl: elliptical rim with a glazed highlight, smooth rounded wall with subtle throwing lines, visible footed base resting on a soft contact shadow. Perspective of rim, body and foot is consistent; no distortion or detached parts observed.", "score": 0.85 }, "broth": { "reason": "Rich miso-toned liquid with speckled particle texture, restrained sheen and warm depth gradient. Toppings show believable contact/occlusion: bok choy stalks and noodle nest edges sink into the surface, tofu and mushrooms have subtle contact shading. Foreground surface is slightly uniform but reads as liquid.", "score": 0.8 }, "composition": { "reason": "Centered focal dish in a dark atmospheric scene with a warm restrained back-glow, balanced proportions and coherent lighting; no text, UI or clutter. Food remains the clear focus.", "score": 0.85 }, "noodles": { "reason": "Central nest of thin yellow strands with layered overlap, curving bends and highlights giving volume, believably seated in the broth. Slightly repetitive S-wave rhythm across strands and a few fine orange thread-like garnish strands, but no rigid ribbons or maze scribbles.", "score": 0.78 }, "steam": { "reason": "Translucent, irregular gray wisps rise from the soup and fade without hiding food; wisps clearly change shape and position across desktop-1, desktop-2, desktop-3 and pointer frames, indicating continuous motion. Static wisps in reduced frames correctly respect the reduced-motion preference.", "score": 0.85 }, "toppings": { "reason": "Good variety rendered organically: veined bok choy leaves, speckled tofu cubes, scored shiitake caps with gill detail, scallion rings and red chili slices. Minor shortcomings: tofu cubes are a bit uniform and the pale mushroom-slice shapes behind the caps read semi-abstract.", "score": 0.8 }, "vegan": { "reason": "Visible ingredients are noodles, bok choy, tofu, mushrooms, scallions and chili in plant broth; no meat, fish, egg or dairy appears.", "score": 1.0 } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 50, "frameIntervalP95Ms": 66.69999999999891, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "0412817af1bf036627342d2c0ece0a89becab7269c79fce268d8831c4b0005f9", "gradedAt": "2026-10-01T01:10:05.354738+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.7055555555555555, "toppings": 0.7477777777777778, "bowl": 0.8222222222222222, "broth": 0.7188888888888889, "steam": 0.7366666666666666, "composition": 0.81 }, "reward": 0.7558888888888888, "runId": "openai/gpt-6.1-sol/max", "rewardStdDev": 0.09635875289385629, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.76, "toppings": 0.79, "bowl": 0.9033333333333333, "broth": 0.75, "steam": 0.8233333333333333, "composition": 0.8866666666666667 }, "reward": 0.8173333333333334, "sampleStdDev": 0.0010408329997330585, "judgments": [ { "noodles": { "score": 0.76, "reason": "Thin continuous golden strands have rounded bends, shading and readable overlaps. The central pile is somewhat flattened and densely looped, with limited submerged volume." }, "toppings": { "score": 0.79, "reason": "Scored mushroom caps and stems, porous tofu, veined greens, corn and scallions provide readable variety. Leaves look slightly stiff and scalloped, while nori has overly straight rectangular edges." }, "bowl": { "score": 0.91, "reason": "A coherent elliptical rim, rounded walls, subtle foot and contact shadow establish convincing depth and support. Fine ceramic texture and restrained rim highlights are polished; no material construction defect is observed." }, "broth": { "score": 0.75, "reason": "Warm amber gradients, small oil marks and restrained highlights suggest appetizing soup. Ingredients have some contact shading, but immersion is shallow and the surface feels slightly opaque and flat." }, "steam": { "score": 0.82, "reason": "Normal frames show changing bends in faint translucent wisps emerging from the soup and fading upward without obscuring food. The wisps remain somewhat narrow and similarly arranged, with limited visible spreading." }, "composition": { "score": 0.89, "reason": "The single dish remains the clear focus against a restrained dark background, with balanced green and gold accents and coherent soft lighting. Mobile framing preserves the whole bowl; desktop placement is slightly bottom-heavy." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, nori, corn, scallions and noodles plausibly form vegan ramen, with no visibly animal-derived ingredients." } }, { "noodles": { "score": 0.76, "reason": "Thin continuous strands have curved bends, shading and readable overlaps beneath the toppings. The pile is somewhat flattened, with several similarly shaped loops and limited submerged depth." }, "toppings": { "score": 0.79, "reason": "Mushroom caps and stems, porous tofu, veined greens, corn and scallions are clearly detailed and varied. Leaves look slightly stiff and scalloped, while the textured nori has unusually straight, card-like edges." }, "bowl": { "score": 0.9, "reason": "A coherent elliptical rim joins smoothly to rounded walls and a supported base; fine ceramic texture, restrained highlights and a contact shadow give convincing weight. No material construction defect is observed." }, "broth": { "score": 0.75, "reason": "Warm amber shading, small oil marks and restrained surface highlights suggest appetizing soup. Ingredient contact is readable, but the opaque surface offers little immersion or translucency around the noodle pile." }, "steam": { "score": 0.82, "reason": "Normal frames show changing bends and positions in several translucent wisps rising from the soup and fading against the dark background without obscuring food. The narrow, similarly vertical plumes show limited lateral spreading." }, "composition": { "score": 0.89, "reason": "The single focal bowl, warm soup against cool dark ceramic, subtle directional lighting and uncluttered background form a polished atmospheric presentation. Mobile framing preserves the entire dish; slightly generous upper negative space is the main refinement." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, nori, corn, scallions and noodles plausibly form vegan ramen, with no visibly animal-derived ingredient." } }, { "noodles": { "score": 0.76, "reason": "Thin continuous golden strands have curved bends, shading and readable overlaps; the pile is somewhat flattened and densely looped, with limited submerged depth." }, "toppings": { "score": 0.79, "reason": "Varied toppings include porous tofu, scored mushroom caps with visible stems, veined greens, corn and textured nori. Leaves are rather uniformly scalloped and nori edges unusually straight, giving parts a cutout appearance." }, "bowl": { "score": 0.9, "reason": "Coherent elliptical rim, consistent wall curvature, subtle ceramic grain and graduated highlights establish convincing thickness and volume. The base sits in a contact shadow; no material construction defect is observed." }, "broth": { "score": 0.75, "reason": "Warm amber shading, small oil-like marks and restrained highlights read as appetizing soup. Ingredient contact is readable, but the surface is quite opaque and toppings mostly appear perched above it rather than partially immersed." }, "steam": { "score": 0.83, "reason": "Normal frames show changing bends and positions in translucent wisps emerging from the soup and fading upward without obscuring food. The plumes remain somewhat narrow and similarly shaped, with limited visible spreading." }, "composition": { "score": 0.88, "reason": "The isolated bowl is a clear focal point, with warm soup contrasting against green toppings and a restrained dark setting. Lighting is coherent and there is no clutter or text; mobile presentation leaves somewhat excessive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6133333333333333, "toppings": 0.6533333333333333, "bowl": 0.7133333333333333, "broth": 0.6133333333333333, "steam": 0.57, "composition": 0.6933333333333334 }, "reward": 0.6448333333333334, "sampleStdDev": 0.013363507523600774, "judgments": [ { "noodles": { "score": 0.62, "reason": "The noodle nest has fine, overlapping, curved strands with a plausible volume, green onion flecks and chili threads. It is small in scale and a bit busy, and the strands read as thin scribbled loops with little depth." }, "toppings": { "score": 0.68, "reason": "The shiitake caps with star cuts, bok choy, waffle-textured tofu cubes, corn and nori are all recognizable and varied. Nori sheets are rectangular and look slightly rigid and translucent, and the bok choy is glossy and cartoonish. The tofu cubes are fairly uniform." }, "bowl": { "score": 0.7, "reason": "The dark ceramic bowl has a coherent elliptical rim, a curved body with subtle texture, a foot and a soft shadow. The rim highlight is thin, and the nori and bok choy overlap the rim in an unclear way, which slightly weakens the contact." }, "broth": { "score": 0.6, "reason": "The warm amber broth has subtle oily rings and some depth gradient. It looks fairly flat, and the toppings sit on it with little ripple, occlusion or immersion." }, "steam": { "score": 0.58, "reason": "Thin translucent wisps rise and change shape across frames, so the steam is visibly animated and does not hide the food. The wisps are faint and sparse, and a bit mechanical." }, "composition": { "score": 0.7, "reason": "The bowl is centered, the dark background is restrained and the warm broth against the green gives good color contrast. The bowl sits slightly low with a lot of empty dark space, but there is no clutter or text." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, corn, nori and chili are all plant-based, and no animal ingredients are visible." } }, { "noodles": { "score": 0.62, "reason": "Fine strands form a tangled nest with believable overlap and a few red chili threads. They read as thin noodles, but the mass is small and looks somewhat scribbly and flat. It sits on top of the broth rather than being immersed in it." }, "toppings": { "score": 0.66, "reason": "Shiitake caps with white scoring, bok choy, tofu cubes, nori sheets, corn and scallion dots are all recognizable and varied. The tofu cubes look waffle-like and a bit rigid. The nori reads as cards with a faint transparent overlap, and the bok choy is glossy and slightly plasticky." }, "bowl": { "score": 0.74, "reason": "The dark ceramic bowl has a coherent elliptical rim, a curved body, a foot and subtle texture lines. Perspective is consistent. The rim highlight is subtle and the bowl contact shadow is weak. The nori and bok choy overlap the rim slightly with transparency." }, "broth": { "score": 0.62, "reason": "The warm amber miso-like broth has a gentle gradient and faint ripples. It looks fairly flat and opaque, and toppings sit on it with little immersion, occlusion or reflection." }, "steam": { "score": 0.58, "reason": "Faint translucent wisps rise and shift between frames, spreading and fading without hiding the food. They are thin and low in contrast, and a little repetitive." }, "composition": { "score": 0.68, "reason": "The single bowl is centered in a dark, restrained scene with warm food colors. The bowl sits slightly low, leaving a large empty top area. There is no clutter or text." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn, chili and noodles appear to make a vegan bowl. No egg, meat or fish is visible." } }, { "noodles": { "score": 0.6, "reason": "The noodle nest has thin, fine strands with plausible overlap and a few green scallion bits. It is small and slightly scribbly, with a repetitive thread texture, and it sits as a flat mat on the broth." }, "toppings": { "score": 0.62, "reason": "The shiitake caps with white scoring, the bok choy, the nori sheets, the corn and the chili rings are all recognizable. The tofu cubes have an odd waffle pattern, and the bok choy is somewhat glossy and plasticky. The nori looks like rigid cards, partly translucent and floating over the rim." }, "bowl": { "score": 0.7, "reason": "The dark ceramic bowl has a coherent elliptical rim, a curved body and a foot, with subtle texture and rim highlights. Its perspective is consistent. It is slightly flat, and the nori overlapping the rim looks detached." }, "broth": { "score": 0.62, "reason": "The warm amber broth has subtle oil rings and depth near the edges. It reads slightly flat and opaque, and the toppings sit on it with limited immersion or contact ripples." }, "steam": { "score": 0.55, "reason": "Faint translucent wisps rise and change shape between frames, and they do not hide the food. They are thin and low in contrast, and the plumes look somewhat similar to each other." }, "composition": { "score": 0.7, "reason": "A single centered bowl against a dark background gives a clear focal dish and warm colors with good contrast. The bowl sits somewhat low and small in the frame, with a lot of empty space above." }, "vegan": { "score": 0.95, "reason": "The visible ingredients are tofu, mushrooms, bok choy, nori, corn, scallion and chili. No animal products are visible." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7433333333333333, "toppings": 0.8, "bowl": 0.85, "broth": 0.7933333333333333, "steam": 0.8166666666666667, "composition": 0.85 }, "reward": 0.8055, "sampleStdDev": 0.012757350822173113, "judgments": [ { "noodles": { "score": 0.7, "reason": "Fine cream-colored strands form a coherent nest with believable curls and varied bends sitting in the broth; slightly thread-thin and tangle-like in spots, but clearly noodles rather than scribbles or rigid ribbons." }, "toppings": { "score": 0.8, "reason": "Varied, recognizable ingredients: grid-marked fried tofu cubes, shiitake slices with star-cut patterns, crisp bok choy halves, leaning nori sheets, corn, chili ring and scallion flecks, all organically shaped. Minor critique: the tofu grid texture is a bit repetitive and the greens have a slightly pasted outline where they overlap the rim." }, "bowl": { "score": 0.85, "reason": "Dark charcoal ceramic bowl with a thick elliptical rim, smooth tapering walls, visible foot and soft contact shadow; rim and body perspectives agree and a subtle side highlight sells the glaze. No material geometric defect observed." }, "broth": { "score": 0.78, "reason": "Warm tan broth with a soft central sheen, darker meniscus ring near the rim, oil specks and gentle texture; ingredients cast small shadows and appear immersed rather than floating. Slightly flat/uniform in tone at the edges." }, "steam": { "score": 0.8, "reason": "Translucent tapering wisps rise from the soup and fade without hiding the food. Comparing desktop-1/2/3 and the pointer frames, the wisp shapes and positions shift continuously (and parallax with pointer movement), while reduced-motion frames hold steady, confirming real animated steam." }, "composition": { "score": 0.85, "reason": "Centered focal bowl on a restrained dark vignette with a warm backlight glow; balanced proportions, coherent lighting and no clutter, text or competing elements. Mobile framing also stays clean with no overflow." }, "vegan": { "score": 1, "reason": "Visible toppings are tofu, shiitake mushrooms, bok choy, nori, corn, scallion and chili in plant broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.78, "reason": "Thin yellow strands form a believable tangled nest with varied curvature and scallion flecks tucked in; the mass is slightly jumbled and strands are uniformly thin, but they read as supple noodles, not wire or maze scribbles." }, "toppings": { "score": 0.8, "reason": "Good variety and detail: shiitake caps with cross-cut asterisks, fried tofu puffs with scored grid, leafy greens, standing nori sheets, corn, chili rings and threads. Tofu grid pattern is a bit repetitive and greens slightly glossy, but all ingredients are recognizable and food-like." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with clean elliptical rim, visible rim thickness, textured charcoal walls and a proper footed base with soft contact shadow; perspective is consistent from rim to base with only minor stiffness in the silhouette." }, "broth": { "score": 0.8, "reason": "Warm tonal gradient with restrained sheen, subtle oil specks and darker edge ring; toppings sit in the surface with contact shadows rather than floating, though the liquid reads slightly flat/painted in the mid-region." }, "steam": { "score": 0.85, "reason": "Translucent wisps rise above the bowl and visibly change shape and position across desktop frames 1-3 and the pointer frames, indicating continuous animated motion; they stay irregular and never obscure the food." }, "composition": { "score": 0.85, "reason": "Centered bowl with warm backlight glow against a dark atmospheric background, balanced proportions, no text or clutter; the dish is the clear focal point and it adapts cleanly to the mobile viewport." }, "vegan": { "score": 1, "reason": "Visible toppings are tofu, shiitake mushrooms, greens, nori, corn, chili and scallion in what appears to be plant broth; no animal-derived ingredients are depicted." } }, { "noodles": { "score": 0.75, "reason": "Central nest of thin, curvy strands with readable individual lines, plausible bends and broth immersion; slightly uniform/tangled in places but not rigid or maze-like." }, "toppings": { "score": 0.8, "reason": "Good variety and readability: crosshatched tofu cubes, shiitake caps with star cuts, leafy greens, nori sheets, corn, chili ring and scallion bits; tofu cubes are somewhat repeated and waffle-like, and greens are slightly blobby." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl with elliptical rim, visible wall thickness, footed base, consistent perspective, subtle highlights and contact shadow on the ground." }, "broth": { "score": 0.8, "reason": "Appetizing brown liquid with gradient depth, restrained sheen swirls and oil speckles; ingredients sit in the surface, though the corn cluster at the edge reads slightly pasted." }, "steam": { "score": 0.8, "reason": "Across desktop-1/2/3 and pointer frames the translucent wisps change shape and position, indicating continuous rising motion; irregular, non-opaque and does not obscure the food." }, "composition": { "score": 0.85, "reason": "Centered, well-proportioned dish with warm rim lighting against a dark atmospheric backdrop, no clutter or UI, food remains the clear focal point." }, "vegan": { "score": 1.0, "reason": "Visible toppings are tofu, mushrooms, greens, nori, corn, chili and scallion in plant broth; no animal-derived ingredients apparent." } } ], "diagnostics": { "vegan": 1.0 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "0ea392821f305148805893872fe4aaa57b3f529dc9d6242c66d13117f288cb24", "gradedAt": "2026-10-01T02:12:18.195459+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5966666666666667, "toppings": 0.7522222222222222, "bowl": 0.8066666666666666, "broth": 0.6555555555555556, "steam": 0.4288888888888889, "composition": 0.8244444444444444 }, "reward": 0.6810555555555555, "runId": "openai/gpt-6.1-sol/ultra", "rewardStdDev": 0.07397490815835489, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.78, "bowl": 0.87, "broth": 0.6333333333333333, "steam": 0.38666666666666666, "composition": 0.8733333333333333 }, "reward": 0.6776666666666666, "sampleStdDev": 0.036675377753110236, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin golden strands have readable bends and overlaps, but repeated elongated loops form a flat, densely outlined mat rather than a supple volume immersed in soup." }, "toppings": { "score": 0.78, "reason": "Veined greens, porous tofu, scored mushroom caps with gill details, corn and textured nori provide convincing variety. Leaves are somewhat flat and uniformly scalloped, and the upright stems look stiff." }, "bowl": { "score": 0.87, "reason": "A coherent elliptical rim, rounded deep walls, attached foot and fine ceramic texture establish a convincing vessel. Warm highlights and darker right-side shading reinforce its volume; the foot's contact shadow is slightly diffuse." }, "broth": { "score": 0.64, "reason": "Warm amber color and scattered oil-like flecks suggest appetizing broth, but the surface has limited liquid sheen and depth. Ingredients mostly sit above it with little visible immersion or contact meniscus." }, "steam": { "score": 0.66, "reason": "Faint translucent wisps above the soup change shape across the normal desktop frames and leave the food unobscured. Their low contrast and diffuse emergence make the rising motion difficult to read clearly." }, "composition": { "score": 0.88, "reason": "The single dish remains a clear focal point against a restrained dark background, with balanced topping groups and coordinated warm gold, brown and green colors. No text or clutter distracts; the mobile framing leaves somewhat excessive upper space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, corn, scallions, nori and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands have readable bends and overlaps, but the dense, predominantly flat horizontal loops resemble a patterned mat rather than a supple immersed mound." }, "toppings": { "score": 0.78, "reason": "Veined greens, porous tofu, scored mushroom caps with edge gills, corn and textured nori are clearly differentiated. Leaves and stems remain stiff and diagrammatic, and mushroom shapes are somewhat repetitive." }, "bowl": { "score": 0.87, "reason": "Consistent elliptical rim, rounded walls, attached foot and directional shading create convincing depth. Fine horizontal ceramic texture and layered rim highlights are polished; the foot's contact with the soft shadow is slightly weak." }, "broth": { "score": 0.63, "reason": "Warm amber coloring and small oil-like flecks make the liquid readable, but its surface is rather flat and matte, with limited meniscus or wet contact around the ingredients." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps rise behind the toppings without obscuring them, but their shapes appear unchanged across the three normal desktop frames, indicating static steam." }, "composition": { "score": 0.88, "reason": "A single clearly framed dish, warm food colors against a restrained dark background, and coherent left-side illumination establish a strong focus without clutter. Mobile framing retains the whole bowl, though it leaves substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, scallions, nori and noodles plausibly form vegan ramen; no visibly animal-derived topping is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands have readable outlines and overlapping bends, but the mass is dominated by tightly packed, repeated flat loops with little volume or immersion." }, "toppings": { "score": 0.78, "reason": "Veined greens, porous tofu, scored mushroom caps with edge details, corn and textured nori provide clear variety. Leaves are unusually uniform and upright, and the nori remains rather rigid." }, "bowl": { "score": 0.87, "reason": "A coherent elliptical rim, shaded curved walls, fine ceramic texture and attached foot convincingly establish depth and material. The foot's contact with the broad shadow is slightly weak." }, "broth": { "score": 0.63, "reason": "Warm brown liquid and scattered oil-like flecks read as broth, but the surface has little reflective depth and most ingredients appear laid on top rather than partially immersed." }, "steam": { "score": 0.25, "reason": "Faint translucent wisps rise behind the greens without obscuring food, but their shapes appear unchanged across the normal time-ordered desktop frames." }, "composition": { "score": 0.86, "reason": "The single dish is clearly framed against a restrained dark background, with warm directional shading and balanced green and gold accents. No clutter or text distracts; the mobile framing leaves somewhat excessive upper empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy vegetables, corn, scallions, nori and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6066666666666667, "toppings": 0.6933333333333334, "bowl": 0.7, "broth": 0.5666666666666667, "steam": 0.2833333333333333, "composition": 0.75 }, "reward": 0.6088333333333333, "sampleStdDev": 0.008892881047969389, "judgments": [ { "noodles": { "score": 0.6, "reason": "Noodles are curly, overlapping strands with a plausible tan color and volume, but they look like repeated flat loops and scribbles, partly maze-like, and they sit in a small patch with the ends cut off at the bottom." }, "toppings": { "score": 0.7, "reason": "Toppings are varied and recognizable: bok choy with veined leaves and pale stems, speckled tofu cubes, shiitake with cross-cut scoring, crinkled nori and corn. The nori is a fairly rigid block and the tofu is a bit boxy, but the textures are good." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent rim, cavity, curved wall with fine ceramic banding, and a foot. The foot is small and reads a bit detached, the rim is a flat ring, and the shadow beneath is faint, but the perspective is consistent." }, "broth": { "score": 0.55, "reason": "The broth is an orange-brown surface with oil speckles and some depth near the rim, but it reads fairly flat and opaque. Ingredients sit on top with little immersion or reflection." }, "steam": { "score": 0.25, "reason": "Steam is barely visible, as faint haze above the bowl. The frames show little change, so it reads as static and very weak." }, "composition": { "score": 0.75, "reason": "The bowl is centered on a dark, restrained background with warm colors and a clear focal dish. There is no clutter, and the toppings are balanced, though slightly crowded toward the back and right." }, "vegan": { "score": 0.95, "reason": "Only tofu, mushrooms, bok choy, corn, nori, noodles and scallions are visible, with no animal products." } }, { "noodles": { "score": 0.6, "reason": "Noodles are thin, curly strands with plausible overlap and a nice cream color. They look like repeated flat loops and scribbles, with a small clustered mass and no real volume or sense of being lifted from the broth." }, "toppings": { "score": 0.68, "reason": "Toppings are varied and readable: bok choy with veins, tofu cubes with speckles, shiitake with cross scoring, crinkled nori and corn. Stylized and somewhat flat, and the nori is a blocky rectangle, but no plastic spheres." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent rim, curved body with ceramic ridge texture, a foot, and a contact shadow. The rim looks like a lid or tray with a flat inner edge, and the foot is small and slightly detached." }, "broth": { "score": 0.55, "reason": "Broth is an orange-brown surface with speckled oil spots, which reads as miso broth. It is fairly flat, with little depth or reflection, and toppings sit on it without much immersion." }, "steam": { "score": 0.3, "reason": "Steam is very faint, with barely visible wisps above the bowl that change only slightly across frames. It is too subtle to read as steam and mostly static." }, "composition": { "score": 0.75, "reason": "Centered bowl on a dark background with a good focal point, warm palette and soft shadow. Clean and uncluttered, with no distracting UI." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn and noodles are all plant-based. No meat or egg is visible." } }, { "noodles": { "score": 0.62, "reason": "Fine, curly golden strands overlap with plausible volume and bends, but they read as repeated flat loops, sit mostly on top of the broth, and form a somewhat scribbly mass." }, "toppings": { "score": 0.7, "reason": "Bok choy leaves with veins, seasoned tofu cubes, shiitake with cross scoring, nori and corn are all readable and varied. The nori is a dark, blocky slab and the mushrooms look a bit like discs." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, curved wall with ceramic banding, and a foot with a ground shadow. The foot looks small and slightly detached, and the rim is thick with a faint inner ring." }, "broth": { "score": 0.6, "reason": "Warm orange broth with oil droplets reads as miso-like, but it is flat and opaque with little depth. Toppings sit on top rather than being immersed, with weak contact." }, "steam": { "score": 0.3, "reason": "Very faint wisps are visible above the bowl, and the desktop frames show almost no visible change. The steam is barely perceptible and effectively static." }, "composition": { "score": 0.75, "reason": "A centered bowl on a dark background with warm colors and good focus. Lighting is coherent, with no clutter or text. The toppings are crowded to one side." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, bok choy, nori, corn and noodles are all plant-based. No animal ingredients are visible." } } ], "diagnostics": { "vegan": 0.95 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6833333333333333, "toppings": 0.7833333333333333, "bowl": 0.85, "broth": 0.7666666666666667, "steam": 0.6166666666666667, "composition": 0.85 }, "reward": 0.7566666666666667, "sampleStdDev": 0.03449033681095811, "judgments": [ { "vegan": { "score": 1, "reason": "Tofu cubes, shiitake mushrooms, corn, leafy greens, nori, scallions and chili threads only; no animal-derived toppings visible." }, "toppings": { "score": 0.75, "reason": "Varied and readable: speckled tofu cubes, bok choy leaves with pale stems, jagged nori sheet, corn cluster, scored shiitake caps, scallion rings. Mushrooms and leaves are slightly flat and stylized, but organic and well-textured." }, "steam": { "score": 0.75, "reason": "Thin translucent wisps rise above the bowl and differ in shape and position across desktop-1/2/3 (~1.6s apart), supporting continuous drift and fade; subtle enough not to obscure food, though quite sparse and line-like." }, "noodles": { "score": 0.7, "reason": "Continuous pale strands with varied bends, curls and believable broth placement, garnished with scallion rings and chili thread. Slightly uniform thickness and a somewhat repeated loop pattern flatten the mass." }, "composition": { "score": 0.85, "reason": "Single centered focal bowl on a dark atmospheric backdrop, warm broth against cream ceramic, soft ground shadow, steam adding depth; no clutter, text or competing elements. Same presentation holds on mobile." }, "broth": { "score": 0.75, "reason": "Rich brown-orange liquid with soft lightening toward the rim, subtle oil speckle texture, and shadowing that seats mushrooms and tofu in the surface. Gradient is slightly smooth and uniform, but reads as liquid." }, "bowl": { "score": 0.85, "reason": "Coherent elliptical rim with inner lip highlight, smoothly tapering speckled ceramic body, proper pedestal foot and grounded contact shadow; consistent perspective and material throughout with no visible distortion." } }, { "vegan": { "score": 1, "reason": "Toppings are tofu cubes, shiitake mushrooms, corn, leafy greens, scallions, chili and nori over noodle broth; no animal-derived ingredients visible." }, "toppings": { "score": 0.85, "reason": "Well-drawn stylized ingredients: scored shiitake caps with gills, speckled tofu cubes, bok-choy-like leaves with veins, corn cluster, textured nori sheet with irregular top edge, scallion rings and chili threads. Minor shortcoming: corn and mushroom scoring are slightly graphic/repetitive." }, "steam": { "score": 0.7, "reason": "Faint translucent gray wisps rise from the soup and visibly change shape and position across the time-ordered frames (barely visible in desktop-1, stronger right-side plumes in desktop-3 and pointer frames), so it is animated, not static. Wisps are somewhat faint and generic vertical plumes." }, "noodles": { "score": 0.65, "reason": "Pale yellow strands with waviness, varied direction and overlapping curls, garnished with scallion bits. However many strands are short wavy strokes that read slightly scribbly and repetitive, and the noodle patch boundary against the broth at the left is abrupt." }, "composition": { "score": 0.85, "reason": "Clean centered presentation on a dark atmospheric background with soft contact shadow, warm lighting focused on the bowl, no text or clutter; the dish is clearly the focal point in both desktop and mobile framing." }, "broth": { "score": 0.8, "reason": "Rich orange-brown broth with speckled oil/chili texture and edge gradient; toppings sit in the liquid with soft shadows and believable immersion. Slightly flat in the larger open areas with a somewhat uniform speckle pattern." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl: elliptical rim with visible lip thickness, consistent perspective, speckled glazed body with highlight and horizontal texture, supported foot and grounded shadow. Minor nit: body silhouette is quite spherical and the surface banding is a touch regular." } }, { "bowl": { "reason": "Coherent ceramic bowl with consistent rim, curved wall, speckled texture, footed base and contact shadow; no perspective breaks.", "score": 0.85 }, "broth": { "reason": "Rich orange broth with speckled surface and believable immersion of toppings; no flat target discs, though surface reads slightly flat in places.", "score": 0.75 }, "composition": { "reason": "Balanced centered dish on restrained dark background, attractive warm palette, no clutter or UI; food remains the focus.", "score": 0.85 }, "noodles": { "reason": "Continuous yellow strands with plausible bends and scallion garnish, but some stiff, ribbon-like flat loops dominate the center mass.", "score": 0.7 }, "steam": { "reason": "Faint translucent wisps rise above the bowl but frames are near-identical across temporal captures, indicating static steam.", "score": 0.4 }, "toppings": { "reason": "Good variety: tofu cubes, leafy greens, corn, nori sheet, scored shiitake, scallions and chili; mushrooms read slightly sticker-like but vegetables are convincing.", "score": 0.75 }, "vegan": { "reason": "All visible ingredients are plant-based: tofu, mushrooms, greens, corn, nori, scallions; no animal-derived toppings.", "score": 1 } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9833333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.25, "frameIntervalP95Ms": 33.399999999999636, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "ffb685e35854c7374f9c053053f8fea40f3a4ba8f2ff45e0d5cbeb238dc29f55", "gradedAt": "2026-10-06T15:53:11.269279+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.40444444444444444, "toppings": 0.4444444444444444, "bowl": 0.548888888888889, "broth": 0.4722222222222222, "steam": 0.5877777777777777, "composition": 0.5255555555555556 }, "reward": 0.4897222222222222, "runId": "mistralai/mistral-large-4-0/default", "rewardStdDev": 0.06174731962993602, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.5, "bowl": 0.6466666666666667, "broth": 0.45, "steam": 0.68, "composition": 0.6266666666666667 }, "reward": 0.56, "sampleStdDev": 0.00476969600708475, "judgments": [ { "noodles": { "score": 0.48, "reason": "Pale continuous strands are recognizable, but mostly straight parallel lengths with a few simple arches; they lack supple tangling, volume, and convincing immersion." }, "toppings": { "score": 0.5, "reason": "Scallion rings, corn, tofu, and brown mushroom-like slices provide variety, but remain smooth geometric symbols. The nori resembles a thick rounded card, with little organic food texture." }, "bowl": { "score": 0.64, "reason": "The elliptical opening, curved walls, decorative bands, and ground shadow establish a readable bowl. Angular diagonal shading and a pointed lower-right silhouette weaken the otherwise coherent ceramic form." }, "broth": { "score": 0.43, "reason": "Warm brown coloration suggests broth, but the surface is nearly featureless and solid-looking, with little liquid sheen or depth; ingredients appear laid over it rather than partially immersed." }, "steam": { "score": 0.67, "reason": "Normal frames show changing translucent wisps above the soup without obscuring food. Their narrow, similarly sinusoidal shapes remain mechanical, with limited visible spreading and dissolution." }, "composition": { "score": 0.62, "reason": "Restrained warm colors and a clean dark background keep the dish readable. The bowl sits very low with excessive empty space, especially on mobile, and an isolated floating leaf distracts from the dish." }, "vegan": { "score": 0.95, "reason": "Visible tofu-like white pieces, mushrooms, corn, scallions, greens, nori, and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Pale strands are continuous and readable, but mostly form straight parallel strokes with a few uniform bends; little tangled volume or convincing immersion is shown." }, "toppings": { "score": 0.5, "reason": "Mushrooms, tofu, corn and scallion rings provide recognizable variety, but use smooth disks and simple blocks. The nori resembles a thick rounded card rather than thin irregular seaweed." }, "bowl": { "score": 0.66, "reason": "The elliptical opening, curved walls, decorative bands and foot establish a coherent bowl. Angular dark shading and sparse rim thickness make the material and lower silhouette less convincing." }, "broth": { "score": 0.46, "reason": "Warm brown coloring suggests soup, but the surface is largely a flat opaque ellipse with little liquid sheen or depth; ingredients appear laid on top rather than partially immersed." }, "steam": { "score": 0.68, "reason": "Normal frames show changing translucent wisps above the soup without obscuring it. Their narrow, similarly sinusoidal shapes have limited spreading and a somewhat mechanical appearance." }, "composition": { "score": 0.62, "reason": "Restrained warm colors and a dark uncluttered background keep the dish readable. Excessive empty space, especially on mobile, places the small bowl very low, while an isolated floating leaf distracts from it." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, corn, scallions, leafy greens and nori alongside noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.48, "reason": "Yellow strands are readable and continuous, but mostly straight parallel strokes with a few simple arches; they lack supple tangling, overlap, and substantial noodle volume." }, "toppings": { "score": 0.5, "reason": "Mushrooms, tofu, scallion rings, corn, and a veined leaf provide variety, but remain simplified disks and blocks. The nori resembles a thick rounded rectangular card rather than irregular thin seaweed." }, "bowl": { "score": 0.64, "reason": "The elliptical opening, curved walls, decorative bands, and foot establish a coherent bowl. Angular shading and the pointed lower-right silhouette weaken its rounded ceramic volume, while rim thickness is poorly defined." }, "broth": { "score": 0.46, "reason": "Warm brown color suggests broth, but the surface is largely a flat opaque ellipse with little liquid sheen or depth. Ingredients appear layered on top rather than convincingly immersed." }, "steam": { "score": 0.69, "reason": "Normal frames show changing positions, heights, and visibility of translucent wisps above the soup without obscuring it. The plumes remain narrow, similarly sinusoidal lines with limited spreading or irregular breakup." }, "composition": { "score": 0.64, "reason": "The restrained dark palette and warm focal lighting keep the dish readable without text or clutter. Excessive upper empty space, especially on mobile, pushes the small dish low, and the isolated floating leaf distracts." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, corn, scallions, greens, and seaweed alongside noodles; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.36666666666666664, "toppings": 0.4166666666666667, "bowl": 0.4166666666666667, "broth": 0.45, "steam": 0.6, "composition": 0.48333333333333334 }, "reward": 0.4441666666666667, "sampleStdDev": 0.015069284433354257, "judgments": [ { "noodles": { "score": 0.35, "reason": "Noodles are a few thin, flat yellow strands and one small loop on the surface. There is no volume or noodle mass, and they look sparse and wire-like." }, "toppings": { "score": 0.45, "reason": "The toppings are recognizable: tofu cube, nori rectangle, shiitake slices, scallion rings, corn and a leaf. They are tiny and simple, and the nori is a rigid card. The leaf floats away from the bowl in the normal frames." }, "bowl": { "score": 0.4, "reason": "The bowl has a rim and a decorative band, but a large dark diagonal wedge cuts across the body and reads as a geometry defect. The bowl is also small and has little material detail." }, "broth": { "score": 0.4, "reason": "The broth is a flat orange ellipse with a mild gradient. It has little depth, texture or immersion of the ingredients, so it reads like a disc." }, "steam": { "score": 0.6, "reason": "Thin translucent wavy wisps rise, drift and fade, and their positions change between frames. They are fairly uniform and thin, but they do not hide the food." }, "composition": { "score": 0.5, "reason": "The dark warm atmosphere is pleasant and the bowl is the focus. The bowl is small, off-center and low in a large empty frame, and the leaf floating on its own is distracting." }, "vegan": { "score": 0.95, "reason": "Tofu, mushrooms, nori, scallions, corn and a leaf are all plant-based. No animal ingredients are visible." } }, { "noodles": { "score": 0.35, "reason": "Noodles are small thin pale strands and a loop barely visible at the top of the broth; they are sparse and cramped and read as scribbles rather than a supple volume in the soup." }, "toppings": { "score": 0.4, "reason": "Tofu cube, nori card, mushrooms, scallion rings and corn are recognizable but tiny and generic. The nori is a rigid dark rounded card, and the floating leaf is detached from the bowl." }, "bowl": { "score": 0.4, "reason": "The bowl has a rim and a base ellipse, but a large hard-edged diagonal dark wedge cuts across the body and looks like a rendering error. The shading is inconsistent and the form reads as flat." }, "broth": { "score": 0.45, "reason": "The orange broth surface has a warm gradient and a rim glint, but it is a flat, thin ellipse with little depth. Toppings sit on top more than in it." }, "steam": { "score": 0.6, "reason": "Thin translucent wisps rise, sway and change positions across frames, and they do not hide the food. They are slender and fairly uniform, though." }, "composition": { "score": 0.45, "reason": "The dark moody scene with a warm glow focuses on the bowl, but the bowl is small and sits low, leaving a lot of empty space. A stray leaf floats far to the left." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, nori, scallions, corn and noodles appear vegan, with no visible animal products." } }, { "noodles": { "score": 0.4, "reason": "Only a few thin pale strands and a small curl are visible on the surface. They are partly hidden under chopsticks and toppings, with little volume and no sense of immersion in the broth." }, "toppings": { "score": 0.4, "reason": "Tofu cube, shiitake slices, nori, scallion rings and corn are recognizable but tiny and simplistic. The nori is a rigid dark card standing upright, and the toppings are sparse and mostly flat symbols." }, "bowl": { "score": 0.45, "reason": "The bowl has a rim ellipse, a curved body and a base with a shadow. A hard diagonal dark wedge cuts across the body and looks like a geometry or shading error. The band decoration is thin." }, "broth": { "score": 0.5, "reason": "The orange surface has a warm gradient and a slight sheen, so it reads as a liquid. It is fairly flat, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "The wisps are translucent and wavy, drift and change between frames, and don't hide the food. They are thin and similar to each other, and a few look like detached streaks." }, "composition": { "score": 0.5, "reason": "The dark, warm atmosphere is moody and the bowl is the focal point. The bowl is small and the frame is mostly empty. A stray floating leaf in the upper left is distracting." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, scallions, corn and nori are all plant-based, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.36666666666666664, "toppings": 0.41666666666666663, "bowl": 0.5833333333333334, "broth": 0.5166666666666666, "steam": 0.48333333333333334, "composition": 0.4666666666666667 }, "reward": 0.465, "sampleStdDev": 0.17385338650713705, "judgments": [ { "noodles": { "score": 0.25, "reason": "No readable noodle strands are visible; at most an indistinct mass inside the small dark bowl, far below the legibility needed for credit." }, "toppings": { "score": 0.25, "reason": "Only a few tiny yellow and green specks can be discerned on the broth surface; no recognizable vegetable, mushroom, tofu or nori forms are readable at this scale." }, "bowl": { "score": 0.5, "reason": "A shallow bowl with a coherent rim and a soft base reflection is present and sits plausibly on the surface, but the dish is very small in frame and its wall/base detail is nearly lost in darkness, making construction only crudely verifiable." }, "broth": { "score": 0.45, "reason": "An orange, warm-lit liquid surface glows in the bowl, which reads as broth, but it is murky and there is no convincing immersion or occlusion of ingredients into the liquid." }, "steam": { "score": 0.35, "reason": "Thin gray wisps rise above the bowl and taper with height, which is a plausible steam silhouette, but they read as dark, opaque smoke-like streaks rather than translucent steam; across the time-ordered frames their shape appears essentially unchanged, suggesting static rather than flowing wisps." }, "composition": { "score": 0.4, "reason": "A dark atmospheric scene with a single dish and no clutter or text is a coherent idea, but the focal bowl is disproportionately tiny in the frame, most detail is swallowed by darkness, and an unexplained green glowing orb floats at the left edge as a distracting artifact." }, "vegan": { "score": 1.0, "reason": "No animal-derived ingredients are visible; only vegetable-like specks and noodles in broth appear, consistent with a vegan ramen." } }, { "noodles": { "score": 0.25, "reason": "No readable noodle strands are visible; the bowl contents read as an indistinct mass with only a faint pale swirl, no coherent strand structure or volume." }, "toppings": { "score": 0.3, "reason": "Only tiny green flecks and an ambiguous pale spiral are discernible; ingredients are too small and undefined to read as recognizable vegetables, mushrooms, or seaweed." }, "bowl": { "score": 0.55, "reason": "A shallow dark-rimmed bowl with a warm orange interior is recognizable with a plausible silhouette and soft glow, but the dish is rendered so small that rim thickness, wall depth, and base/contact detail cannot be convincingly verified." }, "broth": { "score": 0.5, "reason": "An orange liquid surface with a gentle sheen and some textural variation is present, but at this scale immersion and occlusion of ingredients are not believable or readable." }, "steam": { "score": 0.4, "reason": "Thin vertical wisps rise from the bowl in all frames, but they appear as repeated thin strands rather than translucent spreading plumes, and no clear evolution is discernible across the time-ordered frames." }, "composition": { "score": 0.35, "reason": "The dish occupies only a small fraction of a large dark empty frame, with a distracting unexplained green glowing orb floating in the upper area, leaving the focal food underpowered and the scene sparse." }, "vegan": { "score": 0.5, "reason": "No clearly animal-derived toppings are visible, but the pale spiral in the broth resembles a narutomaki fish cake; the appearance is ambiguous at this scale, so only partial credit is warranted." } }, { "noodles": { "score": 0.6, "reason": "Yellow wavy noodle strands peeking under toppings read as noodles with plausible bends, but they are thin flat ribbons mostly hidden, with little volume or coherent overlap." }, "toppings": { "score": 0.7, "reason": "Recognizable tofu cube with a red dot, naruto-like white slice, corn kernels, scallion rings, dark nori sheet, mushroom slices and chopsticks; stylized and readable though simple flat shapes." }, "bowl": { "score": 0.7, "reason": "Coherent ellipsoid rim, warm orange body with decorative band, dark walls and base with a soft ground shadow; a subtle mismatch between rim width and body silhouette, but construction is believable for the flat style." }, "broth": { "score": 0.6, "reason": "Orange liquid surface with restrained sheen and rim highlight; toppings appear to sit on the surface, but the broth is fairly flat with little depth or immersion cue." }, "steam": { "score": 0.7, "reason": "Translucent wavy wisps whose shapes clearly differ across the desktop frames, rising and fading without obscuring the food; somewhat mechanical vertical squiggles." }, "composition": { "score": 0.65, "reason": "Attractive dark atmospheric scene with the bowl as focal point, warm glow and no clutter; a stray floating green leaf near the top-left is a distracting unexplained element, and generous empty space at top." }, "vegan": { "score": 0.9, "reason": "Visible toppings are tofu, corn, scallions, nori, mushrooms and noodles; no animal ingredients apparent." } } ], "diagnostics": { "vegan": 0.8 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 16.699999999999818, "frameIntervalP95Ms": 16.700000000000728, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8ed7459fe0a778c4d5d347aa13799988bc1adbd49991caca5287bbd72c4c55ad", "gradedAt": "2026-10-08T21:12:35.218637+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5577777777777778, "toppings": 0.5522222222222223, "bowl": 0.7822222222222223, "broth": 0.5055555555555555, "steam": 0.6488888888888888, "composition": 0.7966666666666666 }, "reward": 0.6336111111111111, "runId": "anthropic/haiku-5.5/medium", "rewardStdDev": 0.06990814078594963, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.47333333333333333, "toppings": 0.5166666666666667, "bowl": 0.79, "broth": 0.4666666666666667, "steam": 0.5733333333333333, "composition": 0.8333333333333333 }, "reward": 0.5978333333333333, "sampleStdDev": 0.005484827557301472, "judgments": [ { "noodles": { "score": 0.48, "reason": "Golden strands have consistent thickness and some curved overlaps, but the dense, mostly straight crisscross lattice reads as flat rigid ribbons rather than supple noodles." }, "toppings": { "score": 0.5, "reason": "Tofu, mushroom-like caps, scallion rings and a green leaf are recognizable, but simplified rounded tiles and disks dominate; the perfectly rectangular striped nori lacks organic texture." }, "bowl": { "score": 0.79, "reason": "The elliptical rim joins a coherent deep curved body, with consistent thickness and restrained teal shading. The broad graphic highlight and weakly established base/contact limit material conviction." }, "broth": { "score": 0.46, "reason": "Amber patches establish soup beneath the noodles, but the surface is largely flat with little liquid sheen or depth; toppings appear layered atop it rather than visibly immersed." }, "steam": { "score": 0.58, "reason": "Normal frames show changing soft translucent cloud shapes above the soup without obscuring food. However, bright diffuse blobs dominate rather than distinct rising, spreading wisps; continuous flow cannot be established from these samples." }, "composition": { "score": 0.84, "reason": "A clear single-dish focal point, dark teal and golden palette, generous negative space and diagonal chopsticks create a balanced atmospheric presentation across desktop and mobile. No material compositional defect or distracting UI is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, seaweed and green vegetables alongside noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.46, "reason": "Numerous continuous yellow strands are readable, but mostly form a flat lattice of straight ribbons and angular bends, with little supple curvature or volumetric overlap." }, "toppings": { "score": 0.55, "reason": "Tofu, radially marked mushrooms, scallion rings and a green leaf provide recognizable variety. Shapes remain schematic: tofu resembles rounded tiles, the leaf lacks organic detail, and nori is a rigid striped rectangle." }, "bowl": { "score": 0.79, "reason": "The elliptical rim, thick lip and rounded walls join coherently, with dark teal shading and a curved highlight conveying depth. The cavity is shallow-looking and the base has weak visible contact with its surroundings." }, "broth": { "score": 0.48, "reason": "Amber liquid is visible between strands, with small surface flecks. It reads largely as a flat fill, lacking liquid sheen and convincing immersion around the toppings." }, "steam": { "score": 0.57, "reason": "Normal frames show changing positions and shapes of softly translucent vapor above the soup without obscuring food. Bright rounded glow patches dominate rather than distinct rising, spreading wisps; continuous flow is not established by these captures." }, "composition": { "score": 0.83, "reason": "A single clear focal bowl, warm food against dark teal ceramic, and diagonal chopsticks create a balanced atmospheric presentation without text or clutter. Mobile preserves the dish cleanly, though with considerable empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, seaweed and greens alongside noodles; no visibly animal-derived ingredient or material compliance defect is observed." } }, { "noodles": { "score": 0.48, "reason": "Golden strands are distinct and overlap, but the dense, mostly straight crisscrossing ribbons lack supple bends and convincing piled volume." }, "toppings": { "score": 0.5, "reason": "Tofu, radial-textured mushrooms, scallion rings and a green leaf are recognizable, but simplified rounded tiles and a rigid striped rectangular nori sheet dominate the depiction." }, "bowl": { "score": 0.79, "reason": "The elliptical rim joins a coherent deep curved body with consistent dark teal shading and a restrained highlight. The base and supporting contact are weakly defined." }, "broth": { "score": 0.46, "reason": "Amber liquid is visible between noodles with a few bright flecks, but its surface is largely flat and ingredients appear layered on top rather than convincingly immersed." }, "steam": { "score": 0.57, "reason": "Normal frames show changing soft clouds above the soup without obscuring the food. The steam resembles bright diffuse blobs more than thin rising wisps, and continuous rise and fade are not clearly established." }, "composition": { "score": 0.83, "reason": "A single readable dish, diagonal golden chopsticks and teal-versus-amber contrast create a balanced dark presentation without text or clutter. Mobile framing remains intact; the large empty upper region slightly weakens balance." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, greens, nori and noodles plausibly form vegan ramen. No visibly animal-derived ingredient or material compliance defect is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.55, "toppings": 0.5233333333333333, "bowl": 0.7066666666666667, "broth": 0.45, "steam": 0.6066666666666667, "composition": 0.7066666666666667 }, "reward": 0.5888333333333333, "sampleStdDev": 0.009278110439811225, "judgments": [ { "noodles": { "score": 0.55, "reason": "Dense yellow strands with some overlap and bends fill the surface, reading as ramen. They are thin and fairly uniform, with repeated short flat segments, and the mass is somewhat indistinct." }, "toppings": { "score": 0.5, "reason": "Nori, shiitake, tofu cubes, a leaf and scallion rings are all recognizable. The tofu blocks are rounded rigid cards, the scallion rings are glossy beads, and the mushrooms are partly hidden. The set is varied but generic." }, "bowl": { "score": 0.7, "reason": "The dark teal bowl has a coherent rim, a curved body with a gradient and a side highlight, and a grounding shadow. The ellipse perspective is consistent. The base is hidden and the material is a bit flat. No major distortion." }, "broth": { "score": 0.45, "reason": "The broth is mostly covered by noodles, with only brown patches visible at the edges. There is little sheen or depth, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent glowing plumes rise above the bowl and change shape between frames. The glow is blobby, and some bright hotspots read as light blooms rather than wisps." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point on a dark, atmospheric background with warm lighting. The chopsticks add a diagonal. There is no clutter or text. The bowl is slightly large and the chopsticks float, unsupported." }, "vegan": { "score": 0.9, "reason": "Mushrooms, tofu, nori, a leaf and scallions are all plant-based. No visible meat, egg or fish." } }, { "noodles": { "score": 0.55, "reason": "Many thin golden strands fill the surface with overlap and some curves, so they read as noodles. They look like fine straw-like scribbles with flat repeated texture, and the strand definition is limited." }, "toppings": { "score": 0.55, "reason": "Nori, shiitake, tofu cubes, a green leaf and scallion rings are all recognizable and varied. The tofu blocks are rigid rounded cards and the scallion dots look like beads, so the depiction is somewhat generic." }, "bowl": { "score": 0.72, "reason": "The dark teal bowl has a coherent elliptical rim, a rounded body, a smooth gradient and a rim highlight, which give it convincing depth. The base looks a bit floating, and the interior wall is thin, but there is no material distortion." }, "broth": { "score": 0.45, "reason": "The broth is mostly hidden under noodles, with only brown patches visible at the edges. It has little sheen or depth, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent warm plumes rise above the bowl and change shape between frames, with glowing blobs that drift. They look somewhat blobby and glow-like, and sit above the chopsticks, so they read as slightly artificial." }, "composition": { "score": 0.72, "reason": "A centered bowl on a dark background with warm lighting makes the food the clear focal point. The chopsticks add a diagonal, and there is no clutter. The lower half is mostly the bowl, which leaves the food area small." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are tofu, mushrooms, nori, a leaf, scallions and noodles, and none are clearly animal-derived. The tofu is slightly ambiguous in appearance." } }, { "noodles": { "score": 0.55, "reason": "Dense golden strands fill the surface and read as noodles with some overlap, but they are thin, short, and uniform, close to a scribbly mass. Little volume or sense of strands sinking into the broth." }, "toppings": { "score": 0.52, "reason": "Nori, shiitake, tofu blocks, a leaf and scallion rings are all recognizable. The tofu are rigid rounded cards, and the scallion rings and red dots look like generic beads. Variety is decent but the rendering is flat." }, "bowl": { "score": 0.7, "reason": "Coherent elliptical rim, smooth curved walls and a lacquer-like teal gradient with a highlight stroke. The base is not visible and the bowl seems to float above the dark background with only a faint shadow, but there is no distortion." }, "broth": { "score": 0.45, "reason": "Little broth is visible between the noodles, only brown patches at the edges. It looks like a flat surface, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.62, "reason": "Soft translucent warm wisps that change shape between frames and rise above the bowl. Some glowing blob hotspots look like light blooms rather than steam, and the plume is fairly localized." }, "composition": { "score": 0.7, "reason": "The bowl is centered and a clear focal point against a dark, moody background with warm lighting. The chopsticks add a diagonal. The bowl is large compared with the food and there is a lot of empty space, but there is no clutter." }, "vegan": { "score": 0.92, "reason": "Visible ingredients are noodles, mushrooms, tofu, nori, a leaf and scallions, with no meat, fish or egg." } } ], "diagnostics": { "vegan": 0.9066666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.6166666666666667, "bowl": 0.85, "broth": 0.6, "steam": 0.7666666666666667, "composition": 0.85 }, "reward": 0.7141666666666667, "sampleStdDev": 0.01181453906563152, "judgments": [ { "vegan": { "score": 1, "reason": "Visible toppings are nori, mushrooms, tofu, scallion rings, chili flecks and a herb leaf over noodle broth; no animal-derived ingredients appear." }, "toppings": { "score": 0.6, "reason": "Good variety: striped nori sheet, brown mushroom domes with radial gill lines, scallion rings, chili dots and a leaf, all scaled plausibly. However the tofu pieces read as pale rounded cards with ruled lines, looking somewhat note-like rather than food, and shapes are otherwise quite flat-geometric." }, "steam": { "score": 0.75, "reason": "Normal frames show the plume clearly changing shape over time: distinct wisps and glow positions differ across desktop-1/2/3 and pointer frames, rising above the bowl without obscuring food. Reduced frames are frozen as expected. The bright core reads partly as round glowing blobs rather than fine wisps." }, "noodles": { "score": 0.7, "reason": "Thin continuous yellow strands with varied bends form a coherent tangled bed filling the bowl, believable within the flat vector style. Slight deductions for fairly uniform thickness, flat placement with little sense of dipping into broth, and some repetitive squiggle patterns." }, "composition": { "score": 0.85, "reason": "Centered, well-balanced scene with the bowl as clear focal point, diagonal chopsticks adding interest, warm glow against dark atmosphere, subtle background specks and no clutter or UI. Presentation stays restrained and the dish remains the focus." }, "broth": { "score": 0.7, "reason": "Golden broth shows around the noodle bed with darker liquid at the rim, oil specks and sesame-like highlights; toppings sit on the surface with soft shadows, feeling integrated rather than pasted. Somewhat flat and uniformly bright, with limited depth in the liquid." }, "bowl": { "score": 0.85, "reason": "Coherent teal bowl: consistent rim ellipse with interior cavity, smooth hemispherical body with gradient shading, left highlight streak, plausible rim thickness, and grounded contact shadow. Chopsticks rest convincingly across the rim; only minor stiffness in the very circular geometry." } }, { "noodles": { "score": 0.65, "reason": "Dense tangle of thin yellow wavy strands reads as noodles within the flat vector style, with some shaded overlaps, but strands are uniform, flat against the surface, and borders on scribble-like crossing lines rather than showing volume or depth in the broth." }, "toppings": { "score": 0.6, "reason": "Shiitake caps with radial gills, a green herb leaf, scallion rings and chili dots are recognizable, but the beige rounded rectangles with ruled lines dominate and read as card-like blocks, and the nori is a rigid grid-patterned rectangle rather than an irregular seaweed sheet." }, "bowl": { "score": 0.85, "reason": "Coherent teal bowl: wide elliptical rim with dark interior band, smooth gradient walls, subtle curved highlight, soft ground shadow; chopsticks rest believably on the rim. Only minor simplification of the deep hemisphere form." }, "broth": { "score": 0.5, "reason": "Broth is mostly a flat dark brown/orange fill behind the noodle mat with a thin reddish ring at the edge; little sheen, depth or liquid texture, and toppings sit atop the noodle layer rather than clearly immersed in liquid." }, "steam": { "score": 0.8, "reason": "Translucent warm wisps rise from the soup and visibly change shape, spread and fade across the time-ordered frames without obscuring the food; slightly blob-like bright cores are the only shortcoming." }, "composition": { "score": 0.85, "reason": "Centered focal bowl with warm golden/teal palette against a dark atmospheric background, tasteful glow and subtle specks, dynamic chopstick diagonal; no text or clutter, and the mobile framing keeps the dish readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are nori, mushrooms, tofu-like blocks, leaf, scallion and chili; no meat, fish, egg or dairy forms are present." } }, { "noodles": { "score": 0.6, "reason": "Dense yellow strands fill the broth with some waviness and crossing, but many read as long, uniform straight strokes like straw or scratchy line-work rather than supple curved strands; coherent but flat and repetitive texture." }, "toppings": { "score": 0.65, "reason": "Readable variety: striped dark nori sheet, brown shiitake slices with radial gill lines, scallion rings, a green leaf, and pale tofu-like pieces. The beige pieces read slightly as rigid rounded cards with line details, and a few scallion rings sit pasted near the rim edge rather than in the broth." }, "bowl": { "score": 0.85, "reason": "Coherent dark teal bowl: clean elliptical rim with thick highlighted lip, dark inner wall crescent on the left consistent with the viewing angle, smooth curved body with a tasteful highlight arc, and a contact shadow grounding it; chopsticks rest believably across the rim." }, "broth": { "score": 0.6, "reason": "Warm amber liquid visible at the rim edges with glow and sheen, and toppings appear embedded, but the surface is almost entirely covered by the noodle layer, so the liquid itself has limited readable depth or texture." }, "steam": { "score": 0.75, "reason": "Normal frames at 1.0s, 2.6s, 4.2s, 6.1s and 7.9s show the translucent warm plumes changing shape, drifting and fading while never obscuring the food; reduced frames are properly frozen. The wisps are slightly blobby/cloudy rather than finely filamented." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with a dynamic diagonal chopstick line, warm amber steam glow against a restrained dark background with sparse ember particles; the bowl stays the clear focal point with no text or clutter, and the mobile crop keeps it well framed." }, "vegan": { "score": 0.75, "reason": "Visible toppings are seaweed, mushroom slices, scallion, leaf and pale beige pieces that lack any fish-cake spiral or pink marking, so they plausibly read as tofu/aburaage; appearance is somewhat ambiguous but no clearly animal-derived ingredients are shown." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9411111111111111, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.399999999999636, "frameIntervalP95Ms": 50, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "93decc8098632a447819780e1bc17c550842e9fc5b9a733e59b36037b466ffea", "gradedAt": "2026-10-08T21:05:15.086743+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.31, "toppings": 0.3277777777777778, "bowl": 0.7077777777777778, "broth": 0.3333333333333333, "steam": 0.34, "composition": 0.5244444444444445 }, "reward": 0.4222777777777778, "runId": "anthropic/haiku-5.5/low", "rewardStdDev": 0.015078805336311318, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.28, "toppings": 0.35, "bowl": 0.7766666666666667, "broth": 0.36666666666666664, "steam": 0.18666666666666668, "composition": 0.54 }, "reward": 0.41750000000000004, "sampleStdDev": 0.004358898943540678, "judgments": [ { "noodles": { "score": 0.28, "reason": "Pale strands are visible, but mostly form straight, angular crisscrossed sticks rather than supple noodles with coherent bends and volume." }, "toppings": { "score": 0.35, "reason": "Green scallion rings and a dark seaweed-like sheet are readable, but most toppings are featureless colored circles, flat blocks, and blurred orange patches with little organic texture." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, inner cavity, and rounded walls have coherent perspective and restrained teal shading. The base lacks a clear foot or convincing surface contact, giving slight floating ambiguity." }, "broth": { "score": 0.36, "reason": "An amber surface is visible around the ingredients, but it looks flat and opaque, with little liquid sheen, depth, or convincing immersion." }, "steam": { "score": 0.18, "reason": "Normal frames show the plume rising and changing shape, but it is an intensely luminous opaque white-yellow blob rather than translucent wisps; it obscures food and lacks readable fading." }, "composition": { "score": 0.52, "reason": "The isolated bowl and diagonal chopsticks create a clear layout against a restrained dark background. However, the blown-out plume competes strongly with the dish, and chopsticks extend beyond the mobile frame." }, "vegan": { "score": 0.88, "reason": "Scallion-like rings, seaweed, and pale tofu-like pieces plausibly suggest plant-based ramen, with no unmistakable animal ingredients. Several generic topping shapes remain ambiguous." } }, { "noodles": { "score": 0.28, "reason": "Pale strands are visible, but predominantly form rigid straight segments and angular crossings rather than supple noodles with coherent bends and volume." }, "toppings": { "score": 0.35, "reason": "Green scallion rings and a dark seaweed-like piece are recognizable, but most toppings are flat circles, pale blocks, and diffuse orange patches with little organic texture or readable detail." }, "bowl": { "score": 0.77, "reason": "The elliptical rim, rounded walls, and subtle teal shading form a coherent vessel with convincing depth. The heavy black inner band and weak grounding make it somewhat diagrammatic and floating." }, "broth": { "score": 0.36, "reason": "An amber surface fills the cavity, but it looks largely flat and opaque, with little liquid sheen or depth; ingredients appear layered over it rather than convincingly immersed." }, "steam": { "score": 0.18, "reason": "Normal frames show the plume changing position and shape upward, but it is an intensely luminous, opaque white-yellow mass rather than translucent wisps, and obscures food substantially in several captures." }, "composition": { "score": 0.55, "reason": "The isolated teal bowl and diagonal chopsticks create a clean focal arrangement against the dark background. Excessive white glare competes with the dish, and the chopsticks extend outside the mobile framing." }, "vegan": { "score": 0.86, "reason": "Visible scallion-like rings, seaweed-like material, pale tofu-like pieces, and yellow vegetable-like dots plausibly suggest vegan ramen. No unmistakable animal ingredient is visible, though several toppings remain ambiguous." } }, { "noodles": { "score": 0.28, "reason": "Pale strands are visible, but mostly form rigid straight segments and angular crossings rather than supple noodles with coherent bends and volume." }, "toppings": { "score": 0.35, "reason": "Green scallion-like rings and a dark seaweed-like sheet provide some recognition, but flat cream blocks, colored disks, and blurred orange patches dominate; organic texture and ingredient detail are weak." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, continuous curved walls, and restrained teal highlights give a coherent vessel with readable thickness and depth. The base and surface contact are insufficiently defined, making it appear slightly suspended." }, "broth": { "score": 0.38, "reason": "An amber surface fills the cavity, but it looks flat and opaque, with little liquid sheen or depth. Ingredient immersion and contact are poorly articulated." }, "steam": { "score": 0.2, "reason": "Normal frames show the plume changing shape and rising, but it appears as intensely luminous opaque blobs rather than translucent wisps, obscuring food near the source." }, "composition": { "score": 0.55, "reason": "The isolated teal bowl and diagonal chopsticks make a clear arrangement against a dark background without UI clutter. Excessive white-yellow glare competes with the dish, and the mobile chopsticks extend beyond the frame." }, "vegan": { "score": 0.85, "reason": "Visible scallion-like rings, seaweed-like material, and pale tofu-like pieces plausibly suggest plant-based ramen. No unmistakable animal ingredient is visible, though several toppings remain ambiguous." } } ], "diagnostics": { "vegan": 0.8633333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.36666666666666664, "toppings": 0.3, "bowl": 0.6133333333333333, "broth": 0.31666666666666665, "steam": 0.38333333333333336, "composition": 0.5 }, "reward": 0.4101666666666667, "sampleStdDev": 0.00583809329604567, "judgments": [ { "noodles": { "score": 0.35, "reason": "Noodles are thin straight crisscross sticks, like a maze of rigid scribbles. They have little volume or supple bends, and they are mostly hidden under the toppings and the glare." }, "toppings": { "score": 0.3, "reason": "Toppings are mostly white and yellow dots, green rings and blurry red blobs. A dark nori-like piece and tofu-like blocks are barely readable. The result looks like generic particles rather than organic food." }, "bowl": { "score": 0.6, "reason": "The dark bowl has a coherent rim, a smooth hemispherical body and a soft highlight, with a shadow beneath it. The rim and body perspectives are consistent. The bowl looks a little floating and has no foot, and the material is plain." }, "broth": { "score": 0.3, "reason": "The broth surface is mostly hidden under the clutter and glare. Only a brown-orange tint shows at the edges, with little sheen or depth, so the ingredients look pasted on top." }, "steam": { "score": 0.4, "reason": "The steam moves and changes across the frames, drifting upward and shifting side to side. It is an overexposed, opaque glowing blob rather than translucent wisps, and it hides the food. In the reduced-motion frames it blows out the whole bowl surface." }, "composition": { "score": 0.5, "reason": "The bowl is centered on a dark vignette with chopsticks that add a diagonal. The focal dish reads clearly. The blown-out white glare dominates the frame and distracts from the food." }, "vegan": { "score": 0.8, "reason": "The visible ingredients look plant-based: noodles, tofu-like blocks, a seaweed-like piece and corn-like dots. No meat, fish or egg is evident, though the red blobs are ambiguous." } }, { "noodles": { "score": 0.35, "reason": "Noodles are thin straight crisscrossed sticks, like scattered straw or a maze of lines. Few are supple or continuous, and they read as a flat pattern rather than a volume sitting in broth." }, "toppings": { "score": 0.3, "reason": "Toppings are mostly white and yellow dots, green rings and blurry red blobs. There is a dark nori-like square and some pale tofu blocks, but little organic detail, so they read as generic symbols." }, "bowl": { "score": 0.62, "reason": "The dark teal bowl has a coherent elliptical rim, a smooth hemispherical body with a soft highlight, and a contact shadow. It looks slightly floating and plain, and the rim and interior perspective are simple but consistent." }, "broth": { "score": 0.3, "reason": "The broth is a small brown-orange surface mostly hidden under noodle lines and blobs. It has little sheen or depth, and the ingredients look pasted on top rather than immersed." }, "steam": { "score": 0.4, "reason": "Steam moves and drifts upward across frames, so it is animated. It is a very bright, opaque white-hot glow that looks like an explosion or light blob rather than translucent wisps. It also hides the food in the reduced-motion frames." }, "composition": { "score": 0.5, "reason": "The bowl is centered in a dark vignette, with chopsticks adding a diagonal. The overexposed glow dominates and distracts, and the food detail is weak." }, "vegan": { "score": 0.8, "reason": "No visible meat, fish or egg. The ingredients look like noodles, tofu, nori and vegetable bits, so it is plausibly vegan, though the red blobs are ambiguous." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin straight criss-cross strands that look like a rigid maze or scribble. They are visible but small, with no supple bends or volume." }, "toppings": { "score": 0.3, "reason": "Toppings are mostly white dots, yellow dots, green rings, and red blurry blobs. The nori and tofu are barely readable, so these are generic symbols rather than organic food." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a smooth rounded body, a gradient highlight, and a shadow beneath. It looks a bit floaty and flat, and the base has no real contact, but the geometry is consistent." }, "broth": { "score": 0.35, "reason": "The broth is a thin brown sliver at the edges, largely covered by noodles and toppings. It has little sheen or depth and no believable immersion." }, "steam": { "score": 0.35, "reason": "Steam is animated and rises and drifts across frames. It is an overexposed, opaque white glow blob that hides the food, and in the reduced-motion frames it blows out the whole bowl." }, "composition": { "score": 0.5, "reason": "The bowl is centered with a dark moody background and chopsticks. The blown-out glow dominates and washes out the food, which hurts the focus." }, "vegan": { "score": 0.75, "reason": "No obvious meat, fish, or egg is visible. The toppings look like tofu, vegetables, and nori, though the red blobs are ambiguous." } } ], "diagnostics": { "vegan": 0.7833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.2833333333333333, "toppings": 0.3333333333333333, "bowl": 0.7333333333333334, "broth": 0.31666666666666665, "steam": 0.45, "composition": 0.5333333333333333 }, "reward": 0.4391666666666667, "sampleStdDev": 0.054848275573014464, "judgments": [ { "vegan": { "score": 0.6, "reason": "Toppings are green scallion-style rings, white puffs, dark nori-like squares and amber broth with no explicit meat, egg or fish; however the pink-orange rounded blobs are ambiguous and could read as chashu or fish-cake slices, so full compliance is not established." }, "toppings": { "score": 0.4, "reason": "Toppings are mostly simple symbols: green outlined circles, plain white dots, soft pink-orange blobs and flat dark squares. Little texture, scale variety or organic detail; many read as generic dots and disks pasted on the surface rather than recognizable vegetables or mushrooms." }, "steam": { "score": 0.55, "reason": "Frames at 1017ms through 9494ms show the plume rising, expanding and changing shape, so motion is real. But it is a very bright, near-opaque white fog that washes out the center of the bowl and hides the food, rather than translucent wisps; reduced frames show it properly suppressed." }, "noodles": { "score": 0.25, "reason": "Noodles are thin straight cream-colored sticks scattered radially across the surface, reading as rigid splinters or matches with no supple bends, coherent bundles or believable immersion in the broth." }, "composition": { "score": 0.6, "reason": "Dark atmospheric presentation with a centered bowl, chopsticks as an accent and soft ground shadow is balanced, but the huge white steam blowout dominates and obscures the dish, which should remain the focal point; mobile view also crops the chopsticks off-screen." }, "broth": { "score": 0.4, "reason": "Only narrow amber margins of broth are visible around the noodle mass; the liquid shows no sheen or depth and the center is blown out by the white glow, making the soup murky and unreadable, and toppings appear flat on top rather than integrated." }, "bowl": { "score": 0.75, "reason": "Coherent stylized bowl: elliptical rim with a light teal edge and inner lip, smooth hemispherical body with soft gradient shading, plausible chopstick contact at the rim and a grounded shadow below. Minor issues are the slightly detached-looking highlight arcs at the rim ends and thin wall treatment." } }, { "vegan": { "score": 0.75, "reason": "Visible ingredients are pale noodles, scallion-like green rings, yellow corn-like dots and a dark seaweed patch; no clear meat or egg, though blurry orange smudges are ambiguous (possibly naruto-like), so not full credit." }, "toppings": { "score": 0.3, "reason": "Toppings are dominated by scattered glossy white/yellow dots, faint green rings and blurred orange smudges that read as confetti rather than organic food; only the small dark patch vaguely suggests nori." }, "steam": { "score": 0.45, "reason": "The plume clearly animates across desktop-1/2/3, rising, spreading and changing shape, so it is not static; however it is a blown-out white-hot opaque mass that reads more like a flare than translucent wisps and hides the center of the dish." }, "noodles": { "score": 0.3, "reason": "The bowl is filled with thin, perfectly straight pale sticks crossing at random angles like scattered toothpicks; rigid, uncurved strands dominate, violating supple continuous noodle structure." }, "composition": { "score": 0.6, "reason": "Centered bowl, warm dark background, diagonal chopsticks and grounding shadow give a balanced atmospheric layout, but the enormous blown-out glow overwhelms the focal dish and washes out most food detail." }, "broth": { "score": 0.35, "reason": "The liquid surface is a flat pale plane obscured by the white-hot glow; colored smudges and dots appear pasted on top, with no believable sheen, depth or ingredient immersion." }, "bowl": { "score": 0.8, "reason": "A coherent dark teal bowl with consistent elliptical rim, visible inner wall, smooth shaded body, soft highlight and grounded shadow; only minor rim-edge line artifacts." } }, { "vegan": { "score": 0.85, "reason": "Visible contents suggest vegan ramen: noodle strands, green scallion-like rings, dark green nori-like patches, and translucent vegetable-toned blobs; no egg, meat, or fish forms are evident." }, "toppings": { "score": 0.3, "reason": "Toppings read as generic decorative elements: glossy translucent red/pink blobs, scattered white and yellow dots, and flat dark green rectangles. Only faint scallion-like rings and nori-like patches hint at real ingredients; nothing has organic vegetable, mushroom, or tofu texture." }, "steam": { "score": 0.35, "reason": "The plume does change shape and position across the time-ordered frames (low blob in frame 1, taller wispy column by frames 2-3), so it is not static. However, it is an opaque blazing white fog that glows like fire rather than translucent wisps, and it washes out the food beneath it." }, "noodles": { "score": 0.3, "reason": "Noodles are thin, perfectly straight pale-yellow sticks radiating in random directions like matchsticks or straw, with no bends, curls, or coherent layering; they read as rigid scattered sticks rather than supple strands in broth." }, "composition": { "score": 0.4, "reason": "Centered bowl with a pleasant chopstick diagonal and dark atmospheric gradient is a decent layout, but a huge blinding white-yellow glare dominates the frame and destroys the food as focal point; the dish is the least readable element on screen." }, "broth": { "score": 0.2, "reason": "No believable liquid surface is visible: the interior is dominated by an overexposed white-hot glow, and ingredients sit as flat decals on a light-colored plane rather than immersed in soup with depth or occlusion." }, "bowl": { "score": 0.65, "reason": "The bowl is a coherent dark teal hemisphere with a consistent elliptical rim, smooth ceramic sheen, a highlight, and a soft contact shadow; chopsticks rest plausibly on the rim. Minor flaws: a thin lighter band along the outer rim edges looks like a slight seam, and the stylized surface is fairly plain." } } ], "diagnostics": { "vegan": 0.7333333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.7933333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.64999999999964, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8723613c665c267d0db2513ff935605bb5874d903fa552c3c035b54cfbda5652", "gradedAt": "2026-10-08T21:16:25.376674+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.31555555555555553, "toppings": 0.38666666666666666, "bowl": 0.6244444444444445, "broth": 0.4355555555555556, "steam": 0.4477777777777778, "composition": 0.5933333333333334 }, "reward": 0.45472222222222225, "runId": "anthropic/haiku-5.5/high", "rewardStdDev": 0.06503894844487276, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.28, "toppings": 0.46, "bowl": 0.6466666666666667, "broth": 0.44, "steam": 0.42666666666666664, "composition": 0.6466666666666667 }, "reward": 0.47300000000000003, "sampleStdDev": 0.0005000000000000004, "judgments": [ { "noodles": { "score": 0.28, "reason": "Pale strands are visible, but predominantly form straight, rigid crisscrossing sticks with abrupt bends rather than supple noodles; overlap reads as a flat wire mesh." }, "toppings": { "score": 0.46, "reason": "Mushrooms, tofu, greens and seaweed are recognizable and varied, but leaves are simple oval disks, tofu has little texture, and nori resembles a thick striped card." }, "bowl": { "score": 0.64, "reason": "The teal rim and shaded inner wall establish a coherent cavity and ceramic-like sheen. The shallow front wall and barely visible base make the vessel read more like a tilted dish with weak support." }, "broth": { "score": 0.45, "reason": "Warm brown coloration and localized highlights suggest soup, but the surface is largely flat and ingredients appear layered on top with little immersion or liquid contact." }, "steam": { "score": 0.43, "reason": "Faint translucent trails above the soup change across normal frames, supporting some motion. They are mostly narrow and straight, while expanding bright haze obscures food rather than forming convincing irregular wisps." }, "composition": { "score": 0.65, "reason": "A single centered bowl, restrained dark background and teal-orange palette create a clear focal scene without clutter. Strong bloom in later frames washes out a substantial portion of the dish." }, "vegan": { "score": 0.96, "reason": "Visible tofu-like cubes, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen; no material visible animal-derived ingredient is observed." } }, { "noodles": { "score": 0.28, "reason": "Pale strands are clearly visible, but mostly form straight, rigid crisscrossing sticks with abrupt ends rather than supple noodles or coherent piled volume." }, "toppings": { "score": 0.46, "reason": "Mushrooms, tofu, greens and seaweed are recognizable and varied, but leaves are flat oval disks, tofu is minimally textured, and nori resembles a thick striped card." }, "bowl": { "score": 0.64, "reason": "The teal rim and shaded inner wall convey a coherent cavity and material, but the very shallow front wall and barely visible base make the vessel read more like a tilted dish than a deep ramen bowl." }, "broth": { "score": 0.44, "reason": "Warm amber shading suggests broth, but the surface is smooth and solid-looking, with little immersion or liquid contact around the ingredients." }, "steam": { "score": 0.43, "reason": "Faint translucent trails and changing haze appear above the soup across normal frames, but the effect becomes a bright diffuse bloom that obscures food rather than clearly articulated rising, fading wisps." }, "composition": { "score": 0.65, "reason": "A single centered dish, restrained dark background and teal\u2013amber palette create a clear focal composition without clutter; excessive white glare in later frames washes out much of the food." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed and small vegetable garnishes plausibly form vegan ramen; no visibly animal-derived topping is observed." } }, { "noodles": { "score": 0.28, "reason": "Pale strands are readable, but predominantly straight, angular crisscrosses resemble scattered sticks or wires rather than supple noodles; overlap lacks convincing volume." }, "toppings": { "score": 0.46, "reason": "Mushrooms, tofu, greens and seaweed are distinguishable, with some speckling and shading. Leaves are simple oval disks, tofu is blocky, and seaweed resembles a thick striped card." }, "bowl": { "score": 0.66, "reason": "The teal rim and shaded inner wall form a coherent cavity with readable thickness. The exterior is very shallow and the base/contact is unclear, leaving a floating saucer-like impression." }, "broth": { "score": 0.43, "reason": "Warm brown color and highlights suggest soup, but the surface is largely flat and ingredients appear layered atop it with little immersion or liquid contact." }, "steam": { "score": 0.42, "reason": "Faint translucent trails above the soup change across normal frames, supporting motion. They are narrow and beam-like, while expanding bright haze obscures substantial food detail rather than forming restrained irregular wisps." }, "composition": { "score": 0.64, "reason": "A single centered bowl, dark uncluttered background and teal\u2013amber palette establish a clear focal dish. Excessive bloom washes out the upper food in later frames, weakening readability and lighting restraint." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, corn-like kernels and seaweed plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 0.9866666666666667 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.3, "toppings": 0.3, "bowl": 0.5, "broth": 0.35, "steam": 0.4166666666666667, "composition": 0.5 }, "reward": 0.3825, "sampleStdDev": 0.004330127018922197, "judgments": [ { "noodles": { "score": 0.3, "reason": "Noodles are straight, rigid criss-crossing ribbons forming a maze-like scribble across the surface. There are no supple bends, and the strands look flat and pasted on rather than sitting in broth." }, "toppings": { "score": 0.3, "reason": "The toppings are tiny and generic: green ellipses for leaves, a flat dark rectangle for nori, small brown mushroom blobs, and red and yellow dots. Variety is present but the shapes are crude and read as symbols. They are also hard to see under the bloom." }, "bowl": { "score": 0.5, "reason": "A teal bowl with a readable rim, wall and base, and a plausible viewing angle. The inner rim and the broth ellipse do not join convincingly, and the wall edge has a visible gap on the right. The bowl looks flat and thin overall." }, "broth": { "score": 0.35, "reason": "The broth is an orange disc with a flat gradient, and the noodles and toppings look pasted on top rather than immersed. It has little depth, and heavy bloom washes out the surface." }, "steam": { "score": 0.4, "reason": "A glowing plume rises above the bowl and changes between frames. However, it is an overexposed white blob that hides the food in later frames, rather than translucent wisps." }, "composition": { "score": 0.5, "reason": "The bowl is centered and focal against a dark background with warm colors. The blown-out bloom obscures the food, and the bowl is small within the frame." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are noodles, mushrooms, greens, nori and corn-like dots. No meat, fish or egg is apparent." } }, { "noodles": { "score": 0.3, "reason": "Noodles are straight, rigid criss-crossing ribbons scattered flat across the surface like a maze of sticks. There are no supple curves or believable volume, and they look pasted on top of the broth." }, "toppings": { "score": 0.3, "reason": "Toppings are tiny and generic: flat green ovals, a dark rectangle for nori, small red dots, and yellow corn specks. Mushrooms are barely readable and the tofu is a flat tan square. They lack texture and variety." }, "bowl": { "score": 0.5, "reason": "The teal bowl has a readable rim, wall and base with a soft highlight. The rim and interior perspective look slightly off, with a visible rim break at the left and right, and the broth surface looks like a flat disc." }, "broth": { "score": 0.35, "reason": "The broth is a flat orange-brown disc with a radial gradient. It has little depth or immersion, and ingredients sit on top rather than in the liquid. Blown-out glare washes it out in later frames." }, "steam": { "score": 0.4, "reason": "A glowing steam plume rises and changes across frames. It grows into an opaque, blown-out white bloom that hides the food, and it reads more as light glare than as wisps." }, "composition": { "score": 0.5, "reason": "The bowl is centered on a dark background with a warm glow, so it is a clear focal point. The bowl is fairly small in the frame, and the overexposed bloom obscures the food in several frames." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are noodles, nori, greens, mushrooms, corn and tofu. There is no meat, egg or fish, so the bowl appears vegan." } }, { "noodles": { "score": 0.3, "reason": "Noodles are straight rigid ribbons crisscrossing in a maze-like scribble over the surface. They look like flat sticks with no supple bends or volume." }, "toppings": { "score": 0.3, "reason": "Toppings are tiny and generic: green discs for leaves, a dark rectangle for nori, small mushroom blobs, and red and yellow dots. They lack organic detail, and the blown-out glare hides them in later frames." }, "bowl": { "score": 0.5, "reason": "The teal bowl has a readable rim, wall and base with a plausible oblique angle. The inner rim looks slightly mismatched and the wall has a odd seam on the right, so the construction is flat and somewhat artificial." }, "broth": { "score": 0.35, "reason": "The broth reads as a flat orange disc with little depth or immersion. The noodles lie on top like a pasted layer, and the glare washes out the surface." }, "steam": { "score": 0.45, "reason": "A glowing plume rises and changes across frames, so it is clearly animated. It becomes a blown-out white bloom that hides the food and reads more as a light flare than as translucent wisps." }, "composition": { "score": 0.5, "reason": "The bowl is centered on a dark background and is the clear focal point. Its small size and the overexposed glare in later frames weaken the presentation." }, "vegan": { "score": 0.85, "reason": "Visible items are noodles, mushrooms, greens, nori and corn, with no obvious animal products." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.36666666666666664, "toppings": 0.4, "bowl": 0.7266666666666667, "broth": 0.5166666666666667, "steam": 0.5, "composition": 0.6333333333333333 }, "reward": 0.5086666666666667, "sampleStdDev": 0.017221594970656336, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are uniform straight cream rods scattered like pick-up-sticks across the surface\u2014no bends, curves, clumping, or volume; rigid crossed segments dominate, which the rubric caps at .50." }, "toppings": { "score": 0.5, "reason": "Some variety: dark green nori-like sheet, yellow corn clusters, brown mushroom slices, tan tofu squares, red balls, green disks\u2014but many read as flat generic disks and small spheres, and white lens-flare sparkles glare over several toppings." }, "bowl": { "score": 0.78, "reason": "Coherent teal top-down bowl: elliptical rim, inner wall shading, right-side thickness line, plausible highlight; perspective is consistent, though depth is shallow." }, "broth": { "score": 0.4, "reason": "Broth is a flat brown-orange gradient ellipse with speckles\u2014solid-looking, no liquid sheen or texture; toppings and noodles appear pasted on top with no immersion or occlusion, and a central glare washes it out." }, "steam": { "score": 0.45, "reason": "Steam is animated\u2014faint wisps in desktop-1 grow into a large plume by desktop-3 and pointer frames\u2014so it is not static, but it reads as an opaque bright light bloom rather than translucent irregular wisps and hides the food." }, "composition": { "score": 0.6, "reason": "Dark starfield backdrop with warm palette and a centered focal dish is tasteful, but the bowl is small with much empty space, and the intensifying central glow distracts from and obscures the food." }, "vegan": { "score": 0.95, "reason": "Visible ingredients\u2014noodles, greens, nori-like seaweed, corn-like clusters, mushrooms, tofu-like squares, red peppers in brown broth\u2014plausibly form a vegan ramen with no animal-derived toppings apparent." } }, { "noodles": { "score": 0.35, "reason": "Strands are uniform straight beige rods laid in a crisscross pick-up-sticks tangle; rigid, sharp-bent segments with no supple waves, no sense of submersion or volume \u2014 rigid sticks dominate, capped at .50 by the rubric." }, "toppings": { "score": 0.3, "reason": "Toppings are generic plastic-like forms: flat green disks with a slit, a rigid dark-green gridded card standing in for nori, glossy red spheres, and pale square blocks; only the brown mushroom lumps and yellow corn clusters read as food. Generic disk/card/sphere forms dominate, capped at .50." }, "bowl": { "score": 0.7, "reason": "Teal elliptical bowl with coherent rim, visible inner wall, gloss highlight and grounded base; stylized top-down perspective is internally consistent, though a faint detached-looking pale arc sits near the inner rim and wall depth is shallow." }, "broth": { "score": 0.6, "reason": "Glossy orange-brown liquid with gradient depth and ingredients in contact, but heavy white bloom hotspots and glitter-like specks wash out the center in later frames, and noodles appear pasted on the surface rather than immersed." }, "steam": { "score": 0.55, "reason": "Wisps evolve across normal frames (faint threads at ~1s, fuller plumes by ~5s), showing genuine motion, but the plumes read as bright opaque fog blobs with excessive bloom rather than translucent irregular wisps." }, "composition": { "score": 0.7, "reason": "Centered dish as clear focal point in a dark star-dotted atmosphere with pleasing teal/warm contrast and no clutter or UI; the intense central glow periodically distracts from the food and the bowl feels small on mobile." }, "vegan": { "score": 0.75, "reason": "Visible toppings (greens, corn, tofu, mushrooms, nori) plausibly form vegan ramen; the small glossy red spheres are ambiguous (could read as roe), preventing full credit." } }, { "noodles": { "score": 0.35, "reason": "Noodles render as completely straight, rigid rods criss-crossing like pick-up-sticks across the broth; no curvature, supple bends, or clumped nest volume \u2014 exactly the rigid maze-like strand failure the rubric caps at .50, and here it dominates the dish." }, "toppings": { "score": 0.4, "reason": "Some variety (tofu squares, brown mushroom caps, corn clusters, scallion rings, green discs, red spheres) but forms are generic: flat green disks, glossy red plastic-like spheres, and a thick rigid dark-green grid card standing in for nori; simplified sticker-like shapes dominate." }, "bowl": { "score": 0.7, "reason": "Teal ceramic bowl is coherent: consistent elliptical rim, visible wall thickness, smooth gradient body, rim highlight, and matching inner-cavity tilt; no broken joins or perspective clash, though it is shallow with no foot ring and floats without support context." }, "broth": { "score": 0.55, "reason": "Warm amber-brown miso-style surface reads plausibly around the strands with a soft gradient, but it is fairly flat, and by the pointer frames a white-hot glare blows out the upper broth, obscuring the surface and ingredient contact." }, "steam": { "score": 0.5, "reason": "Steam is genuinely animated across normal frames \u2014 thin wisps in desktop-1 grow into shifting diffuse plumes by desktop-3 and the pointer frames, and reduced-motion frames are static as expected. However, later plumes become a bright opaque fog/bloom that hides the food rather than translucent wisps." }, "composition": { "score": 0.6, "reason": "Centered bowl on a moody near-black starfield with warm appetizing palette; the early frame is well balanced, but the growing central glare overexposes and distracts from the dish in subsequent frames, and the bowl sits unsupported in empty space." }, "vegan": { "score": 1.0, "reason": "All visible toppings are plant-plausible: tofu, mushrooms, corn, scallion rings, cucumber-like green slices, nori-like sheet, and red chili/tomato spheres; no egg, meat, fish, or dairy forms appear." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9122222222222223, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.29999999999927, "frameIntervalP95Ms": 99.89999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "ac86dbc8981e96e4f1212be86f28b0d16abe0d77c2d2c22044b8c31f9ac2ec37", "gradedAt": "2026-10-08T21:27:43.786516+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5333333333333333, "toppings": 0.41444444444444445, "bowl": 0.7644444444444445, "broth": 0.6411111111111111, "steam": 0.5788888888888889, "composition": 0.7844444444444444 }, "reward": 0.5925555555555556, "runId": "anthropic/haiku-5.5/xhigh", "rewardStdDev": 0.08109362444684494, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.42666666666666664, "bowl": 0.8266666666666667, "broth": 0.6466666666666667, "steam": 0.6033333333333333, "composition": 0.8433333333333333 }, "reward": 0.6114999999999999, "sampleStdDev": 0.0040000000000000036, "judgments": [ { "noodles": { "score": 0.5, "reason": "Distinct cream strands overlap coherently, but angular, wire-like bends and a shallow scattered arrangement undermine supple noodle volume." }, "toppings": { "score": 0.42, "reason": "Green garnish, pale tofu-like pieces and brown mushroom-like slices provide variety, but mostly read as flat geometric tokens; the seaweed is a rigid rectangle with little organic detail." }, "bowl": { "score": 0.83, "reason": "A consistent elliptical rim, visible inner wall, rounded dark body and grounded foot establish convincing construction. Restrained edge highlights suggest ceramic, though the exterior shading is somewhat flat." }, "broth": { "score": 0.64, "reason": "Warm amber shading, small surface glints and dim submerged strands suggest liquid depth. The surface lacks convincing contact ripples and several ingredients appear laid over it." }, "steam": { "score": 0.61, "reason": "Translucent plumes emerge from the soup and subtly change shape and intensity across normal frames without obscuring food. Their broad, nearly parallel columns lack intricate irregular wisps and clearly readable rising motion." }, "composition": { "score": 0.85, "reason": "The single dish remains a clear focal point against a restrained dark backdrop, with warm food colors contrasting the black bowl and no distracting UI or text. Mobile framing retains the complete bowl; excessive grain is a minor distraction." }, "vegan": { "score": 1, "reason": "Visible noodles, green garnish, seaweed and tofu- or mushroom-like pieces plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Distinct pale strands overlap and include submerged sections, but angular, wire-like bends and a mostly flat scattered tangle undermine supple noodle volume." }, "toppings": { "score": 0.43, "reason": "Green scallion rings, pale tofu-like pieces and dark seaweed are discernible, but toppings are mostly flat geometric tokens with little organic texture; the seaweed is a rigid rectangle." }, "bowl": { "score": 0.83, "reason": "Coherent elliptical rim, visible inner wall, rounded dark body and grounded foot establish convincing depth and support. Restrained rim highlights work well, though the exterior material shading is somewhat uniform." }, "broth": { "score": 0.65, "reason": "Warm amber shading, small surface flecks and dim submerged strands suggest liquid depth. The broad surface remains rather matte, with limited contact effects around ingredients." }, "steam": { "score": 0.62, "reason": "Translucent columns emerge above the soup without obscuring it and subtly change shape and density across normal frames. Their evenly spaced, diffuse vertical forms lack distinct irregular curling wisps." }, "composition": { "score": 0.84, "reason": "A single clearly framed bowl, warm food illumination and restrained dark surroundings create a strong focal dish without text or clutter. Mobile framing preserves the bowl, though generous upper empty space slightly weakens balance." }, "vegan": { "score": 1, "reason": "Visible noodles, green vegetables, seaweed and pale tofu-like pieces plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Distinct pale strands overlap with some shading and submerged portions, but angular, wire-like bends and a flattened scattered arrangement dominate rather than supple noodle volume." }, "toppings": { "score": 0.43, "reason": "Pale tofu-like pieces, green rings, small brown mushroom-like shapes and a dark seaweed strip provide variety, but mostly read as flat geometric symbols with little organic texture or detail." }, "bowl": { "score": 0.82, "reason": "A coherent elliptical rim, visible inner wall, rounded dark body and grounded foot establish convincing depth and support. Restrained rim highlights work well, though the body material is subdued and somewhat featureless." }, "broth": { "score": 0.65, "reason": "Warm amber shading, small surface flecks and dimmer submerged strands suggest liquid depth. The broad surface remains rather matte, and ingredient contact lacks convincing ripples or menisci." }, "steam": { "score": 0.58, "reason": "Translucent plumes emerge above the soup without obscuring it, and their contours and brightness shift across normal frames. They remain broad, nearly parallel columns with limited irregular curling or visibly spreading wisps." }, "composition": { "score": 0.84, "reason": "The isolated bowl remains a clear focal point against a restrained dark background, with warm amber light contrasting the charcoal vessel and no distracting UI or text. Mobile framing preserves the dish, although generous upper negative space makes it feel low-set." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly read as tofu, greens, mushrooms and seaweed alongside noodles; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.3, "bowl": 0.65, "broth": 0.5933333333333333, "steam": 0.43333333333333335, "composition": 0.6766666666666666 }, "reward": 0.5036666666666667, "sampleStdDev": 0.020744477176668833, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin cream strands with wavy bends cluster in a central tangle and read as noodles. They are thin and wiry, with flat uniform strokes and a radiating spiky pattern. They look more like scribbled lines than supple noodles, and they sit on the surface rather than in the broth." }, "toppings": { "score": 0.3, "reason": "The toppings are tiny and sparse: a dark green rectangle for nori, pale tofu-like squares, a few green dots and one green leaf. They are generic flat shapes with little texture, and there are no mushrooms or recognizable vegetables of real scale." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a dark rounded body, a base shadow and a warm rim light. The perspective is consistent. The material is fairly flat and matte, with weak ceramic highlights and a dull wall." }, "broth": { "score": 0.58, "reason": "The warm amber-brown broth has a radial glow and a darker edge that give some depth, and it reads as soup. It is fairly flat and smooth, with little sheen or texture, and the ingredients look laid on top rather than immersed." }, "steam": { "score": 0.4, "reason": "Faint vertical soft columns rise above the bowl and shift slightly between frames. They look like uniform blurry streaks rather than irregular wisps, and they are subtle and quite mechanical." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, moody scene with a warm glow, and the food is the clear focus. There is no clutter or text. The palette is cohesive, though the bowl is dark and the toppings are small." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are noodles, tofu-like cubes, nori, greens and a plant-looking broth. There is no meat, egg or fish." } }, { "noodles": { "score": 0.55, "reason": "Noodles are thin, wavy and overlap in a dense central tangle, but they look like flat thin ribbons radiating out from a hub. They read as scribbly and uniform, and sit on the surface rather than in the broth." }, "toppings": { "score": 0.3, "reason": "Toppings are tiny and generic: beige tofu squares, a dark teal rectangle for nori, a green blob and small dots of scallion. They are barely readable, with no mushrooms or leafy vegetables, and read as flat blocks." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a deep rounded body, a foot and a grounded shadow, with consistent perspective. Its dark matte surface is a bit flat and low in detail, with only a subtle rim highlight." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a radial glow and a darker edge that give depth and an appetizing color. The surface is somewhat flat and smooth with little sheen, and the noodles look pasted on top." }, "steam": { "score": 0.5, "reason": "Soft vertical translucent plumes rise above the bowl and shift subtly between frames. They are somewhat uniform, columnar and faint, and mostly static-looking, but they do not hide the food." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point in a dark atmospheric scene with a warm glow, and there is no clutter or text. The food is small within the bowl, and the lower half is a large dark mass." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are noodles, tofu-like cubes, nori and scallions, with no meat, egg or fish visible. The broth appearance is ambiguous, but the bowl plausibly reads as vegan." } }, { "noodles": { "score": 0.55, "reason": "Thin, wavy cream strands overlap into a coherent central tangle with some organic bends. They read as thin wires radiating from a center, a bit flat and uniform, with a few loose stray ends. They sit on the broth rather than being immersed." }, "toppings": { "score": 0.3, "reason": "The toppings are sparse and tiny: a dark green rectangle for nori, pale square tofu cubes, a small green leaf and scattered dots. They are generic blocks with little texture, there are no clear mushrooms or vegetables, and little variety or detail." }, "bowl": { "score": 0.65, "reason": "The bowl has a coherent elliptical rim, a curved body, a base and a contact shadow, and the perspective is consistent. The dark matte ceramic is rather plain, with minimal highlights, and the body is murky against the dark background." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a glowing center and a darker vignette edge, which looks appetizing. The surface is fairly flat with little ripple or sheen. A few floating specks add some texture, but the noodles look slightly pasted on top." }, "steam": { "score": 0.4, "reason": "Soft vertical translucent plumes rise above the bowl and shift slightly between frames. They look like uniform vertical streaks that are blurry and mechanical, and they appear to start above the rim rather than from the soup surface." }, "composition": { "score": 0.65, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with warm lighting and no clutter or text. The food is small within the bowl, and the lower half of the bowl is dull." }, "vegan": { "score": 0.85, "reason": "The bowl shows only noodles, tofu-like cubes, nori, greens and broth. There are no visible meat, egg or fish items." } } ], "diagnostics": { "vegan": 0.85 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5666666666666667, "toppings": 0.5166666666666667, "bowl": 0.8166666666666667, "broth": 0.6833333333333333, "steam": 0.7, "composition": 0.8333333333333334 }, "reward": 0.6625000000000001, "sampleStdDev": 0.028831406486677005, "judgments": [ { "noodles": { "score": 0.6, "reason": "Pale curvy strands with tapered ends, highlight lines and overlap read as noodles, but they lie as scattered separate squiggles splayed flat over the broth rather than a coherent nested mass, with visible gaps and slightly worm-like repetition." }, "toppings": { "score": 0.55, "reason": "Scallion rings and red chili flecks are readable, but tofu appears as plain featureless pale blocks, the nori is a rigid glossy rectangle card-like in shape, and overall variety is sparse; an ambiguous brown blob sits at right." }, "bowl": { "score": 0.85, "reason": "Coherent dark ceramic bowl: clean elliptical rim with inner wall band, smooth tapering body, believable foot with contact shadow, subtle sheen; consistent perspective across all frames with no broken joins." }, "broth": { "score": 0.7, "reason": "Warm orange-brown surface with restrained sheen, oil/chili specks, bubbles and soft shadows under the noodles suggesting immersion; slightly flat and glossy in places, and a few tofu squares sit with minimal occlusion." }, "steam": { "score": 0.75, "reason": "Translucent irregular wisps rise from the soup and fade without hiding food; comparing desktop-1/2/3 and pointer frames the plume shapes visibly shift and spread, confirming gentle motion rather than static smoke." }, "composition": { "score": 0.85, "reason": "Balanced centered dish in a dark atmospheric setting with warm coherent lighting, soft vignette, bowl and food clearly focal, no clutter, text or competing elements; mobile framing also keeps the dish readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, scallions, tofu blocks, nori and chili flecks in plant broth; no meat, fish, egg or dairy-like toppings appear." } }, { "noodles": { "score": 0.5, "reason": "Thin pale strands sprawl over the surface as tangled, root-like squiggles with slight shadowing, but they read more as wire scribbles scattered flat on top than a coherent noodle nest resting in broth." }, "toppings": { "score": 0.5, "reason": "Scattered scallion rings are recognizable, but the tofu pieces are featureless pale parallelograms and the nori is a flat dark card; variety is sparse and several shapes read as generic blocks rather than food." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with a clean elliptical rim, inner-wall highlight band, smoothly tapering body, foot shadow and believable ground contact; the matte material reads well with only minor lack of surface detail." }, "broth": { "score": 0.65, "reason": "Warm tonkotsu-toned liquid with a central glow, oil specks, bubbles and sesame dots, but the surface is somewhat flat and glossy, and the noodles/toppings appear pasted onto it rather than believably immersed." }, "steam": { "score": 0.7, "reason": "Translucent wisps rise from the soup and change shape across the time-ordered frames (distinct silhouettes at 1.0s, 3.0s, 4.9s and pointer frames), fading without hiding food; the plumes are slightly columnar/mechanical but clearly not static." }, "composition": { "score": 0.85, "reason": "Balanced centered bowl on a dark atmospheric ground with warm rim glow, readable focal dish, no clutter or UI; the dimness subdues ingredient readability slightly but presentation is clean on desktop and mobile." }, "vegan": { "score": 1, "reason": "Visible ingredients are noodles, scallion rings, tofu-like cubes, a nori sheet, chili flakes and plant broth; no animal-derived toppings appear." } }, { "noodles": { "score": 0.6, "reason": "Individually the cream-colored strands are continuous with tapered ends and soft under-shadows, but they sprawl as a flat squiggle field across the whole surface rather than piling with volume or dipping believably into the broth, reading somewhat scribble-like." }, "toppings": { "score": 0.5, "reason": "Sparse, small toppings: scallion rings read acceptably, but the beige rounded blocks are featureless pill/card forms, the dark rectangles read as rigid flat cards, and remaining brown bits are vague ovals; generic flat tokens dominate with little variety or texture." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: clean elliptical rim with thin inner wall highlight, smooth tapering body, footed base, ground shadow, and consistent perspective; only shortcoming is the very dark body shows minimal material detail." }, "broth": { "score": 0.7, "reason": "Amber-brown surface with tonal swirls, small bubbles and specks gives a plausible liquid read, but the bright central glow is somewhat light-effect-like and toppings/noodles sit on top with weak immersion and occlusion." }, "steam": { "score": 0.65, "reason": "Translucent pale wisps rise from the soup and change shape/position between desktop-1/2/3 and pointer frames, fading with height without hiding food; however they are fairly straight vertical columns rather than irregular billowing wisps." }, "composition": { "score": 0.8, "reason": "Centered, well-proportioned focal dish in a warm dark vignette with subtle ambient specks; lighting supports the dish, no text/UI/clutter, though the scene is quite simple and the background glow does some of the atmospheric work." }, "vegan": { "score": 0.95, "reason": "Visible ingredients are noodles, scallion rings, dark seaweed sheets, tofu-like beige pieces and plant-broth specks; no meat, egg, fish or dairy forms appear anywhere." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9444444444444444, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.60000000000036, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "9f63a85cc15bdc7b1f565f169dc4a9fac01d796c4df9ba527b33c30cade763ed", "gradedAt": "2026-10-08T21:41:59.814307+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5033333333333334, "toppings": 0.47333333333333333, "bowl": 0.7133333333333334, "broth": 0.6233333333333333, "steam": 0.6722222222222223, "composition": 0.7488888888888889 }, "reward": 0.5997222222222223, "runId": "anthropic/haiku-5.5/max", "rewardStdDev": 0.05200275989826925, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.46, "toppings": 0.48, "bowl": 0.76, "broth": 0.6366666666666667, "steam": 0.6666666666666667, "composition": 0.7966666666666667 }, "reward": 0.6073333333333334, "sampleStdDev": 0.0012583057392117633, "judgments": [ { "noodles": { "score": 0.46, "reason": "Continuous pale strands are readable, but uniformly crimped, wire-like loops dominate and sit in a flat spiral rather than a supple, layered noodle mound." }, "toppings": { "score": 0.48, "reason": "Scallion rings, corn, pale tofu pieces and green vegetables provide variety, but most are simple flat polygons or glossy ovals with little organic texture; mushrooms are poorly resolved." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, shaded inner wall and dark exterior form a coherent supported vessel. The abruptly narrowed lower silhouette and sparse material highlights make it somewhat diagrammatic." }, "broth": { "score": 0.64, "reason": "Warm brown gradients, small oil-like highlights and a darker perimeter suggest liquid depth. Ingredients mostly appear laid over the surface, with limited convincing immersion or contact." }, "steam": { "score": 0.66, "reason": "Normal frames show translucent vapor developing upward from the soup without obscuring the food. The vapor consists largely of soft rounded patches in similar columns rather than finely irregular wisps." }, "composition": { "score": 0.79, "reason": "A restrained amber-and-black palette, clear focal bowl, diagonal chopsticks and soft contact shadow create an attractive uncluttered scene. Desktop framing leaves substantial empty space above the low-positioned dish." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly read as tofu, greens, scallions, corn and mushrooms; no discernible animal-derived ingredient is present." } }, { "noodles": { "score": 0.46, "reason": "Thin highlighted strands are readable and overlap, but uniformly crimped, angular loops form a flat wire-like spiral rather than supple noodles with volume." }, "toppings": { "score": 0.48, "reason": "Scallion rings, corn, pale tofu pieces and dark mushroom-like shapes offer variety, but most are simple flat polygons or glossy ovals; greens lack organic edges and texture." }, "bowl": { "score": 0.76, "reason": "Elliptical rim, shaded inner wall and connected rounded body establish a coherent cavity and support. The broad rim and abruptly narrowed lower body feel somewhat schematic, with limited ceramic highlights." }, "broth": { "score": 0.64, "reason": "Warm brown gradients, small oil-like highlights and perimeter shading suggest liquid depth. Ingredient contact and immersion are weak, leaving much of the food looking layered on top." }, "steam": { "score": 0.67, "reason": "Normal frames show translucent steam developing upward from the bowl and spreading without obscuring food. The plumes consist largely of soft rounded puffs rather than finely irregular wisps; continuous fading is not clearly demonstrated." }, "composition": { "score": 0.8, "reason": "A restrained amber-and-black palette, clear focal bowl and diagonal chopsticks make a cohesive atmospheric presentation without text or clutter. The dish sits rather low, leaving substantial empty space above." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, scallion rings, corn and tofu- and mushroom-like toppings plausibly form vegan ramen; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.46, "reason": "Continuous pale strands are clearly readable, but repetitive tight zigzags and broad flat spiral loops resemble crimped wire rather than supple noodles with convincing volume." }, "toppings": { "score": 0.48, "reason": "Scallion rings, corn, pale tofu pieces and dark mushroom-like shapes offer variety, but most are simple flat polygons or smooth blobs; greens lack organic edges and readable leaf texture." }, "bowl": { "score": 0.76, "reason": "The elliptical rim, shaded inner wall and connected rounded body establish a coherent vessel with a contact shadow. The abrupt inward shoulder and subdued material highlights make the construction somewhat schematic." }, "broth": { "score": 0.63, "reason": "Warm brown gradients and small oily highlights suggest an appetizing liquid surface. Ingredient immersion is weak, with much of the food appearing layered atop the surface rather than partially submerged." }, "steam": { "score": 0.67, "reason": "Normal frames show translucent steam developing above the soup and extending higher without obscuring the dish. The wisps are mostly soft rounded patches arranged in vertical columns, with limited fine curling structure." }, "composition": { "score": 0.8, "reason": "A single readable bowl and diagonal chopsticks remain the focus against a restrained dark amber background, with no distracting text or UI. The dish sits rather low, leaving substantial empty space above, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, greens, scallions, corn and mushroom-like toppings plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.4066666666666667, "bowl": 0.63, "broth": 0.6, "steam": 0.5666666666666667, "composition": 0.65 }, "reward": 0.5443333333333333, "sampleStdDev": 0.005057996968497859, "judgments": [ { "noodles": { "score": 0.55, "reason": "Wavy cream strands with a ruffled texture read as ramen noodles and overlap with some volume. They look thin and wiry, like scribbled loops lying on the surface rather than sitting in the broth, and a few are loose with no clear mass." }, "toppings": { "score": 0.4, "reason": "Toppings are tiny and sparse: a white tofu cube, corn kernels, scallion dots, a dark green leaf and mushroom-like dark ovals. They are hard to read and mostly small generic shapes, and the nori and mushrooms are not convincing." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, an inner wall, a pedestal foot and a soft shadow, with chopsticks resting on the rim. The body-to-foot join is slightly awkward, and the bowl looks a little floaty." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a radial glow, specks of oil and a believable color. It is somewhat flat and gradient-like, and the noodles seem to sit on top rather than being immersed." }, "steam": { "score": 0.55, "reason": "Soft translucent plumes grow and rise over the frames. They are blurry, blobby column shapes that look somewhat uniform, but they do not hide the food." }, "composition": { "score": 0.62, "reason": "The dark, moody scene has a centered bowl and warm lighting, and the food is the focus. The bowl sits low with a large empty area above, and the toppings are small." }, "vegan": { "score": 0.85, "reason": "Visible ingredients are tofu, corn, scallions, mushrooms, greens and noodles. No meat or egg is visible." } }, { "noodles": { "score": 0.55, "reason": "Wavy cream strands with a ruffled texture read as ramen noodles and overlap with some volume. They look like thin wire scribbles laid on top of the broth, and several loops repeat. A few strands end abruptly and the mass is somewhat maze-like." }, "toppings": { "score": 0.4, "reason": "The toppings are small and hard to read. A white tofu cube, a few corn kernels, dark mushroom ovals, green scallion rings and a green leaf are present, but they are tiny, flat and partly hidden by the noodles. Nothing has convincing texture or detail, and the nori is a dark flat strip." }, "bowl": { "score": 0.65, "reason": "The dark bowl has a coherent elliptical rim, an inner wall, a foot and a ground shadow. The rim highlight is subtle and the perspective is consistent. The body looks slightly floaty and the foot is narrow, and the material is plain, but no clear distortion is visible." }, "broth": { "score": 0.6, "reason": "The warm amber broth has a radial glow, small oil droplets and some depth. It looks somewhat flat and gradient-like. The noodles sit on top of it rather than being immersed, with little occlusion at the contact points." }, "steam": { "score": 0.55, "reason": "Soft translucent wisps grow and rise across the frames, from none in frame 1 to a taller plume later. They are blurry, roughly blob-like columns and look somewhat uniform, but they are subtle and do not hide the food." }, "composition": { "score": 0.65, "reason": "The bowl is a centered focal point on a dark, atmospheric, warm-toned background with chopsticks as an accent. The bowl sits low in the frame with a lot of empty space above it. There is no clutter or text, and the lighting is coherent." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are tofu, mushrooms, corn, scallions, greens and noodles. No meat, egg or fish is visible." } }, { "noodles": { "score": 0.5, "reason": "Noodles are wavy, thin, golden strands with decent overlap, but they look like scribbled wire loops floating on the surface. They are thin and repetitive, with little volume or immersion in the broth." }, "toppings": { "score": 0.42, "reason": "Toppings include a tofu cube, corn, scallion rings, dark mushroom circles, and a green leaf and nori strip. They are tiny and generic. Most are flat shapes or dots with little texture, and none are clearly readable." }, "bowl": { "score": 0.62, "reason": "The bowl has a coherent elliptical rim, a dark glazed wall, a base with a foot, and a soft shadow. The wall flares oddly into the foot, and the rim and body look slightly disjointed. The bowl seems to float with weak contact." }, "broth": { "score": 0.6, "reason": "The broth is a warm amber with a radial glow and small oil droplets, so it reads as appetizing liquid. It is fairly flat, and the noodles sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps build up over the frames and rise above the bowl without hiding the food. They are blobby and a bit uniform, but they do move and fade." }, "composition": { "score": 0.68, "reason": "The bowl is centered as the focal point on a dark, atmospheric background with warm lighting and no distracting text. The bowl sits low in the frame, with a large empty area above it. Chopsticks add a nice accent." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are tofu, mushrooms, corn, scallions, greens, and noodles. No meat or egg is visible." } } ], "diagnostics": { "vegan": 0.8833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.5166666666666667, "toppings": 0.5333333333333333, "bowl": 0.75, "broth": 0.6333333333333333, "steam": 0.7833333333333333, "composition": 0.8 }, "reward": 0.6475000000000001, "sampleStdDev": 0.039686269665968894, "judgments": [ { "noodles": { "score": 0.55, "reason": "Continuous cream-colored strands with plausible thickness and coherent tangling, but drawn as uniform repeating sine-wave squiggles that read as flat coiled ribbons lying on the surface, with no volume or immersion shading; a few strands drape over the front rim." }, "toppings": { "score": 0.55, "reason": "Varied elements are readable: green leaf pieces, scallion rings, corn kernels, a white tofu square, dark seaweed-like strips, chili flecks. However most are simple flat squares, circles and dots with little organic texture, scattered like confetti on the surface." }, "bowl": { "score": 0.75, "reason": "Coherent black bowl: smooth elliptical rim, visible interior wall, gently curving body and attached pedestal foot, with chopsticks resting believably on the rim with a shadow. Slightly generic and thin-walled, but no broken joins or incompatible perspective." }, "broth": { "score": 0.65, "reason": "Warm brown liquid surface with a soft glowing sheen, floating oil droplets and small bubbles; ingredients sit on it acceptably, though noodles and toppings look more pasted-on than submerged, and the broth reads a bit like a flat gradient disc." }, "steam": { "score": 0.8, "reason": "Time-ordered frames show steam absent at 1s, forming wisps by 2.8s, and taller diffuse plumes by 4.5s and in pointer frames \u2014 translucent, irregular, rising and fading without obscuring the food. Slightly blobby particle clusters, but clearly dynamic." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric scene: the bowl is the clear focal point slightly left of center with chopsticks counterweighting right, warm brown glow and dust particles, no clutter or UI; upper frame is somewhat empty but stylistically restrained." }, "vegan": { "score": 1.0, "reason": "Visible toppings are vegetables, greens, corn, tofu square, mushrooms and seaweed-like strips in a plant-broth-colored soup; no egg, meat, fish or dairy visible." } }, { "noodles": { "score": 0.55, "reason": "Continuous pale strands form a nest across the broth with plausible thickness, but each strand is a uniform repeating sine-wave squiggle with a beaded texture, reading as mechanical wiggle-lines lying flat on the surface rather than supple noodles with volume." }, "toppings": { "score": 0.55, "reason": "Recognizable variety is present \u2014 green leaf/scallion shapes, a white tofu square, dark mushroom-like ellipses, a yellow corn cluster, scallion rings and chili specks \u2014 but several pieces are small featureless tan rectangles and plain dark blobs, so the forms read flat and schematic rather than organic." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: clean elliptical rim with a highlight arc, visible inner wall gradient, consistent perspective on the broth ellipse, stable pedestal foot with contact shadow, and chopsticks resting believably on the rim. Only minor shortcoming is the rather generic stem-bowl profile." }, "broth": { "score": 0.7, "reason": "Warm brown-orange liquid surface with a radial glow, restrained glints, oil droplets at the edges and a believable rim line where broth meets the wall. Ingredients sit on the surface with light contact shading, though the strong central bloom and flat-lying toppings slightly reduce the sense of immersion." }, "steam": { "score": 0.8, "reason": "Steam is absent in the first frame, fades in by ~2.8s, then grows taller and more diffuse across subsequent frames with changing irregular wisp shapes, rising, spreading and fading without obscuring the food; translucent and animated, only somewhat blobby/columnar in structure." }, "composition": { "score": 0.8, "reason": "Balanced dark atmospheric scene: the bowl is the clear focal point, slightly offset with chopsticks counterweighting to the right, warm coherent lighting, subtle dust particles, no clutter or text. The large empty dark area above the bowl in the earliest frame is only partly a defect since steam later fills it; mobile framing also stays readable." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are noodles, greens, tofu, mushroom-like pieces, corn and chili specks in plant broth; no meat, fish, egg or dairy appears." } }, { "noodles": { "score": 0.45, "reason": "Noodles are uniform wavy squiggle strands lying flat on the broth surface like decorative wire scribbles; some overlap exists but there is no believable volume, nesting, or immersion in the soup." }, "toppings": { "score": 0.5, "reason": "Recognizable but very flat cutout shapes: scallion rings, green leaf blobs, a white tofu-like square, beige rectangles, dark mushroom dots and yellow corn clusters; several read as featureless blocks and small plastic-like balls with minimal texture." }, "bowl": { "score": 0.7, "reason": "Coherent wide dark bowl with elliptical rim, cavity, flared body and foot, consistent high-angle perspective, and chopsticks resting plausibly on the rim; it is quite plain with little ceramic material variation and a heavy pedestal-like base." }, "broth": { "score": 0.55, "reason": "Orange-brown surface with soft gradient, oil specks and scattered bits reads as liquid, but it is essentially a flat disc; toppings and noodles rest on top of the surface rather than appearing submerged, limiting depth and integration." }, "steam": { "score": 0.75, "reason": "Normal frames show temporal evolution: no steam at ~1s, wisps appearing at ~2.8s, and a taller translucent irregular column by ~6.9-9.5s, so animated rising is supported; wisps stay translucent and do not obscure the food, though they become somewhat blobby." }, "composition": { "score": 0.8, "reason": "Balanced centered dish in a dark atmospheric scene with warm spotlight glow, chopsticks accent, no text or clutter, and clean mobile framing; presentation is attractive though somewhat sparse." }, "vegan": { "score": 0.9, "reason": "Visible ingredients (noodles, scallions, greens, mushrooms, corn, tofu-like white square, chili flecks) plausibly form vegan ramen; the white square with a tiny red fleck is slightly ambiguous but reads most like tofu, not an animal product." } } ], "diagnostics": { "vegan": 0.9666666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.95, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "8c393902f229acffad11846b144a964fdb6755112623e140f6a477273d18b305", "gradedAt": "2026-10-12T20:01:37.793888+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6144444444444445, "toppings": 0.6311111111111111, "bowl": 0.8077777777777778, "broth": 0.6288888888888888, "steam": 0.6422222222222222, "composition": 0.83 }, "reward": 0.6844444444444444, "runId": "meta/muse-wonder-1.0/xhigh", "rewardStdDev": 0.0680435590333569, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.61, "toppings": 0.5866666666666667, "bowl": 0.8566666666666667, "broth": 0.57, "steam": 0.5533333333333333, "composition": 0.8533333333333333 }, "reward": 0.6653333333333333, "sampleStdDev": 0.0018929694486000733, "judgments": [ { "noodles": { "score": 0.61, "reason": "Lifted strands are continuous and gently bent, but sparse and nearly parallel; the submerged strands form flat bands with little convincing overlap or volume." }, "toppings": { "score": 0.59, "reason": "Leaves with veins, capped mushrooms, tofu and scallion rings provide readable variety. Shapes remain simplified and largely untextured, especially the smooth mushrooms and rigid rectangular nori." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, curved walls, attached foot and subtle dark highlights establish a convincing lacquer bowl. Decorative bands follow the body consistently; contact with the dark tabletop is slightly indistinct." }, "broth": { "score": 0.57, "reason": "Warm amber gradients and restrained surface highlights suggest soup, but the surface is quite flat and opaque, with weak immersion cues around toppings." }, "steam": { "score": 0.54, "reason": "Translucent blurred patches change position across normal frames without obscuring the food. They read more as detached diffuse clouds than connected wisps emerging from the soup." }, "composition": { "score": 0.85, "reason": "The single centered bowl, diagonal chopsticks and warm-on-dark palette create a clear, uncluttered focal composition across desktop and mobile. No text or UI distracts; the broad background glow slightly dominates the upper space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, scallions, corn and seaweed; no visibly animal-derived ingredients are present." } }, { "noodles": { "score": 0.61, "reason": "Lifted strands are continuous, evenly thick and gently bent, but remain nearly parallel; noodles in the soup form sparse flat lines with little tangled volume." }, "toppings": { "score": 0.58, "reason": "Leaf veins, mushroom caps and stems, tofu edges and scallion rings provide readable variety. Shapes are simplified and mostly flat, while nori resembles a rigid rounded card." }, "bowl": { "score": 0.85, "reason": "Consistent elliptical rim, curved walls, aligned decorative bands and a small attached foot create coherent depth and perspective. Subtle highlights suggest lacquer, though the dark base and contact are weakly defined." }, "broth": { "score": 0.57, "reason": "Warm amber shading and restrained curved highlights suggest soup, but the surface is rather solid and flat; most ingredients lack convincing immersion or liquid contact." }, "steam": { "score": 0.56, "reason": "Translucent blurred patches change position across normal frames and leave the food visible. They read more as floating diffuse blobs than connected wisps emerging from the soup." }, "composition": { "score": 0.86, "reason": "A centered dish, diagonal chopsticks, restrained warm palette and ample dark negative space establish a clear focal point without text or clutter. Mobile framing remains balanced; the broad background haze slightly dominates the atmosphere." }, "vegan": { "score": 1, "reason": "Visible mushrooms, leafy greens, tofu-like cubes, seaweed and vegetable garnishes plausibly form vegan ramen; no animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.61, "reason": "Lifted strands are continuous, evenly thick and gently curved, but sparse and nearly parallel; noodles in the soup form flat bands with little overlapping volume." }, "toppings": { "score": 0.59, "reason": "Leaf veins, mushroom caps and stems, tofu and scallion rings provide readable variety. Shapes remain simplified and largely textureless, especially the rounded rectangular nori and smooth tofu." }, "bowl": { "score": 0.86, "reason": "Coherent elliptical rim, curved walls, attached foot and subtle highlights convincingly support the illustrated bowl. Decorative bands follow its curvature; no material construction defect is observed." }, "broth": { "score": 0.57, "reason": "Warm amber shading and fine surface arcs suggest soup, but the surface is quite flat and opaque, with limited immersion cues and several garnishes appearing pasted on top." }, "steam": { "score": 0.56, "reason": "Normal frames show translucent haze changing position above the soup without obscuring it. Steam mostly reads as diffuse floating patches rather than connected irregular wisps emerging from the liquid." }, "composition": { "score": 0.85, "reason": "The centered dish, diagonal chopsticks and restrained warm palette create a clear focal composition against the dark setting, with no distracting text or UI. Mobile retains the complete dish; the broad background haze slightly competes for attention." }, "vegan": { "score": 1, "reason": "Visible toppings read as mushrooms, leafy greens, tofu, seaweed, scallions and corn; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5166666666666667, "toppings": 0.6066666666666667, "bowl": 0.7333333333333333, "broth": 0.5833333333333334, "steam": 0.6066666666666667, "composition": 0.77 }, "reward": 0.628, "sampleStdDev": 0.014730919862656249, "judgments": [ { "noodles": { "score": 0.5, "reason": "The lifted noodle strands are smooth and warm yellow, with believable hang from the chopsticks. The noodles in the bowl are thin, pale, and hard to read as a mass, so the bowl itself shows little volume or overlap." }, "toppings": { "score": 0.62, "reason": "Bok choy, shiitake caps, tofu cubes, nori, scallion rings and corn are all recognizable and varied. The shapes are simple and flat, the nori is a rigid dark rectangle, and the tofu is a bit blocky." }, "bowl": { "score": 0.74, "reason": "The bowl has a coherent elliptical rim, a curved body with red and gold bands, a base, and a contact shadow. The lacquer highlights are subtle, and the base looks slightly small. No material distortion is visible." }, "broth": { "score": 0.6, "reason": "The broth is a warm orange with faint concentric ripples and a glow. It looks somewhat flat and opaque, and toppings sit on it with little immersion or occlusion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and shift across the frames. They sit high above the bowl and look blobby, with little visible connection to the broth surface. They do not hide the food." }, "composition": { "score": 0.78, "reason": "The bowl is a clear central focus on a dark, atmospheric background with a warm glow. The proportions are balanced and there is no clutter. The lower third is empty." }, "vegan": { "score": 0.93, "reason": "The toppings are tofu, mushrooms, bok choy, nori, corn and scallions, with no visible animal products." } }, { "noodles": { "score": 0.55, "reason": "The lifted noodle strands are smooth and golden with a chopstick pinch, but the noodles in the bowl are thin streaks buried in the surface. Volume and overlap are weak." }, "toppings": { "score": 0.62, "reason": "Bok choy, shiitake with stems, tofu cubes, nori, scallion rings, corn and carrot are all readable and varied. They are small and flat-stylized, and the tofu is a bit blocky." }, "bowl": { "score": 0.74, "reason": "The bowl has a coherent elliptical rim, a dark lacquer body with red and gold bands and dots, a base and a grounded shadow. The perspective is consistent. The body looks slightly flat, and the rim is thick with only subtle highlights." }, "broth": { "score": 0.6, "reason": "The warm orange broth has a gradient sheen and faint ripple rings. It reads somewhat flat and opaque, and the toppings sit on it with little immersion." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and shift across the frames and fade out. They are blobby and rise high above the bowl, away from the soup surface, and the reduced-motion frame looks different." }, "composition": { "score": 0.78, "reason": "The bowl is centered as a clear focal point in a dark, moody scene with a warm glow and a restrained palette. There is no clutter. The bowl is slightly small in the frame and there is a lot of empty space below." }, "vegan": { "score": 0.93, "reason": "Only tofu, mushrooms, bok choy, nori, vegetables and noodles are visible, with no egg or meat." } }, { "noodles": { "score": 0.5, "reason": "The lifted noodle strands are smooth and supple, but the noodles in the broth are only thin flat lines that barely read as noodles. The mass is indistinct and low in volume." }, "toppings": { "score": 0.58, "reason": "Bok choy, shiitake caps, tofu cubes, nori and scallion rings are all recognizable and varied. They are small and flat, and the nori reads as a dark rigid card. The tofu cubes look like plain blocks." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim, a curved body, a red and gold band with dots, a foot and a contact shadow. The shading is smooth and the perspective is consistent. The rim is somewhat thick and the body is a little flat." }, "broth": { "score": 0.55, "reason": "The broth is warm orange with faint concentric ripple lines and a glow. It reads somewhat flat and opaque, and toppings sit on the surface without much immersion or occlusion." }, "steam": { "score": 0.62, "reason": "Soft translucent wisps drift and change shape across frames. They sit high above the bowl and look a bit blobby, and they are disconnected from the soup surface." }, "composition": { "score": 0.75, "reason": "The bowl is a clear focal point against a dark warm glow with balanced proportions and restrained colors. The chopsticks add interest. The lower third is empty." }, "vegan": { "score": 0.95, "reason": "The toppings are tofu, mushrooms, bok choy, nori, scallion and corn, and no animal products are visible." } } ], "diagnostics": { "vegan": 0.9366666666666666 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7166666666666667, "toppings": 0.7, "bowl": 0.8333333333333334, "broth": 0.7333333333333333, "steam": 0.7666666666666667, "composition": 0.8666666666666667 }, "reward": 0.76, "sampleStdDev": 0.03436931771216887, "judgments": [ { "noodles": { "score": 0.7, "reason": "The chopstick-lifted strands are supple, continuous, gently waved and read as noodles, but in the bowl itself noodles are only implied by a few thin squiggle lines on the broth, leaving most of the noodle body unseen." }, "toppings": { "score": 0.65, "reason": "Mushroom caps with stems, tofu blocks, scallion rings, chili bits and greens are recognizable and varied, but shapes are quite flat and stylized; the bok choy reads as two glossy green blobs, and the white rectangles with green spirals look plasticky and narutomaki-like." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with visible wall thickness, dark ceramic body with red and gold decorative bands and dot pattern, stable perspective and a grounded contact shadow; no broken joins, though the foot is barely suggested." }, "broth": { "score": 0.7, "reason": "Warm orange liquid with soft gradient and thin swirl lines; toppings sit on the surface with small shadows and partial occlusion, though the surface reads slightly uniform and toppings appear more pasted than immersed." }, "steam": { "score": 0.75, "reason": "Translucent gray wisps change shape and position across the time-ordered frames (rising near the bowl in pointer-right, drifting top-left in pointer-left), fading without hiding food; slightly blobby but clearly animated, and static in reduced-motion frames as expected." }, "composition": { "score": 0.85, "reason": "Centered dish as clear focal point, warm rim glow against dark atmospheric backdrop, chopsticks adding diagonal interest, no text or clutter; only the slightly murky steam blobs distract." }, "vegan": { "score": 0.5, "reason": "Mushrooms, greens, tofu, scallions, corn/chili and nori are plant-based, but the white rectangles with green spiral centers resemble narutomaki fish cakes, making the ingredient set ambiguous." } }, { "noodles": { "score": 0.7, "reason": "The chopstick-lifted bundle shows continuous cream-colored strands with gentle bends and coherent convergence at the tips, reading clearly as noodles; however, strands are fairly uniform and stiff in their wave pattern, and in-bowl noodles are barely visible beneath the toppings, leaving noodle presence thin." }, "toppings": { "score": 0.7, "reason": "Good variety and readability: green bok-choy-like leaves with vein lines, brown mushroom caps with stems and cut faces, green scallion rings, carrot strips, corn dots, chili flecks, and a dark nori sheet. Defects: tofu is plain featureless white blocks and the nori is a fairly geometric rounded rectangle, both slightly card-like." }, "bowl": { "score": 0.85, "reason": "Coherent construction: elliptical rim with visible wall thickness, body tapering smoothly to a small foot, red band with white dot pattern that follows the bowl curvature, gold accent band, subtle side highlight, and a grounding contact shadow. No broken joins or perspective conflicts observed; only minor flatness in the decorative bands." }, "broth": { "score": 0.7, "reason": "Orange tonkotsu-style surface fills the rim with a warm glowing center, soft swirl arcs, and scattered chili-oil specks; ingredients sit at the surface plane with reasonable contact. Immersion depth is modest \u2014 toppings rest on the surface without waterline darkening, and the light concentric glow reads slightly artificial." }, "steam": { "score": 0.75, "reason": "Across desktop-1\u21923 the translucent gray wisps change shape continuously, drifting and reforming above the bowl without becoming opaque fog or identical mechanical plumes; they do not obscure the food. Wisps stay somewhat diffuse and detached from the visible broth surface. Reduced frames are appropriately static." }, "composition": { "score": 0.85, "reason": "Well-balanced centered bowl against a dark atmospheric backdrop with a restrained warm radial glow; chopsticks add a natural focal diagonal, no text/UI or clutter, and the mobile crop keeps the dish fully framed. Background glow stays subordinate to the dish." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-derived: bok choy, mushrooms, tofu, scallions, carrot, corn, chili oil, and nori; no egg, meat, fish, or dairy shapes appear." } }, { "noodles": { "score": 0.75, "reason": "The chopstick noodle lift is the strongest element: continuous cream-colored strands with gentle curves and consistent thickness descending into the broth. In-broth noodles are faint pale lines mostly hidden under toppings, and the strands hang somewhat parallel/uniform, but nothing reads as broken wire or maze scribble." }, "toppings": { "score": 0.75, "reason": "Good variety: brown shiitake-style mushroom caps, standing dark nori sheet, white tofu cubes, scallion rings, corn dots, carrot shreds, and leafy green bok-choy-like shapes. Stylized but readable; the greens are slightly blobby and the tofu cubes a bit featureless/flat." }, "bowl": { "score": 0.85, "reason": "Coherent ceramic bowl with a clean elliptical rim, consistent tapering dark body, red and gold decorative bands with dot pattern, visible foot with contact shadow, and a subtle sheen highlight on the right wall. Perspective of rim, body and base agree; no distortion or detached parts observed." }, "broth": { "score": 0.8, "reason": "Warm amber surface with pale swirl highlights, scattered oil/seasoning dots, and soft shadows grounding each topping so ingredients appear seated in the liquid rather than pasted on. Slight flatness/repetitive swirl arcs keep it from higher marks." }, "steam": { "score": 0.8, "reason": "Across the normal frames the wisps above the bowl clearly change shape, position and dispersion between desktop-1/2/3 and both pointer frames \u2014 rising translucent grey plumes that thin and drift without ever obscuring the food. Not static; only slightly blobby puff shapes." }, "composition": { "score": 0.9, "reason": "Centered focal dish on a dark atmospheric ground with a warm radial glow, subtle floating specks and vignette; no text, UI or clutter, and the mobile viewport crops cleanly around the bowl. Balanced and attractive presentation." }, "vegan": { "score": 1.0, "reason": "All visible ingredients are plant-based: tofu, mushrooms, leafy greens, nori, scallion, corn, carrot and noodles in a plant broth; no egg, meat or other animal-derived toppings appear." } } ], "diagnostics": { "vegan": 0.8333333333333334 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9233333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.30000000000109, "frameIntervalP95Ms": 83.40000000000146, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "a1c6e71ad18da76d81f18fd24dee3b607d92f0997c74bf248a00c8059323552a", "gradedAt": "2026-10-12T20:01:37.847938+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.3, "toppings": 0.32333333333333336, "bowl": 0.4888888888888889, "broth": 0.29888888888888887, "steam": 0.2488888888888889, "composition": 0.4777777777777778 }, "reward": 0.3536111111111111, "runId": "meta/muse-wonder-1.0/high", "rewardStdDev": 0.3085135659421971, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5633333333333334, "bowl": 0.79, "broth": 0.41333333333333333, "steam": 0.3466666666666667, "composition": 0.7566666666666667 }, "reward": 0.5678333333333334, "sampleStdDev": 0.002362907813126275, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin pale strands have continuous curved bends, but repeated shallow loops form a flat, low-contrast tangle with little convincing overlap or volume." }, "toppings": { "score": 0.58, "reason": "Veined green leaves, scored mushroom slices, corn and tofu provide readable variety. Most remain small flat symbols, and the nori resembles a thick, smooth rectangular card." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, curved walls, decorative bands and grounded shadow form a coherent vessel. The cavity is shallowly described and the rectangular foot looks abrupt rather than naturally integrated." }, "broth": { "score": 0.42, "reason": "Warm golden color and a broad highlight suggest soup, but the sharply bounded central oval reads as a flat disc, with little immersion or liquid contact around toppings." }, "steam": { "score": 0.35, "reason": "Normal frames show changing positions and sizes of translucent particles above the soup. These are disconnected circular bokeh-like blobs rather than irregular rising wisps." }, "composition": { "score": 0.75, "reason": "A centered bowl, restrained warm palette and uncluttered dark setting create a clear focal dish. The tall particle field draws excessive attention, especially in the mobile layout." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, corn and seaweed alongside noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Pale continuous strands have curved bends, but mostly form flat, repeated horizontal loops with little convincing overlap or piled volume." }, "toppings": { "score": 0.56, "reason": "Veined green leaves, scored mushroom pieces, corn and pale tofu provide recognizable variety. Shapes remain simplified and sparsely textured; the nori resembles a thick rounded card rather than irregular seaweed." }, "bowl": { "score": 0.79, "reason": "Consistent elliptical rim, curved walls, fine decorative bands and a contact shadow create a coherent supported bowl. The broad orange interior and blocky foot have limited material and depth detail." }, "broth": { "score": 0.42, "reason": "Warm golden color and a soft highlight suggest soup, but the central pale oval reads as a separate flat disc. Ingredients appear laid onto it with little immersion or liquid contact." }, "steam": { "score": 0.34, "reason": "Translucent particles change position and size across normal frames without obscuring the food. They read as disconnected circular bokeh blobs rather than irregular rising steam wisps." }, "composition": { "score": 0.76, "reason": "A centered dish, restrained amber palette and dark uncluttered setting establish a clear focal point. The food occupies a relatively small area of the large bowl, and the tall particle column competes for attention, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible tofu-like pieces, mushrooms, greens, corn, seaweed and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin continuous yellow strands have curved bends, but mostly form repeated flat arcs with little readable overlap or piled volume." }, "toppings": { "score": 0.55, "reason": "Veined green leaves, scored brown mushrooms, corn and pale tofu provide recognizable variety. Shapes remain very simplified, and the nori resembles a thick, smooth rounded card." }, "bowl": { "score": 0.8, "reason": "The elliptical rim, curved walls, fine decorative bands and grounded shadow form a coherent bowl. Material shading is restrained, though the broad orange interior and rectangular foot feel somewhat schematic." }, "broth": { "score": 0.4, "reason": "Warm golden color is appetizing, but the sharply bounded central ellipse reads as a flat inset disc. Ingredients show little immersion, contact shading or liquid interaction." }, "steam": { "score": 0.35, "reason": "Normal frames show changing positions and sizes of translucent rising particles without obscuring the food. However, isolated circular blobs resemble bokeh rather than irregular connected steam wisps." }, "composition": { "score": 0.76, "reason": "A centered dish, warm palette and uncluttered dark setting create a clear focal presentation. The tall field of glowing particles competes somewhat with the comparatively small food area, especially on mobile." }, "vegan": { "score": 1, "reason": "Visible noodles, greens, mushrooms, corn, tofu-like pieces and dark seaweed plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.4, "toppings": 0.4066666666666667, "bowl": 0.6766666666666666, "broth": 0.48333333333333334, "steam": 0.4, "composition": 0.6766666666666666 }, "reward": 0.493, "sampleStdDev": 0.012529964086141652, "judgments": [ { "noodles": { "score": 0.4, "reason": "Noodles are thin pale scribbled strands confined to a small oval island, with a few arcs. They read as tangled wire and lack volume or supple overlap." }, "toppings": { "score": 0.42, "reason": "Tiny leaves, shiitake slices, tofu cubes, corn and a dark nori rectangle are recognizable. They are very small and sparse, and the nori is a rigid block with no irregular edges." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, tapered walls, a foot, groove lines, a gradient and a cast shadow. It is clean and consistent in perspective, but somewhat flat in material and the base looks slightly boxy." }, "broth": { "score": 0.5, "reason": "The orange broth has a soft gradient and a sheen highlight, but it looks flat and opaque like a target disc. The noodle island appears pasted on top, with little immersion." }, "steam": { "score": 0.4, "reason": "The steam is bokeh-like circular blobs that drift and change between frames. They read as floating discs rather than translucent wisps, and they are disconnected from the soup." }, "composition": { "score": 0.7, "reason": "The bowl is centered and is a clear focal point against a dark, warm atmosphere with good color harmony. The food is small within the bowl, and the lower third is empty." }, "vegan": { "score": 0.9, "reason": "Only tofu, mushrooms, greens, corn, nori and noodles are visible. There is no apparent meat, egg or fish." } }, { "noodles": { "score": 0.4, "reason": "Noodles are thin pale strands clustered in a small central oval, with a few arcs. They look like scribbled lines with little volume or overlap, and the mass is small relative to the bowl." }, "toppings": { "score": 0.4, "reason": "Toppings include small green leaves, mushroom slices, tofu cubes, corn dots and a dark nori rectangle. The set is varied, but the pieces are tiny and generic. The nori reads as a rigid dark block and the tofu as small white blocks." }, "bowl": { "score": 0.65, "reason": "Coherent dark lacquer bowl with a clear rim, tapered walls, decorative lines and a foot. It sits on a contact shadow with consistent perspective. Shading is somewhat flat, and the broth looks very high in the bowl." }, "broth": { "score": 0.45, "reason": "The broth is a smooth orange gradient with a soft highlight. It reads as a flat disc, and the noodle patch looks like a pale oval pasted on top, not immersed in the liquid." }, "steam": { "score": 0.4, "reason": "Steam is made of bokeh-like round translucent circles that drift and change between frames. They read as floating blobs, not wispy steam, though they are soft and do not hide the food." }, "composition": { "score": 0.65, "reason": "Centered bowl with a warm glow on a dark background, no text or clutter. The bowl is a clear focal point and the colors are attractive. Large empty space sits at the bottom." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are tofu, mushrooms, greens, corn, nori and noodles. There is no meat, egg or fish." } }, { "noodles": { "score": 0.4, "reason": "Noodles sit in a small pale oval patch as thin, faint strands, with a few arcs visible. They are tiny, low-contrast and read as scribbly. There is little volume or overlap, and they do not fill the bowl." }, "toppings": { "score": 0.4, "reason": "Small green leaves, mushroom slices, tofu cubes, corn dots and a dark nori rectangle are recognizable. They are tiny and sparse, and the nori is a stiff rounded block. Detail and variety are limited." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, tapered body, foot, groove lines, gradient shading and a contact shadow. It looks competent. Minor defects: the foot is a flat dark rectangle, and the material is somewhat flat." }, "broth": { "score": 0.5, "reason": "The orange broth has a gradient and soft highlight, but it reads as a flat, opaque, saturated disc. The noodle patch looks pasted on top, with little immersion or ripple." }, "steam": { "score": 0.4, "reason": "Steam is made of bokeh-like circular blobs that drift and change between frames. It does not look like wisps and reads as floating discs, though it is translucent and does not hide the food." }, "composition": { "score": 0.68, "reason": "The bowl is centered in a dark, warm atmosphere, so the focus is clear and the palette is attractive. Food detail is small relative to the bowl, and a large empty area sits below it." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are tofu, mushrooms, greens, corn, nori and noodles. There are no animal products." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "toppings": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "bowl": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "broth": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "steam": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "composition": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." }, "vegan": { "score": 0, "reason": "Placeholder evaluation overwritten below \u2014 see corrected entry." } }, { "noodles": { "score": 0, "reason": "placeholder" }, "toppings": { "score": 0, "reason": "placeholder" }, "bowl": { "score": 0, "reason": "placeholder" }, "broth": { "score": 0, "reason": "placeholder" }, "steam": { "score": 0, "reason": "placeholder" }, "composition": { "score": 0, "reason": "placeholder" }, "vegan": { "score": 0, "reason": "placeholder" } }, { "noodles": { "score": 0, "reason": "Noodles are a thin pale scribble of hairline strands with a few wire-like loops confined to a small central patch; flat, sparse and lacking volume or believable tangling in the broth." }, "toppings": { "score": 0, "reason": "Toppings are generic flat symbols: white cubes, a rigid dark rectangle for nori, dot-like corn/chili and simple leaf shapes sitting pasted on the surface with no texture, occlusion or organic detail; variety is present but all forms are featureless." }, "bowl": { "score": 0, "reason": "Bowl geometry is coherent: elliptical rim, tapering body with decorative ring bands, visible foot and grounded shadow, consistent high-angle perspective and rim highlight. Minor: hard edge between inner pool and orange rim reads slightly pasted." }, "broth": { "score": 0, "reason": "Broth surface is a flat opaque yellow ellipse with a hard-edged orange ring; no sheen, depth or texture, and ingredients show no contact shadows or immersion, reading as placed on a disc rather than in liquid." }, "steam": { "score": 0, "reason": "Steam is disconnected translucent bokeh orbs, not wisps; while positions shift and fade upward across the timed frames, the form is explicitly blob-like rather than irregular rising plumes." }, "composition": { "score": 0, "reason": "Balanced centered dish on a warm dark ground, clear focal bowl, atmospheric glow, no clutter or text, and the mobile frame scales cleanly; presentation is restrained and readable." }, "vegan": { "score": 0, "reason": "Visible ingredients are tofu, herbs, corn, small red bits, mushroom-like slivers, nori, noodles and plant-colored broth; no animal-derived toppings appear." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.6333333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 50.100000000000364, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "37d5fa80117395b79c7c432692fb28b870edd6ab831aefb4b4b06f5ce43e68e9", "gradedAt": "2026-10-12T20:03:39.556720+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5511111111111111, "toppings": 0.5466666666666666, "bowl": 0.7711111111111111, "broth": 0.5733333333333334, "steam": 0.63, "composition": 0.7966666666666666 }, "reward": 0.6326111111111111, "runId": "meta/muse-wonder-1.0/medium", "rewardStdDev": 0.07282595945382644, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.52, "toppings": 0.47333333333333333, "bowl": 0.82, "broth": 0.48666666666666664, "steam": 0.5733333333333333, "composition": 0.85 }, "reward": 0.606, "sampleStdDev": 0.004500000000000004, "judgments": [ { "noodles": { "score": 0.5, "reason": "Continuous golden strands and a curved central bundle read as noodles, but most are flat parallel surface lines with little overlap, volume, or convincing immersion." }, "toppings": { "score": 0.48, "reason": "Greens, tofu, corn, mushrooms, and nori provide readable variety, but simplified oval mushroom disks, featureless tofu squares, and rectangular nori dominate; organic texture and detail are sparse." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, smoothly shaded curved walls, integrated foot, and grounding shadow form a coherent vessel. Rim highlights convey thickness, though the nearly black interior edge and understated material detail reduce ceramic richness." }, "broth": { "score": 0.5, "reason": "Warm amber color and gentle shading distinguish the soup surface, but it looks largely like a solid flat fill, with minimal liquid sheen or contact cues around ingredients." }, "steam": { "score": 0.58, "reason": "Normal frames show changing translucent vapor above the soup without obscuring the food. However, it consists mostly of separated blurry puffs rather than connected, irregular rising wisps." }, "composition": { "score": 0.85, "reason": "A single centered dish, diagonal chopsticks, restrained warm palette, and uncluttered dark setting create a clear focal composition across desktop and mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy vegetables, corn, and seaweed with noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.5, "reason": "Continuous golden strands and a curved lifted bundle are readable, but most strands form flat, parallel surface lines with little overlap, volume, or convincing immersion." }, "toppings": { "score": 0.48, "reason": "Greens, tofu, corn, mushrooms, and nori are recognizable, but featureless square tofu, oval mushroom disks, circular kernels, and rectangular nori dominate; organic texture and detail are sparse." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, rounded shaded walls, lower decorative band, and attached foot form coherent depth and perspective. Warm highlights suggest a dark ceramic finish, though the foot and contact shadow are somewhat indistinct." }, "broth": { "score": 0.47, "reason": "Warm amber color is appetizing, but the surface reads as a smooth gradient plane with minimal liquid sheen or texture. Ingredients appear laid on top rather than partially immersed." }, "steam": { "score": 0.57, "reason": "Normal frames show changing translucent patches above the soup without obscuring the food. However, these look like disconnected blurred blobs rather than connected, irregular rising wisps; continuous flow is not established by the captures." }, "composition": { "score": 0.85, "reason": "A centered focal bowl, diagonal chopsticks, restrained warm palette, and uncluttered dark setting create a balanced presentation on desktop and mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, leafy greens, corn, and seaweed alongside noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.56, "reason": "Continuous golden strands and a curved lifted bundle read as noodles, but mostly parallel surface lines have little overlap, volume, or convincing submerged continuation." }, "toppings": { "score": 0.46, "reason": "Greens, tofu, corn, mushrooms, and nori are recognizable, but simple oval mushroom disks, featureless tofu squares, circular kernels, and rectangular nori dominate; organic texture is sparse." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, smoothly shaded curved walls, integrated foot, and contact shadow form a coherent supported vessel. Rim highlights convey thickness, though the material finish is rather plain." }, "broth": { "score": 0.49, "reason": "Warm amber shading defines the soup surface, but it resembles a flat filled plane with little liquid sheen or depth; ingredients lack convincing wet contact and immersion." }, "steam": { "score": 0.57, "reason": "Normal frames show changing translucent clouds above the soup without obscuring it. However, the steam consists largely of disconnected blurry blobs rather than connected, tapering wisps." }, "composition": { "score": 0.85, "reason": "A centered focal bowl, diagonal chopsticks, restrained warm palette, and uncluttered dark setting create a balanced presentation across desktop and mobile. No material compositional defect is observed." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, corn, and seaweed, with noodles and plant-compatible broth; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.5, "bowl": 0.6933333333333334, "broth": 0.55, "steam": 0.5833333333333334, "composition": 0.7066666666666667 }, "reward": 0.5768333333333334, "sampleStdDev": 0.0038837267325770035, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin wavy golden strands read as noodles with some overlap and a small lifted bundle. They are fairly flat, short and scribbly, with limited volume and no real immersion in the broth." }, "toppings": { "score": 0.5, "reason": "Bok choy, tofu cubes, corn kernels, shiitake slices, a nori piece and greens are all identifiable. The tofu blocks are plain and small, the mushrooms are simple ovals, and the nori is a rigid dark rectangle. Variety is good but the execution is generic." }, "bowl": { "score": 0.68, "reason": "The bowl has a coherent elliptical rim, a deep curved body, a foot, a decorative band, a contact shadow and a soft gradient that reads as lacquer. The rim's inner lip is slightly flat and the base is hidden in shadow, but there is no major distortion." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a radial glow and red chili flecks, which makes it appetizing. It looks fairly flat, with little surface depth, ripples or immersion cues, and the toppings sit on it rather than in it." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape between frames, so there is visible motion. They are blobby and muddy, and read more as smoky bokeh than as wispy steam, but they do not hide the food." }, "composition": { "score": 0.72, "reason": "The bowl is centered and clearly the focal point against a dark, moody background with warm lighting. The chopsticks add a diagonal accent. The palette is cohesive, though the lower part of the frame is empty." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are tofu, bok choy, corn, mushrooms, nori and noodles. No meat, egg or fish appears. The orange curl is probably carrot." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands wave across the surface and read as noodles, but they are mostly flat squiggles with little overlap or volume. The small lifted bundle is a nice touch but looks stiff." }, "toppings": { "score": 0.5, "reason": "Bok choy, tofu cubes, corn, shiitake slices, greens and nori are all recognizable. The tofu is simple blocks, the mushrooms are flat ovals, and the nori is a plain dark rectangle. Everything is small and sparse." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, curved walls, a foot ring, a band line and a ground shadow. Shading is smooth and the perspective is consistent. The material looks somewhat matte and flat, with no strong highlights." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a glow gradient and a plausible depth. It looks fairly flat, with scattered red chili dots and little sheen. Ingredients sit on the surface with weak immersion or ripples." }, "steam": { "score": 0.55, "reason": "Soft translucent smoky blobs rise and change shape between frames. They look blurry and blobby rather than wispy, and they sit somewhat detached from the bowl surface." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point on a dark, moody background with warm lighting. The chopsticks add diagonal interest. There is no clutter or text. The lower part of the frame is empty." }, "vegan": { "score": 0.95, "reason": "Tofu, corn, bok choy, mushrooms, nori and noodles are all plant-based. The orange curl could be read as carrot or ambiguous, but nothing visibly animal-derived appears." } }, { "noodles": { "score": 0.5, "reason": "Thin golden strands read as noodles with some wavy bends and a lifted bunch near the chopsticks. They are flat, short and scribble-like, with little volume or overlap, and they sit on the surface rather than in the broth." }, "toppings": { "score": 0.5, "reason": "Bok choy, tofu cubes, corn kernels, shiitake slices, nori and greens are all recognizable and add variety. They are small, simple vector shapes. The tofu cubes are plain blocks, the corn is just dots, and the nori is a dark rectangle." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a curved body, a foot, a decorative band and a contact shadow, with a lacquer-like gradient. It looks competent and consistent in perspective. It is fairly dark and plain, with a faint foot." }, "broth": { "score": 0.55, "reason": "The warm orange broth has a radial glow and a dark rim that give some depth. It reads as a fairly flat gradient with little sheen or immersion detail, and the toppings seem to float on it." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps rise above the bowl and change shape between frames, so there is real motion. They are blobby and blurry, and read a bit like smoke puffs. They are detached from the surface and fade out well." }, "composition": { "score": 0.7, "reason": "The bowl is centered and clearly the focus, set in a dark atmospheric scene with warm lighting and a diagonal chopstick accent. Colors are harmonious and there is no clutter. The bottom third is empty and the scene is quite dark." }, "vegan": { "score": 0.92, "reason": "Tofu, mushrooms, bok choy, corn, nori and noodles all look plant-based. The orange curls could be read as carrot, and no meat, egg or fish is visible." } } ], "diagnostics": { "vegan": 0.9233333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6333333333333333, "toppings": 0.6666666666666666, "bowl": 0.8, "broth": 0.6833333333333333, "steam": 0.7333333333333333, "composition": 0.8333333333333334 }, "reward": 0.7150000000000001, "sampleStdDev": 0.014999999999999958, "judgments": [ { "noodles": { "score": 0.7, "reason": "Thin continuous tan strands with gentle waves spread across the broth, including a lifted bundle and a curly cluster; strands read as noodles with coherent bends, though they lie fairly flat and sparse with limited volume or layered overlap in the stylized scene." }, "toppings": { "score": 0.65, "reason": "Good variety: bok choy with leaf veins, sliced mushrooms, corn, scallion bits, a thin dark nori sheet and chili dots; the three tofu pieces are plain near-featureless white squares and the mushrooms are slightly glossy ovals, so some forms are simplistic." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim with a visible tan ceramic rim edge, smooth dark body with a left-side highlight, decorative band near the base and a grounded soft shadow; foot is slightly stubby but perspective between rim, wall and base is consistent." }, "broth": { "score": 0.7, "reason": "Warm orange surface with a soft radial glow, darker toward the rim, restrained sheen and scattered chili/corn specks; tofu cubes cast small contact shadows so toppings sit in the liquid, but the surface reads somewhat flat with little texture or depth variation." }, "steam": { "score": 0.75, "reason": "Translucent gray-brown wisps above the bowl that visibly shift position and shape between desktop-1, desktop-2, desktop-3 and pointer frames, rising and dispersing without hiding the food; puffs are slightly blobby rather than smoothly drawn-out plumes. Reduced frames correctly freeze it." }, "composition": { "score": 0.85, "reason": "Well-balanced centered bowl with warm broth glow against a dark vignette, chopsticks resting on the rim as an accent, subtle ambient particles and a clean table plane; the dish stays the clear focal point with no text or clutter." }, "vegan": { "score": 0.7, "reason": "Visible ingredients are tofu, bok choy, mushrooms, corn, scallions, nori and chili\u2014plausibly vegan; the orange spiral shape is ambiguous and could read as a fish-cake swirl, so an intermediate score applies." } }, { "noodles": { "score": 0.6, "reason": "Thin continuous golden strands with a lifted bundle at the chopsticks read as noodles, but most strands are repetitive flat arcs spanning the bowl edge-to-edge with little volume or submerged depth, giving a slightly mechanical ribbon pattern." }, "toppings": { "score": 0.7, "reason": "Recognizable variety: bok choy leaves with lighter inner blades, shaded tofu cubes, sliced brown mushrooms, a dark nori sheet at the rim, scallion/spice dots and an orange chili-oil squiggle. Shapes are flat and simple (plain square tofu, generic dots), but nothing reads as plastic or featureless." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: clean elliptical rim with inner lip and top highlight, deep smooth body, gold accent band, small foot with grounded contact shadow. No broken joins or perspective mismatch; only minor shortcoming is limited material detail due to the very dark glaze." }, "broth": { "score": 0.7, "reason": "Warm amber liquid surface with a believable inner glow, restrained sheen and small occlusion shadows under the tofu cubes, so toppings sit in the soup. Slightly gradient-flat look and a faint circular ripple ring keep it from reading fully wet and textured." }, "steam": { "score": 0.7, "reason": "Normal temporal frames show the translucent brownish wisps reconfiguring continuously (desktop-1/2/3 and pointer frames all differ), rising, spreading and fading without hiding the food. Wisps are a bit blobby and loosely connected rather than drawn as fine plumes; reduced frames are correctly frozen." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with dark atmospheric vignette, warm rim light, diagonal chopsticks as accent and rising steam; mobile crop also holds together. Food stays the clear focal point with no text, UI or clutter; presentation is simple but polished." }, "vegan": { "score": 1.0, "reason": "Visible ingredients are tofu, leafy greens, mushrooms, nori, scallions/chili and noodles in plant broth; no eggs, meat, fish or dairy appear anywhere." } }, { "noodles": { "score": 0.6, "reason": "Thin continuous yellow strands with gentle bends and a lifted arc near the chopstick tips read as noodles, but they are mostly flat, parallel ribbon squiggles lying on the surface with little volume, overlap depth or immersion." }, "toppings": { "score": 0.65, "reason": "Recognizable variety: bok choy leaves, four shaded tofu cubes, scallion rings, red chili flecks, a nori rectangle and mushroom-like tan slices. Weaknesses: the nori is a plain rigid rectangle, mushroom slices are simple ellipse-and-dot shapes, and everything sits fairly flat." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: elliptical rim with a lighter back band, smooth symmetric curved wall, base ring line, foot and soft cast shadow; perspective is consistent. Only minor shortcomings: stubby foot and no contact shadow under the chopsticks." }, "broth": { "score": 0.65, "reason": "Warm amber radial glow reads as hot liquid with darker edges for depth, and toppings have small shadows suggesting contact. It lacks sheen, surface texture and stronger occlusion, so ingredients look slightly pasted on rather than immersed." }, "steam": { "score": 0.75, "reason": "Translucent gray-brown wisps above the bowl change arrangement across desktop-1/2/3 and pointer frames, rising and spreading without hiding the food; slightly blobby discrete puffs keep it from scoring higher." }, "composition": { "score": 0.8, "reason": "Balanced centered dish with warm glowing focal point against a dark atmospheric backdrop, diagonal chopstick accent, subtle floor line and sparkles, no text or clutter; steam column sits slightly off-center." }, "vegan": { "score": 0.75, "reason": "Tofu, bok choy, scallion, chili and nori are plausibly vegan, but the tan sliced discs and orange curl pattern are ambiguous and could be read as fish-cake-style toppings, preventing full credit." } } ], "diagnostics": { "vegan": 0.8166666666666667 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9133333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 33.30000000000109, "frameIntervalP95Ms": 33.400000000001455, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "c68b265594e58355dc0694fae42fa0994b70882ca3be8568a1a0c2626f6e29e7", "gradedAt": "2026-10-12T20:03:25.697247+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5877777777777777, "toppings": 0.5333333333333333, "bowl": 0.7822222222222223, "broth": 0.6122222222222222, "steam": 0.6033333333333333, "composition": 0.7911111111111111 }, "reward": 0.6381666666666667, "runId": "meta/muse-wonder-1.0/minimal", "rewardStdDev": 0.1054636588277371, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.48, "toppings": 0.5, "bowl": 0.8133333333333334, "broth": 0.5533333333333333, "steam": 0.5766666666666667, "composition": 0.84 }, "reward": 0.6095, "sampleStdDev": 0.0025000000000000022, "judgments": [ { "noodles": { "score": 0.48, "reason": "Lifted strands have continuous gentle bends, but the soup is dominated by angular, flat zigzag strands with little supple overlap or volume." }, "toppings": { "score": 0.5, "reason": "Green onion rings, leaves, corn and mushroom-like slices provide variety, but remain tiny simplified symbols; the nori looks like a thick rectangular block and textures are largely absent." }, "bowl": { "score": 0.8, "reason": "Coherent elliptical rim, rounded walls and joined foot establish convincing depth. Soft blue-gray shading suggests ceramic, though the rim and material highlights are quite uniform." }, "broth": { "score": 0.55, "reason": "Warm amber color and scattered subtle surface patches suggest soup, but the surface is mostly flat and opaque, with limited immersion or contact around the ingredients." }, "steam": { "score": 0.58, "reason": "Normal frames show translucent vapor appearing and expanding upward without obscuring the food. It forms soft bright blobs rather than fine irregular wisps, and continuous fading is not clearly demonstrated." }, "composition": { "score": 0.84, "reason": "The centered blue bowl contrasts attractively with the amber soup and restrained dark tabletop; diagonal chopsticks create a clear focal gesture with no distracting UI. Mobile framing leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict vegetables, mushrooms, tofu and seaweed alongside noodles; no visibly animal-derived ingredient is present." } }, { "noodles": { "score": 0.48, "reason": "Lifted strands are continuous and gently curved, but the soup is dominated by flat angular zigzags with little supple overlap or volume." }, "toppings": { "score": 0.5, "reason": "Scallion rings, corn, small leaves and mushroom-like slices provide variety, but most are tiny flat symbols; the nori reads as a thick rectangular block and textures are minimal." }, "bowl": { "score": 0.82, "reason": "A coherent elliptical rim, rounded shaded walls and attached foot establish convincing depth and construction. Material highlights are subdued, and the broad shadow gives slightly weak contact with the table." }, "broth": { "score": 0.55, "reason": "Warm amber color and subtle surface patches suggest soup, but the surface is largely flat and opaque, with little sheen or convincing immersion of the ingredients." }, "steam": { "score": 0.57, "reason": "Normal frames show pale steam appearing and expanding upward above the soup without obscuring food. The forms are soft rounded cloudy blobs rather than fine irregular wisps; continuous fading is not established by these captures." }, "composition": { "score": 0.84, "reason": "The centered dish, diagonal chopsticks, warm soup against a cool bowl, and restrained dark tabletop create a clear uncluttered focal scene. Mobile framing remains readable, though generous empty space reduces the dish's prominence." }, "vegan": { "score": 0.95, "reason": "Visible toppings plausibly depict greens, corn, mushrooms, tofu and seaweed, with no clearly animal-derived ingredient. Some small simplified pieces remain hard to identify." } }, { "noodles": { "score": 0.48, "reason": "Lifted strands are continuous and gently curved, but the soup is dominated by angular zigzag ribbons with little supple overlap or volume." }, "toppings": { "score": 0.5, "reason": "Scallion rings, green leaves, corn and mushroom-like slices provide variety, but most are tiny flat symbols; the seaweed resembles a thick rectangular card and textures are sparse." }, "bowl": { "score": 0.82, "reason": "The elliptical rim, curved walls and attached foot form coherent geometry, with restrained blue-gray shading and a grounding shadow. The material remains somewhat flat and lacks refined ceramic highlights." }, "broth": { "score": 0.56, "reason": "Warm amber color and subtle surface patches suggest soup, but the liquid is mostly a flat opaque field, with weak immersion and contact around ingredients." }, "steam": { "score": 0.58, "reason": "Normal frames show pale steam developing and rising above the rear rim without obscuring food. Its rounded blurred patches read more as clustered fog than delicate irregular wisps." }, "composition": { "score": 0.84, "reason": "A clear central dish, diagonal chopsticks, warm food against a cool bowl, and restrained dark tabletop create a balanced presentation without text or clutter. Mobile framing leaves considerable empty space." }, "vegan": { "score": 1, "reason": "Visible greens, corn, seaweed, mushroom-like pieces and noodles plausibly form vegan ramen; no visibly animal-derived topping is present." } } ], "diagnostics": { "vegan": 0.9833333333333333 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.55, "toppings": 0.4, "bowl": 0.7, "broth": 0.5166666666666667, "steam": 0.5333333333333333, "composition": 0.6833333333333333 }, "reward": 0.55, "sampleStdDev": 0.0025000000000000022, "judgments": [ { "noodles": { "score": 0.55, "reason": "Noodles are thin, golden and plentiful, and the lifted bundle on the chopsticks reads well. In the bowl they are angular zigzag lines that look somewhat rigid and wire-like. Overlap is flat and the strands lack supple curves." }, "toppings": { "score": 0.4, "reason": "Toppings are tiny and sparse: a nori rectangle, scallion rings, corn bits, small mushroom slices, a chili ring. They are barely readable and look like symbols. The white sesame specks add little variety and no convincing texture." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a gradient-shaded wall, a foot and a ground shadow. The perspective is consistent and the form is solid. The material is plain and matte, with few highlights. The base looks slightly small." }, "broth": { "score": 0.5, "reason": "The orange broth has a radial glow and looks warm, but it is flat. There is little sense of depth, and toppings sit on top rather than being immersed. There are no ripples or reflections." }, "steam": { "score": 0.55, "reason": "Steam grows and drifts across the frames as translucent puffs, so it is clearly animated. The puffs are blobby and look somewhat grainy. They sit offset to the left and get fairly dense without hiding the food." }, "composition": { "score": 0.65, "reason": "The dark, warm scene is atmospheric, with the bowl centered as the clear focal point. The chopsticks lifting noodles add interest. The lower half is empty, and the toppings are small relative to the bowl." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are noodles, mushrooms, scallions, nori, corn and chili. There is no meat or egg, so it looks vegan." } }, { "noodles": { "score": 0.55, "reason": "The noodles read as thin, wavy strands, and the lifted bundle on the chopsticks is a nice touch. The strands in the bowl are flat, zigzag, and somewhat wiry, and the bundle looks rigid and parallel. They sit on top of the broth rather than in it." }, "toppings": { "score": 0.4, "reason": "The toppings are tiny and symbolic: green rings for scallion, a dark nori rectangle, small yellow corn blobs, and mushroom slivers. They have little texture and are scattered sparsely. The white specks look like sesame seeds. Nothing in the bowl is a substantial, convincing ingredient." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a smooth gradient on the wall, a foot, and a grounded contact shadow. The dark rim thickness is plausible. The material is somewhat flat and matte, with few specular highlights." }, "broth": { "score": 0.5, "reason": "The orange broth has a radial gradient and is readable as a liquid. It looks flat and opaque, with little surface sheen or depth. Noodles and toppings sit on top with little immersion or occlusion." }, "steam": { "score": 0.55, "reason": "Steam appears and grows across frames, drifting upward as soft translucent puffs. It looks blobby and smoky. It is also offset to the left of the bowl and gets fairly dense without looking wispy." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point against a dark, warm atmosphere with a subtle table surface. The chopsticks add a focal accent. The empty space is a little large and the lighting is simple, but there is no clutter." }, "vegan": { "score": 0.9, "reason": "The visible ingredients are noodles, scallions, corn, mushrooms, and nori. No meat, egg, or fish is visible." } }, { "noodles": { "score": 0.55, "reason": "Noodles are thin, wavy golden strands with a lifted bunch on chopsticks, which reads as ramen. Many strands are angular zigzags and look like flat scribbles. The mass is somewhat maze-like and the lifted strands are rigid parallel lines." }, "toppings": { "score": 0.4, "reason": "Toppings are tiny, sparse symbols: a nori rectangle, green peas or scallion rings, corn flecks, small mushroom slices and a chili ring. They are recognizable but very small and flat, with little texture. Scattered white seeds add clutter, and there is no substantial tofu or vegetable." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a thick dark rim band, a smooth curved wall with soft gradient shading, a foot ring and a contact shadow. The form is consistent and convincing. The material is somewhat flat and the highlights are minimal." }, "broth": { "score": 0.55, "reason": "The orange broth has a radial gradient that gives some depth and warmth. It looks fairly flat and opaque, and noodles and toppings sit on top without much immersion, ripples or sheen." }, "steam": { "score": 0.5, "reason": "Steam is absent in the first frame, then grows into soft translucent puffs across the later frames. It drifts and enlarges over time. It looks blobby and cloud-like rather than wispy, and it is positioned off to the left of the bowl center." }, "composition": { "score": 0.7, "reason": "A centered bowl on a dark, moody background with a warm glow gives a clear focal point and a good palette. The lifted noodles and chopsticks add interest. There is no clutter, though the lower third is empty." }, "vegan": { "score": 0.85, "reason": "The visible ingredients are noodles, nori, scallions, corn, mushrooms and chili, with no meat or egg visible. The pale slices could be ambiguous but read as mushroom or tofu." } } ], "diagnostics": { "vegan": 0.8833333333333333 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.7333333333333333, "toppings": 0.7, "bowl": 0.8333333333333334, "broth": 0.7666666666666667, "steam": 0.7, "composition": 0.85 }, "reward": 0.755, "sampleStdDev": 0.03605551275463986, "judgments": [ { "vegan": { "score": 1.0, "reason": "Visible ingredients are plant-based: noodles, scallion rings, nori sheet, corn, chili slices, sesame, and mushroom-like cream slices with brown rims; no eggs, meat, or fish shapes are present." }, "toppings": { "score": 0.7, "reason": "Recognizable stylized toppings \u2014 scallion rings, nori rectangle, chili arcs, corn kernels, sesame specks, and mushroom-slice half-discs \u2014 but shapes are very simplified and small in scale with limited texture detail." }, "steam": { "score": 0.7, "reason": "Steam is animated across the normal frames: absent at ~1s, faint puffs at ~2.6s, growing cloud at ~4.3s, then a broad drifting plume by ~6\u20137.9s that rises from the broth and fades without hiding the food. It reads slightly as discrete cotton-like puff clusters rather than continuous wisps." }, "noodles": { "score": 0.75, "reason": "In-bowl strands are thin, continuous, and tangled with plausible bends, and the lifted bundle merges into the broth tangle; however, the chopstick-lifted strands are uniformly parallel and rather straight/wire-like, and in-bowl squiggles are fairly uniform." }, "composition": { "score": 0.85, "reason": "Well-balanced centered dish with warm rim glow, dark atmospheric backdrop, subtle table planks and dust specks; chopsticks lifting noodles give a clear focal action with no text, UI, or clutter. Bowl scales sensibly in the mobile viewport." }, "broth": { "score": 0.8, "reason": "Smooth amber surface with a soft top-left gradient, faint oil patches, and a darker edge near the rim; noodles and toppings sit within the liquid with believable contact, though the surface is fairly flat with minimal texture or sheen variation." }, "bowl": { "score": 0.85, "reason": "Coherent footed ceramic bowl: elliptical rim with inner lip matches the body curve, smooth slate-blue gradient with a subtle highlight, proper foot ring and contact shadow on the table; no distortion, detached parts, or perspective mismatch observed." } }, { "vegan": { "score": 0.5, "reason": "Ingredients read vegan overall (scallion rings, sesame, mushroom slices, greens, chili, nori-style rectangle), but the yellow spiral discs strongly resemble narutomaki fish cake, which is animal-derived; appearance is ambiguous, so an intermediate score applies." }, "toppings": { "score": 0.75, "reason": "Good variety and scale: segmented scallion rings, sesame seeds, gilled mushroom slices, red chili rings, a nori-like dark rectangle and greens; shapes are simple flat symbols with basic detail, competent for the style but not richly textured." }, "steam": { "score": 0.75, "reason": "Across desktop frames steam is absent at 1s, small wisps at 2.6s, growing and drifting left by 4.2s and larger still in pointer frames, showing continuous rise and spread; puffs are irregular and translucent enough to keep the food visible, though somewhat blobby rather than finely wispy." }, "noodles": { "score": 0.75, "reason": "Lifted bundle from the chopsticks is a convincing set of parallel curved strands, and the in-bowl tangle uses continuous, evenly thick strands with believable bends; strands are uniformly flat-colored and somewhat repetitively squiggled, but no rigid ribbons or maze scribbles." }, "composition": { "score": 0.85, "reason": "Centered bowl with warm rim glow against a dark atmospheric background, wooden table with soft shadow, chopsticks lifting noodles as a focal accent; balanced proportions, no text or clutter, and the layout adapts cleanly to mobile." }, "broth": { "score": 0.8, "reason": "Warm orange broth with a gentle central sheen and scattered oil flecks; toppings and noodle tangle sit convincingly in the liquid, and the lifted strands emerge plausibly from the surface; no flat target discs or murky areas." }, "bowl": { "score": 0.85, "reason": "Coherent navy bowl with a clean elliptical rim, dark inner wall band, smooth gradient body, subtle left highlight, and a stable footed base resting with a believable shadow on the table; perspective is consistent with no visible distortion or detached parts." } }, { "vegan": { "score": 0.6, "reason": "Visible toppings (nori sheets, scallion pieces, chili rings, corn, sesame, brown mushroom slices) read as plant-based; one small orange X-shaped piece is ambiguous and could read as a fish-cake slice, so an intermediate score is used." }, "toppings": { "score": 0.65, "reason": "Recognizable flat-vector toppings: dark nori rectangles, scallion rings, red chili rings, corn kernels, sesame specks and brown mushroom slices with decent variety; they are quite small, sparse and uniformly icon-like, sitting flat on the surface without texture depth." }, "steam": { "score": 0.65, "reason": "Across normal frames steam emerges, grows and drifts left/up continuously (absent at 1.0s, small puffs at 2.6s, larger diffuse clouds by 6-8s), while the reduced-motion frames correctly show none. The wisps are soft and translucent but read as somewhat blobby cotton-like puffs rather than fine irregular plumes." }, "noodles": { "score": 0.7, "reason": "Continuous curved strands with plausible thickness and wavy bends; the chopstick-lifted bundle is convincing with parallel hanging strands and slight wave. In-bowl noodles form a believable tangle, though they are a uniform flat color with little volume or strand separation, reading slightly flat in places." }, "composition": { "score": 0.85, "reason": "Well-balanced centered dish against a dark warm-glow background, wooden table with subtle plank lines and a soft contact shadow; the lifted-noodle action with chopsticks gives a clear focal point and no clutter or UI distracts. Layout scales cleanly to mobile." }, "broth": { "score": 0.7, "reason": "Warm orange liquid surface with darker edge band and subtle sheen patches; noodles and toppings rest on the surface and the lifted bundle dips believably into the mass. Somewhat flat gradient with limited depth cues, but no solid or murky appearance." }, "bowl": { "score": 0.8, "reason": "Coherent footed bowl: elliptical opening with dark rim band, smooth gradient body with a left-side highlight, flared foot and grounded shadow with consistent perspective. Minor shortcoming: the foot/waist join is slightly simplified and the rim band is uniform." } } ], "diagnostics": { "vegan": 0.7 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8555555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 49.899999999999636, "frameIntervalP95Ms": 50, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "383c753eeac20c6b240493bc7fe623af3a234ce3661706e2c9d57e62c094aec6", "gradedAt": "2026-10-12T20:06:26.520851+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5611111111111111, "toppings": 0.5622222222222222, "bowl": 0.7288888888888889, "broth": 0.5755555555555556, "steam": 0.5888888888888889, "composition": 0.7922222222222222 }, "reward": 0.6236666666666667, "runId": "meta/muse-wonder-1.0/max", "rewardStdDev": 0.08273602601043879, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.48666666666666664, "bowl": 0.7466666666666667, "broth": 0.4766666666666666, "steam": 0.5166666666666667, "composition": 0.82 }, "reward": 0.5781666666666667, "sampleStdDev": 0.001755942292142151, "judgments": [ { "noodles": { "score": 0.5, "reason": "Lifted strands are continuous and gently curved, but the soup is dominated by thin, nearly straight crisscrossed lines with little supple volume or convincing overlap." }, "toppings": { "score": 0.48, "reason": "Mushroom caps have some gill detail and scallion rings are recognizable. However, leaves are flat oval symbols, nori is rigid rectangular panels, and pale central pieces lack readable texture." }, "bowl": { "score": 0.75, "reason": "The elliptical rim, curved walls and attached foot form coherent geometry with warm shading. The rim is heavily outlined and the angular foot and oversized dark shadow weaken the ceramic and contact impression." }, "broth": { "score": 0.48, "reason": "Warm amber coloring and small highlights suggest soup, but the surface is largely a flat opaque fill, with ingredients appearing layered over it rather than convincingly immersed." }, "steam": { "score": 0.52, "reason": "Normal frames show changing translucent puffs above the soup without obscuring food. These remain separated blurry blobs rather than connected, irregular rising wisps." }, "composition": { "score": 0.82, "reason": "The centered dish, diagonal chopsticks, restrained amber-green palette and uncluttered dark setting create a clear focal hierarchy. Mobile framing preserves the dish but leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible mushrooms, green vegetables, scallion rings, seaweed and noodles plausibly form vegan ramen; no clearly animal-derived topping is visible." } }, { "noodles": { "score": 0.5, "reason": "The lifted bundle has continuous, gently bending strands, but the soup is dominated by thin crisscrossing lines with little convincing overlap or noodle volume." }, "toppings": { "score": 0.5, "reason": "Mushroom caps with fine gill marks and scallion rings are recognizable, but leaves are flat ovals, nori is rectangular and rigid, and pale blocks have little food texture." }, "bowl": { "score": 0.73, "reason": "The elliptical rim, rounded walls and attached foot form coherent geometry with warm shading. The angular foot and nearly black contact shadow weaken the ceramic material and support." }, "broth": { "score": 0.47, "reason": "The warm amber surface has subtle shading and scattered glints, but reads mostly as a flat opaque fill; ingredients lack convincing immersion and liquid contact." }, "steam": { "score": 0.51, "reason": "Translucent marks change position across normal frames and leave the food visible, but appear as disconnected blurred puffs rather than connected, spreading wisps." }, "composition": { "score": 0.82, "reason": "A centered bowl, diagonal chopsticks and restrained amber-green palette establish a clear focal dish against a dark, uncluttered background. No material desktop layout defect is observed; mobile leaves considerable empty space." }, "vegan": { "score": 1, "reason": "Visible mushrooms, greens, scallion rings, seaweed, pale tofu-like pieces and noodles plausibly form vegan ramen; no visibly animal-derived topping is apparent." } }, { "noodles": { "score": 0.5, "reason": "Lifted strands have continuous gentle bends, but the soup is dominated by thin, nearly straight crisscrossing lines with little supple volume or coherent overlap." }, "toppings": { "score": 0.48, "reason": "Mushroom caps and gills, scallion rings, greens and pale cubes are recognizable, but leaves are featureless ovals and nori reads as rigid rectangular cards; most ingredients remain flat symbols." }, "bowl": { "score": 0.76, "reason": "Elliptical rim, curved shaded walls and attached foot establish coherent depth and perspective. The angular foot and very dark contact shadow are less refined, and the material lacks distinct ceramic highlights." }, "broth": { "score": 0.48, "reason": "Warm amber color and scattered glossy droplets suggest soup, but the surface looks mostly like a flat filled ellipse, with little liquid sheen or convincing ingredient immersion." }, "steam": { "score": 0.52, "reason": "Translucent patches change positions across normal frames and extend upward without obscuring the food. They resemble disconnected blurry particles more than irregular, connected steam wisps." }, "composition": { "score": 0.82, "reason": "A single centered bowl, diagonal chopsticks and restrained warm-on-dark palette create a clear, uncluttered focal arrangement without text or UI. Mobile framing leaves considerable empty space and makes the dish relatively small." }, "vegan": { "score": 1, "reason": "Visible mushroom forms, greens, scallions, seaweed, pale tofu-like pieces and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.5333333333333333, "toppings": 0.5, "bowl": 0.64, "broth": 0.5333333333333333, "steam": 0.6, "composition": 0.7066666666666667 }, "reward": 0.5736666666666667, "sampleStdDev": 0.01450287327853813, "judgments": [ { "noodles": { "score": 0.55, "reason": "Fine, thin strands read as noodles, and the chopstick lift is a nice touch. The strands in the broth are thin, flat, scribbly lines, and the lifted bundle looks like rigid parallel wires." }, "toppings": { "score": 0.5, "reason": "Nori, shiitake caps, tofu cubes and bok choy are recognizable. They are small and simple, and the scallion rings look like uniform green circles. The nori is a dark rectangle block, so the toppings are only partly convincing." }, "bowl": { "score": 0.62, "reason": "The ellipse rim, tapered body and foot read coherently, with a warm gradient and a ground shadow. The base looks a bit blocky, the shadow is detached, and the material highlights are subdued." }, "broth": { "score": 0.55, "reason": "The orange broth has a warm gradient and small oil droplets. It looks flat and opaque, and the toppings sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent wisps drift and change between frames, and they do not hide the food. The steam is made of blobby, grainy dots that float well above the bowl and are disconnected from the surface." }, "composition": { "score": 0.7, "reason": "The bowl is a centered focal point on a dark, atmospheric background with a warm glow, and there is no clutter. The bowl is somewhat small, and the large empty dark space around it is unused." }, "vegan": { "score": 0.88, "reason": "Tofu, mushrooms, nori, bok choy and noodles all look plant-based. There are no visible animal ingredients, though the red marks on the tofu are slightly ambiguous." } }, { "noodles": { "score": 0.55, "reason": "Fine, thin strands, with a lifted bundle on chopsticks that reads well. The noodles in the broth are flat, scribbly lines with little volume, and the lifted strands look rigid and straight." }, "toppings": { "score": 0.5, "reason": "Nori rectangles, shiitake caps, tofu cubes and bok choy are recognizable. They are small and flat, and the scallion rings are uniform hollow circles that look like symbols. The nori is blocky and the tofu has odd red marks." }, "bowl": { "score": 0.68, "reason": "The rim ellipse, curved walls, foot and cast shadow are coherent, and the lighting is warm. The base looks slightly faceted and the foot is dark and awkward, with a large shadow disc beneath." }, "broth": { "score": 0.55, "reason": "The orange broth is a warm, smooth surface with chili oil droplets. It is fairly flat and gradient-like, with little depth, and the ingredients sit on top rather than being immersed." }, "steam": { "score": 0.6, "reason": "Soft translucent blurry puffs rise and drift differently between frames, so the motion is real. They look like dotted bokeh particles rather than wisps, and they float somewhat disconnected from the bowl." }, "composition": { "score": 0.72, "reason": "The bowl is centered in a dark, atmospheric scene with a warm glow and a clear focal point. The chopsticks add a diagonal. The framing is balanced, with no clutter or text." }, "vegan": { "score": 0.85, "reason": "The toppings are tofu, mushrooms, nori, greens and scallions, with no visible animal products. The red marks on the tofu are slightly ambiguous." } }, { "noodles": { "score": 0.5, "reason": "Thin pale strands read as noodles, with a lifted bundle on chopsticks. The strands in the bowl are flat, thin scribble-like lines and look wiry and sparse, and the lifted bundle is rigid and straight." }, "toppings": { "score": 0.5, "reason": "Nori squares, shiitake caps, tofu cubes, bok choy leaves and scallion rings are all recognizable. They are small and flat, and the nori is a rigid dark rectangle. The scallion rings are uniform green circles that look like generic symbols." }, "bowl": { "score": 0.62, "reason": "The rim ellipse, curved body, foot and shadow are coherent, and the rim has a highlight. The body is very dark and matte with little ceramic detail. The foot looks narrow and a little blocky, and the shadow is a flat ellipse." }, "broth": { "score": 0.5, "reason": "The orange broth is warm and has a gradient and some oil droplets. It looks fairly flat and opaque, and the ingredients sit on top of it with little immersion or ripple." }, "steam": { "score": 0.6, "reason": "Soft translucent puffs rise above the bowl and change shape between frames. They look like blurry dots or blobs rather than wisps, and they float somewhat disconnected from the soup. The reduced-motion frame shows only a faint haze." }, "composition": { "score": 0.7, "reason": "The bowl is centered as a clear focal point on a dark, atmospheric background with a warm glow. The colors are cohesive and there is no clutter or text. The lower half of the frame is empty, and the mobile view is small." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, nori, greens and scallions all appear plant-based. There is no visible meat or egg. The red marks on the tofu are slightly ambiguous." } } ], "diagnostics": { "vegan": 0.8766666666666667 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7, "bowl": 0.8, "broth": 0.7166666666666667, "steam": 0.65, "composition": 0.85 }, "reward": 0.7191666666666667, "sampleStdDev": 0.012829003598617168, "judgments": [ { "vegan": { "score": 0.5, "reason": "Mushrooms, bok choy, nori, corn, scallion rings and broth all read plant-based, but the small white cubes with a red wavy mark resemble naruto fish cake; appearance is ambiguous between seasoned tofu and fish cake, so an intermediate score applies." }, "toppings": { "score": 0.7, "reason": "Good variety: shiitake slices with gill detailing, bok choy leaves with highlights, a torn-edge nori sheet, corn and chili-oil droplets. Scallion rings are uniformly perfect hollow circles scattered like confetti, reading slightly sticker-like; tofu/naruto cubes are minimal." }, "steam": { "score": 0.55, "reason": "Steam is clearly animated across the normal frames (dispersed puffs at 1s, denser cloud at 2.8-4.6s, changed plume at 7-9.5s), so not static. However it is built from soft blurred round blobs; in frames 1 and pointer-left it reads as disconnected floating dots rather than continuous rising wisps." }, "noodles": { "score": 0.7, "reason": "The lifted bundle from the chopsticks is the strongest element: thin continuous strands fanning down into the broth with believable convergence. The in-bowl tangle is readable but the strands are fairly straight and wiry with limited supple bends and little volume." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric scene: bowl slightly left of center with chopsticks as a diagonal counterpoint, warm rim glow, clear focal dish, no text or clutter, and consistent lighting across desktop and mobile crops. Steam blobbiness is the only minor distraction." }, "broth": { "score": 0.8, "reason": "Warm orange broth with a soft radial sheen, floating oil droplets and a darker edge near the rim. Ingredients show small contact shadows and lifted noodles emerge plausibly from the surface; some scallion rings look pasted on top rather than sitting in the liquid." }, "bowl": { "score": 0.8, "reason": "Coherent stylized bowl: elliptical rim with a dark band, smooth tapered body with a ceramic gradient, distinct foot and a grounded contact shadow. Perspective of rim, body and base agree; only minor flatness in the rim/inner wall treatment." } }, { "noodles": { "score": 0.55, "reason": "Noodles are thin, uniform cream strands with gentle curves and a plausible lifted bundle from the chopsticks, but they read as flat wire-like threads with little volume or thickness variation, forming a loose tangled mat on the surface." }, "toppings": { "score": 0.7, "reason": "Good variety and readable identities: shiitake caps with gill lines, a dark nori sheet, green vegetable shapes, scallion rings, corn dots and chili-oil beads. Scallion rings are somewhat uniform graphic donuts and some pieces sit flatly on the surface, keeping it below polished." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl: open elliptical rim with a dark inner band, smooth tapering body, visible pedestal foot and a soft contact shadow. Simple flat styling with only minor stiffness at the foot join; no broken geometry or incompatible perspective." }, "broth": { "score": 0.7, "reason": "Appetizing golden-orange surface with a warm radial glow, scattered oil droplets and white specks; ingredients mostly appear to sit in the liquid. The surface is somewhat flat and glossy-gradient-like, lacking texture and depth cues." }, "steam": { "score": 0.7, "reason": "Steam clearly changes across the normal frames (compact cluster, then a taller plume, then scattered puffs), so it is animated, translucent and does not hide the food. It is built from soft blurred puffs that read slightly blobby rather than continuous wisps." }, "composition": { "score": 0.85, "reason": "Balanced centered dish with a strong diagonal chopstick accent lifting noodles, warm rim glow against a dark atmospheric background, no text or clutter, and the food stays the clear focal point. Whole scene shifts subtly between frames, which is coherent rather than distracting." }, "vegan": { "score": 0.5, "reason": "Most toppings are plant-appropriate (mushrooms, nori, greens, scallion, corn, chili oil), but the white slices with red curved markings strongly resemble narutomaki fish cake; appearance is ambiguous between fish cake and seasoned tofu, so an intermediate score applies." } }, { "noodles": { "score": 0.7, "reason": "Thin cream strands with gentle waves overlap plausibly in the broth, and the lifted bundle converges convincingly at the chopstick tips. However, in-bowl noodles read as flat squiggles scattered on the surface with a few broken or floating strands, lacking volume." }, "toppings": { "score": 0.7, "reason": "Good variety and readable details: shiitake slices with radial gills, a speckled nori sheet, white tofu blocks with pink curls, edamame pods with seeds, and scattered scallion rings. Defects: scallion rings are uniform graphic donuts, tofu/naruto blocks are plain flat squares, and nori is a stiff rectangle \u2014 somewhat generic shapes." }, "bowl": { "score": 0.8, "reason": "Coherent dark ceramic bowl with an even rim ellipse, smooth body gradient, small foot, and a soft contact shadow beneath. The elliptical rim with side-view body is a tasteful stylization; no broken joins or distortion observed, only a slightly thick flat rim ring." }, "broth": { "score": 0.65, "reason": "Warm orange surface with soft gradient and scattered chili-oil droplets reads appetizing in a stylized way, but the liquid is essentially flat: no sheen texture, no reflections, and toppings/noodles appear pasted onto the surface rather than immersed with occlusion." }, "steam": { "score": 0.7, "reason": "Normal frames show the translucent gray wisps changing configuration and rising above the bowl across time, fading without hiding the food. However, steam is built from many small disconnected round puffs rather than continuous wisps, appearing somewhat blobby." }, "composition": { "score": 0.85, "reason": "Well-balanced centered dish against a dark warm-glow backdrop, attractive amber/green palette, dynamic chopstick diagonal, clean shadow anchoring, no text or clutter; mobile crop also frames the bowl nicely. Food remains the clear focal point." }, "vegan": { "score": 0.0, "reason": "White slices with a distinctive pink spiral curl are clearly narutomaki (fish cake), an unmistakable animal-derived topping, alongside otherwise plant-based mushrooms, nori, edamame, tofu and scallions." } } ], "diagnostics": { "vegan": 0.3333333333333333 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.7366666666666667, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 66.69999999999891, "frameIntervalP95Ms": 66.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5b8c72d10631fc8f860f88e90a5997160f3b57a4c1436a2a0203f50028264548", "gradedAt": "2026-10-12T20:07:06.717822+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.35000000000000003, "toppings": 0.35333333333333333, "bowl": 0.3466666666666667, "broth": 0.32666666666666666, "steam": 0.26666666666666666, "composition": 0.47444444444444445 }, "reward": 0.3477777777777778, "runId": "meta/muse-wonder-1.0/low", "rewardStdDev": 0.30119342868243576, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.5, "toppings": 0.56, "bowl": 0.54, "broth": 0.48, "steam": 0.3, "composition": 0.7833333333333333 }, "reward": 0.5193333333333334, "sampleStdDev": 0.010408329997330672, "judgments": [ { "noodles": { "score": 0.5, "reason": "Thin yellow strands have smooth bends and a readable lifted bundle, but repeated flat loops dominate the surface with little coherent overlap or noodle volume." }, "toppings": { "score": 0.56, "reason": "Mushroom caps and stems, leafy greens, tofu pieces and scallion rings are recognizable and varied. Texture is sparse, however, and the nori resembles thick rounded cards rather than irregular thin seaweed." }, "bowl": { "score": 0.5, "reason": "The elliptical rim and shaded left wall establish depth, but the right wall disappears into an abrupt dark wedge and the lower decorative lines extend beyond the body, producing an incoherent join." }, "broth": { "score": 0.48, "reason": "The warm amber surface is readable and contained by the rim, but looks mostly solid and flat; ingredients appear layered on top with little immersion, contact shading or liquid sheen." }, "steam": { "score": 0.3, "reason": "Diffuse translucent patches change position across normal frames without obscuring the food, but resemble disconnected background haze rather than wisps visibly emerging from the soup and rising." }, "composition": { "score": 0.78, "reason": "A single colorful dish remains the clear focus against a restrained dark setting, with diagonal chopsticks balancing the presentation and no distracting text. Mobile framing leaves substantial empty space and crops the chopstick ends." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, scallions and seaweed alongside noodles; no visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin curved strands and the chopstick-held bundle are readable, but repeated flat loops dominate, with little convincing overlap, depth, or gathered volume." }, "toppings": { "score": 0.58, "reason": "Mushroom caps and stems, tofu pieces, leafy greens, and scallion rings provide recognizable variety. Shapes remain schematic and minimally textured; the nori especially resembles thick rounded cards." }, "bowl": { "score": 0.5, "reason": "The elliptical rim clearly encloses a cavity, with warm highlights and a shaded foot. However, the right wall largely disappears into the background while lower decorative arcs extend beyond the visible body, making construction look disconnected." }, "broth": { "score": 0.48, "reason": "The warm amber surface is readable and contrasts with the garnishes, but looks mostly like a flat opaque fill. Ingredients appear layered on top rather than convincingly immersed, with little liquid sheen or contact detail." }, "steam": { "score": 0.3, "reason": "Soft translucent patches change position across normal frames and leave the food visible. They read more as scattered background haze than irregular wisps visibly emerging from the soup; continuous rising and fading is not established." }, "composition": { "score": 0.78, "reason": "A single prominent dish, restrained dark setting, warm palette, and diagonal chopsticks create a clear focal arrangement without text or UI. The mobile view retains the dish but clips the chopsticks and leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, scallions, seaweed, and noodles plausibly form vegan ramen. No visibly animal-derived ingredient is observed." } }, { "noodles": { "score": 0.5, "reason": "Thin yellow strands have smooth bends and a readable lifted bundle, but repeated flat loops dominate the soup, with little convincing overlap or noodle volume." }, "toppings": { "score": 0.54, "reason": "Mushroom caps and stems, leafy greens, tofu pieces and scallion rings are recognizable. Their detailing is mostly symbolic, and the nori resembles thick, smooth rounded cards rather than irregular thin seaweed." }, "bowl": { "score": 0.62, "reason": "The elliptical rim and front wall establish a readable cavity and depth. However, the right wall disappears into an abrupt dark shape, while lower decorative curves extend beyond the visible body, weakening construction and material coherence." }, "broth": { "score": 0.48, "reason": "The warm amber surface is clearly readable, but largely resembles an opaque flat fill. Ingredients appear layered on top with limited immersion, liquid contact or surface sheen." }, "steam": { "score": 0.3, "reason": "Diffuse translucent patches change position across normal frames without obscuring the dish. They mostly resemble disconnected background haze, rather than irregular wisps visibly emerging from the soup and fading upward." }, "composition": { "score": 0.79, "reason": "The warm focal bowl contrasts effectively with the restrained dark setting, and diagonal chopsticks provide balance without text or UI clutter. Mobile framing leaves substantial empty space and clips the chopsticks." }, "vegan": { "score": 1, "reason": "Visible toppings plausibly depict tofu, mushrooms, greens, scallions and seaweed alongside noodles. No visibly animal-derived ingredient is observed." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.55, "toppings": 0.5, "bowl": 0.5, "broth": 0.5, "steam": 0.5, "composition": 0.64 }, "reward": 0.524, "sampleStdDev": 0.001732050807568879, "judgments": [ { "noodles": { "score": 0.55, "reason": "Thin golden strands curve and overlap with some organic bends, but they read as scribbled thin loops on a flat surface with little volume, and few strands read as truly continuous." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, shiitake caps, bok choy leaves, carrot slices, scallion rings and a nori sheet are all recognizable. The nori is two thick rigid rectangles standing like cards, the tofu is blocky, and the red blobs and scallion rings are generic symbols." }, "bowl": { "score": 0.5, "reason": "The bowl has a clear rim, a lacquer-like gradient and a highlight line, but the lower body has an odd dark wedge on the right and the base looks disjointed and floating over the shadow. The perspective is inconsistent between rim and body." }, "broth": { "score": 0.5, "reason": "The surface is a warm orange glow that reads as soup with some sheen, but it is fairly flat and opaque. Toppings look pasted on top rather than immersed, with little occlusion at the waterline." }, "steam": { "score": 0.5, "reason": "Soft translucent smoky wisps drift up and to the left and change between frames, so there is real motion. They look like blurry blobs that are faint and sit offset from the bowl, and they are not clearly connected to the soup." }, "composition": { "score": 0.62, "reason": "The bowl is a clear focal point in a dark, moody scene with warm colors and a tidy chopstick accent. It sits somewhat large in the frame, and the floating nori and the odd base shadow weaken the presentation." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, bok choy, carrot, scallion, nori and noodles all look plant-based. The red blobs are ambiguous but read as chili." } }, { "noodles": { "score": 0.55, "reason": "Thin yellow strands curve and overlap with a golden sheen and look supple. They read as scribbly loops on a flat surface, with little volume or clear immersion in the broth." }, "toppings": { "score": 0.5, "reason": "Mushrooms, tofu cubes, bok choy leaves, carrot slices and scallion rings are all recognizable and varied. They are small and flat, the tofu looks like blocks, and the nori is a thick dark rectangle standing upright like a card. The red chili blobs are generic." }, "bowl": { "score": 0.5, "reason": "The rim is an ellipse with a believable lip and a warm lacquer gradient. The lower body has an odd dark wedge on the right and a base that looks detached, floating above a faint shadow. The perspective between rim and body is slightly inconsistent." }, "broth": { "score": 0.5, "reason": "The orange surface has a warm glow and depth. It is fairly flat and opaque, and ingredients sit on top rather than being immersed, with little occlusion at the edges." }, "steam": { "score": 0.5, "reason": "Soft, translucent smoky blobs drift upward and shift between frames. They are blurry and sit off to the left, and they read more like fog patches than wisps. They do not hide the food." }, "composition": { "score": 0.65, "reason": "The bowl is a clear central focal point in a dark, moody scene with warm colors and a chopstick accent. The nori overlaps awkwardly at the top left, and the large empty table area below weakens the framing." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, greens, carrot, nori and noodles all read as plant-based, and no animal ingredients are visible. The pink-red bits are ambiguous but read as chili." } }, { "noodles": { "score": 0.55, "reason": "Thin yellow strands curve and overlap with some organic bends. They read as scribbled loops lying flat on the surface, with little volume or depth and no sense of sitting in the broth." }, "toppings": { "score": 0.5, "reason": "Tofu cubes, mushrooms, bok choy, carrot slices, scallion rings and nori are all recognizable and varied. They look like flat symbols, and the nori is two rigid dark rectangles standing up like cards, which limits the score." }, "bowl": { "score": 0.5, "reason": "The bowl has a clear rim, a glossy red lacquer look and a base ring. The dark wedge shape beneath the body is odd, and the bowl appears to float above the table with a mismatched shadow, so the geometry and contact are flawed." }, "broth": { "score": 0.5, "reason": "The orange broth glows warmly and has a subtle sheen. It is flat and mostly hidden behind the noodle pattern, and the toppings look pasted on top rather than immersed." }, "steam": { "score": 0.5, "reason": "Soft translucent smoke drifts and changes between frames, so it clearly moves. It rises mostly from the left and above the nori rather than from the soup, and it reads as blurry blobs." }, "composition": { "score": 0.65, "reason": "The bowl is centered on a dark, moody background with warm colors and a clear focal point. The nori and chopsticks add some balance, and there is no clutter, but the floating bowl and the empty lower third are weaknesses." }, "vegan": { "score": 0.9, "reason": "The bowl shows tofu, mushrooms, greens, carrot, scallion and nori, with no visible animal ingredients. The red bits are ambiguous, but they read as chili or tomato." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0, "toppings": 0, "bowl": 0, "broth": 0, "steam": 0, "composition": 0 }, "reward": 0.0, "sampleStdDev": 0.0, "judgments": [ { "noodles": { "score": 0, "reason": "Placeholder" }, "toppings": { "score": 0, "reason": "Placeholder" }, "bowl": { "score": 0, "reason": "Placeholder" }, "broth": { "score": 0, "reason": "Placeholder" }, "steam": { "score": 0, "reason": "Placeholder" }, "composition": { "score": 0, "reason": "Placeholder" }, "vegan": { "score": 0, "reason": "Placeholder" } }, { "noodles": { "score": 0, "reason": "placeholder" }, "toppings": { "score": 0, "reason": "placeholder" }, "bowl": { "score": 0, "reason": "placeholder" }, "broth": { "score": 0, "reason": "placeholder" }, "steam": { "score": 0, "reason": "placeholder" }, "composition": { "score": 0, "reason": "placeholder" }, "vegan": { "score": 0, "reason": "placeholder" } }, { "noodles": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "toppings": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "bowl": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "broth": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "steam": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "composition": { "score": 0, "reason": "Placeholder \u2014 see final answer" }, "vegan": { "score": 0, "reason": "Placeholder \u2014 see final answer" } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.6333333333333333, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 83.29999999999927, "frameIntervalP95Ms": 83.39999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "14632858ff30eb15ce36140a27ebe5c6a3a3cacb2fcf44e432e0dbc5125857b1", "gradedAt": "2026-10-15T23:52:44.496677+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.3155555555555556, "toppings": 0.5744444444444444, "bowl": 0.46777777777777774, "broth": 0.5377777777777778, "steam": 0.39111111111111113, "composition": 0.68 }, "reward": 0.4807222222222222, "runId": "google/gemini-4-argon/low", "rewardStdDev": 0.13745366222570884, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.37333333333333335, "toppings": 0.5066666666666667, "bowl": 0.37, "broth": 0.4633333333333333, "steam": 0.24, "composition": 0.7066666666666667 }, "reward": 0.42833333333333334, "sampleStdDev": 0.00929157324317761, "judgments": [ { "noodles": { "score": 0.38, "reason": "A few pale curved strands are visible beneath the toppings, but they form faint parallel loops with little readable overlap or noodle volume." }, "toppings": { "score": 0.52, "reason": "Scored mushroom caps, crosshatched tofu and green vegetables are recognizable. However, the nori resembles rigid rectangular cards, and the leaves and tofu have simplified, hard-edged shapes." }, "bowl": { "score": 0.38, "reason": "Speckled ceramic shading and a grounded base provide material detail, but the tall rear opening meets a straight horizontal front cutoff, creating conspicuously incompatible rim and body geometry." }, "broth": { "score": 0.46, "reason": "The warm orange surface is readable as soup, but its broad uniform gradient lacks liquid sheen and depth; toppings have little convincing immersion or surface contact." }, "steam": { "score": 0.24, "reason": "A translucent haze rises above the soup without fully hiding it, but its broad column shapes remain essentially unchanged across normal frames; shifting specks do not establish rising, fading wisps." }, "composition": { "score": 0.72, "reason": "Warm food colors, dark surroundings and restrained utensils create a clear central focal dish without text or clutter. The oversized front wall dominates the presentation, and the mobile dish is rather small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible mushrooms, tofu-like grilled pieces, greens, seaweed and noodles plausibly form vegan ramen. The spoon contains orange liquid with small garnishes, with no clearly animal-derived topping observed." } }, { "noodles": { "score": 0.38, "reason": "A few thin golden curves are visible beneath the toppings, but they form faint, nearly parallel loops rather than a readable, volumetric tangle of supple strands." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, grilled tofu and greens are recognizable, but the mushrooms resemble smooth disks, the greens lack leaf detail, and the nori resembles rigid rectangular cards." }, "bowl": { "score": 0.38, "reason": "Speckled shading and a grounded base suggest ceramic, but the tall rear opening meets a conspicuous straight horizontal front edge, creating incompatible cavity and body perspective." }, "broth": { "score": 0.46, "reason": "Warm orange shading and scattered garnish suggest soup, but the surface is largely a smooth solid-looking fill with little liquid sheen or convincing immersion around toppings." }, "steam": { "score": 0.24, "reason": "Translucent haze rises above the dish, but its broad column-like silhouette stays essentially unchanged across normal frames; shifting specks do not establish rising, spreading wisps." }, "composition": { "score": 0.72, "reason": "The centered dish, warm restrained palette and dark tabletop create a clear focal scene without text or clutter. The oversized bowl frontage dominates the food, and the mobile presentation leaves substantial empty space." }, "vegan": { "score": 1, "reason": "Visible mushroom caps, tofu, greens, seaweed and noodles plausibly form vegan ramen. The cream-colored oval is a spoon holding orange liquid and garnish, not an egg; no animal-derived topping is apparent." } }, { "noodles": { "score": 0.36, "reason": "Only a narrow patch of faint, nearly parallel looping strands is visible beneath the toppings; thickness and overlapping noodle volume are poorly resolved." }, "toppings": { "score": 0.5, "reason": "Scored mushroom caps, crosshatched tofu and green vegetables are recognizable, but remain simplified symbols. Nori resembles rigid rectangular cards, and leaves lack organic detail." }, "bowl": { "score": 0.35, "reason": "Speckling and curved highlights suggest ceramic, but the tall rear opening joins a straight horizontal front edge with incompatible perspective, making the bowl look cut apart rather than coherently hollow." }, "broth": { "score": 0.47, "reason": "Warm orange shading suggests soup, but the surface is largely a flat gradient with little liquid sheen or depth. Toppings lack convincing immersion and contact." }, "steam": { "score": 0.24, "reason": "A translucent haze rises above the soup without fully obscuring it, but its broad column shapes appear essentially unchanged across normal frames; drifting specks do not establish evolving steam wisps." }, "composition": { "score": 0.68, "reason": "Warm colors, a dark restrained background and a single centered dish create a clear focal point without text or clutter. The large foreground wall dominates the food, and the mobile dish occupies relatively little of the frame." }, "vegan": { "score": 1, "reason": "Visible mushroom caps, tofu-like grilled pieces, greens, seaweed and noodles plausibly form vegan ramen. The spoon appears to contain orange broth with garnish; no clearly animal-derived ingredient is visible." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.13999999999999999, "toppings": 0.5166666666666667, "bowl": 0.3, "broth": 0.48333333333333334, "steam": 0.4, "composition": 0.5166666666666667 }, "reward": 0.3771666666666667, "sampleStdDev": 0.009169696468985954, "judgments": [ { "noodles": { "score": 0.15, "reason": "Noodles are almost invisible; only faint thin golden lines show near the mushrooms. No readable strands or volume." }, "toppings": { "score": 0.55, "reason": "Shiitake with white scoring, a grid-textured tofu, nori sheets and greens are recognizable and vegan-appropriate. The nori reads as stiff rectangular cards, and the lower halves of the toppings are cut off by the bowl." }, "bowl": { "score": 0.3, "reason": "The bowl is split by a hard horizontal line that cuts through the soup. The rim, front wall and base have incompatible perspective, so the bowl looks broken. The speckled glaze and the base are decent." }, "broth": { "score": 0.5, "reason": "The orange broth has a warm glow and some depth, but it is abruptly truncated by the flat bowl wall. Ingredients are cut off rather than sitting in the liquid." }, "steam": { "score": 0.4, "reason": "A soft translucent haze rises above the bowl and drifts slightly between frames. It is a fairly uniform column with little wisp detail." }, "composition": { "score": 0.5, "reason": "The dark moody scene, centered bowl and warm focal color are good. The hard horizontal cut and the awkward spoon placement hurt the presentation." }, "vegan": { "score": 0.9, "reason": "Mushrooms, tofu, nori, greens and an orange broth are visible. No animal products are seen." } }, { "noodles": { "score": 0.15, "reason": "Noodles are almost invisible: only faint thin golden lines near the mushrooms. The lower half of the bowl is cut off by a hard opaque horizontal band that hides the main noodle mass, so there is no readable volume." }, "toppings": { "score": 0.5, "reason": "Shiitake with split-star cuts, a grid-patterned tofu, nori sheets and bok choy are recognizable and varied. The nori reads as stiff rectangles and the tofu as a waffle-like card. The toppings are cut off by the hard horizontal band across the bowl." }, "bowl": { "score": 0.3, "reason": "A flat horizontal cut slices through the bowl at mid-height, so the front wall looks like an opaque slab detached from the rim. Perspective is incompatible between the top ellipse and the lower body. The speckled glaze and the base ring are decent, but the construction is conspicuously broken." }, "broth": { "score": 0.45, "reason": "The warm orange broth has a pleasant gradient and glow, with small sesame flecks. It is clipped by the hard band, so the liquid depth and immersion contact are unreadable, and the toppings partly look pasted on." }, "steam": { "score": 0.4, "reason": "A soft translucent glow plume rises above the bowl and shifts slightly between frames. It is a diffuse vertical fog column rather than distinct wisps, and the motion is subtle." }, "composition": { "score": 0.5, "reason": "The dark moody scene is centered with a clear focal bowl, and the warm orange against the dark background is attractive. The spoon and chopsticks add context. The sliced bowl band is a glaring flaw in the overall presentation." }, "vegan": { "score": 0.9, "reason": "Mushrooms, tofu, nori, greens and an orange broth are all plant-based, with no visible meat, egg or fish." } }, { "noodles": { "score": 0.12, "reason": "Noodles are almost entirely hidden: only faint thin lines are visible near the mushrooms. Ramen noodles are essentially unreadable." }, "toppings": { "score": 0.5, "reason": "The shiitake with white scoring, grid-textured nori, fried tofu and bok choy are recognizable and varied. But the lower halves are cut off by a flat horizontal slice, and the nori reads as stiff cards." }, "bowl": { "score": 0.3, "reason": "The bowl is cut by a hard horizontal line where the dark front wall starts, and the rim is not coherent with the body. The speckled glaze and the base are decent, but the geometry looks broken and the cavity does not read." }, "broth": { "score": 0.5, "reason": "The warm orange broth has a pleasing gradient and glow. It is cut off flatly by the bowl wall, and the toppings look clipped rather than immersed." }, "steam": { "score": 0.4, "reason": "A soft vertical haze rises above the bowl and shifts slightly between frames. It is a diffuse column with no distinct wisps, and it looks mostly static." }, "composition": { "score": 0.55, "reason": "The dark moody scene is centered, with warm broth as the focal point and a spoon and chopsticks for context. The cropped bowl makes the dish feel unbalanced, and the lower half is a large empty dark area." }, "vegan": { "score": 0.9, "reason": "The toppings are mushrooms, tofu, nori, greens and scallion, with no visible animal products." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.43333333333333335, "toppings": 0.7, "bowl": 0.7333333333333333, "broth": 0.6666666666666666, "steam": 0.5333333333333333, "composition": 0.8166666666666667 }, "reward": 0.6366666666666667, "sampleStdDev": 0.030138568866708567, "judgments": [ { "noodles": { "score": 0.45, "reason": "Only a few sparse wavy yellow strands peek out between the mushrooms and tofu; the large open broth areas show no coherent noodle bed, so the noodle presence reads as isolated squiggles rather than a believable ramen mass." }, "toppings": { "score": 0.7, "reason": "Good variety with readable details: two mushroom slices with star-cut cross-sections, grid-textured nori sheets, lattice-patterned fried tofu, leafy greens, scallion, and scattered sesame seeds. Minor flaws: the mushroom discs are flat and near-identical, and the nori squares read slightly as rigid standing cards." }, "bowl": { "score": 0.75, "reason": "Coherent dark speckled ceramic bowl with rim band, accent line, curved glaze strokes, pedestal foot, mat and shadow; spoon and chopsticks rest plausibly on the rim. The straight horizontal seam where the interior meets the front wall and the slightly blob-like foot are noticeable but read as intentional stylization." }, "broth": { "score": 0.65, "reason": "Warm orange miso-toned surface with restrained sheen, a subtle ripple ellipse around the mushrooms, floating sesame, and proper occlusion of topping bases behind the front rim. The strong central glow reads a bit like backlighting rather than liquid depth, and the surface is otherwise fairly flat." }, "steam": { "score": 0.7, "reason": "Translucent wisps above the bowl change shape across desktop-1/2/3 (curling plume shifts and reshapes, particles drift), rising and spreading without obscuring the food; it is somewhat diffuse and glow-like rather than crisply wisped, hence not higher." }, "composition": { "score": 0.8, "reason": "Well-centered focal bowl in a moody dark scene with warm broth against cool darkness, balanced accents from chopsticks, spoon and drifting specks, no text or clutter; mobile framing leaves generous empty space but keeps the dish readable." }, "vegan": { "score": 1.0, "reason": "Visible toppings are mushrooms, nori, tofu, leafy greens, scallion and sesame in plant broth; no meat, fish, egg or dairy is discernible, and the brown star-cut discs read as mushroom slices rather than naruto." } }, { "noodles": { "score": 0.45, "reason": "Noodles are barely represented: only a few thin wavy yellow strands peek out between the mushrooms and tofu pieces; there is no coherent noodle mass, volume, or believable nesting in the broth, reading as sparse decorative squiggles." }, "toppings": { "score": 0.75, "reason": "Good variety and readability: two brown shiitake caps with white cross-cut scores, stacked nori sheets with a grid texture, crosshatched tofu-like pieces, a scallion segment, leafy greens and scattered sesame specks. Greens look slightly glossy/faceted and the tofu lattice pattern is schematic, but nothing reads as a generic plastic stand-in." }, "bowl": { "score": 0.7, "reason": "Coherent dark lacquer bowl with a rim pinstripe, speckled body, curved lower wall, pedestal foot and woven trivet; the spoon and chopsticks rest plausibly on the rim. Main flaw: the broth is terminated by a perfectly straight horizontal front-wall seam across a round opening, which reads as a clipped cutaway rather than a natural near rim." }, "broth": { "score": 0.7, "reason": "Warm orange broth with a soft backlit gradient, sesame specks on the surface and toppings seated with contact shadows. Appetizing and integrated, but the surface is mostly a smooth airbrushed glow with little ripple or liquid texture, and the bright center haze is somewhat artificial." }, "steam": { "score": 0.25, "reason": "A pale translucent plume rises above the bowl without hiding the food, but its silhouette is effectively identical across desktop frames 1-3 (1077/3428/5628 ms), indicating a static glow rather than wisps that rise, spread and fade." }, "composition": { "score": 0.8, "reason": "Centered focal bowl on a restrained dark backdrop with warm rim light, floating particles and no text, UI or competing elements; mobile keeps the dish legible. The large featureless dark front wall occupies much of the lower frame and slightly deadens the presentation." }, "vegan": { "score": 0.9, "reason": "Visible ingredients are nori, scored mushrooms, leafy greens, scallion, sesame and tofu-like lattice pieces in an orange plant-broth-colored soup; no egg, meat, fish or dairy is depicted. Lattice pieces are tofu-like rather than fish-cake-like, so only mild ambiguity." } }, { "noodles": { "score": 0.4, "reason": "Noodles are barely depicted: only a few thin yellow squiggle strands peek between the mushrooms and tofu at the broth edge; no coherent strand volume or overlap is visible, most of the noodle bed is hidden behind the tall front wall." }, "toppings": { "score": 0.65, "reason": "Good variety: two grid-textured nori sheets, scored shiitake-style caps, crosshatch fried-tofu pieces, green leaves with a scallion stick, sesame seeds and a chili-oil spoon. Shapes are clean but flat and graphic; the greens read slightly glossy/blobby and the nori leans oddly over the rim." }, "bowl": { "score": 0.75, "reason": "Coherent dark ceramic bowl with speckled glaze, ridge lines, red rim accent, footed base on a woven tray with contact shadow; spoon and chopsticks rest plausibly. Minor issues: the broth-to-wall boundary is a nearly straight horizontal edge rather than a rim arc, and the pedestal looks slightly thin." }, "broth": { "score": 0.65, "reason": "Warm amber surface with radial glow and subtle ripple arcs; mushrooms and tofu sit with faint contact shadows. It reads somewhat flat and backlit-glowy rather than liquid, and the bright even gradient limits depth." }, "steam": { "score": 0.65, "reason": "Steam is clearly animated across the normal frames: the plume grows, shifts and splits into two wisps by the pointer-left frame, and stays translucent without hiding the food. However it reads as a soft uniform haze rather than defined irregular wisps." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric scene: centered dish with warm glow, red chopstick accent, scattered sesame and vignette; no clutter or text. Mobile framing leaves the bowl small with large empty margins, and the steam haze occupies much of the upper third." }, "vegan": { "score": 0.85, "reason": "Visible ingredients (nori, mushrooms, leafy greens, scallion, sesame, tofu-like crosshatch pieces, plant-toned broth) plausibly form vegan ramen; the crosshatch pieces are ambiguous but read more as fried tofu than fish cake, with no unmistakable animal ingredients." } } ], "diagnostics": { "vegan": 0.9166666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.9388888888888889, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 133.29999999999927, "frameIntervalP95Ms": 133.39999999999964, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "5996b1bc49c75a52b96bf89fefd239f53f5beac3d36d83dc0652def110dcdd61", "gradedAt": "2026-10-15T23:56:58.642171+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.6666666666666666, "toppings": 0.65, "bowl": 0.7611111111111111, "broth": 0.63, "steam": 0.5988888888888889, "composition": 0.8055555555555556 }, "reward": 0.6814444444444445, "runId": "google/gemini-4-argon/medium", "rewardStdDev": 0.037801503987324424, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.6466666666666667, "toppings": 0.5866666666666667, "bowl": 0.7933333333333333, "broth": 0.5666666666666667, "steam": 0.5133333333333333, "composition": 0.84 }, "reward": 0.6523333333333334, "sampleStdDev": 0.012770408502993666, "judgments": [ { "noodles": { "score": 0.66, "reason": "Lifted strands have continuous bends, warm highlights and readable separation. Those in the soup are flatter, uniformly thick strokes with limited tangled volume and immersion." }, "toppings": { "score": 0.6, "reason": "Scored mushroom caps, porous tofu, scallion rings and veined greens provide recognizable variety. Leaves are overly smooth and paddle-shaped, while nori resembles rigid gridded cards rather than irregular thin seaweed." }, "bowl": { "score": 0.81, "reason": "The elliptical gold-edged rim, shaded red cavity and dark exterior form a coherent supported vessel. Layered highlights suggest lacquer, although the lower body is very dark and shallow, weakening its volume." }, "broth": { "score": 0.58, "reason": "Warm reddish shading and scattered surface flecks suggest seasoned broth. The surface lacks convincing liquid reflections and menisci, leaving several ingredients looking arranged on top rather than immersed." }, "steam": { "score": 0.53, "reason": "Translucent plumes change subtly across normal frames and leave the food visible. Their broad, nearly straight ribbon shapes and abrupt upper haze look more like light shafts than irregular rising and dissipating wisps." }, "composition": { "score": 0.84, "reason": "Warm gold and red food contrasts clearly against the restrained dark background; diagonal chopsticks guide attention toward the lifted noodles, with no distracting text or UI. Mobile framing leaves substantial empty space and makes food details small." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.64, "reason": "Lifted strands have continuous rounded bends and warm highlights, but hang in an overly parallel bundle; noodles in the soup look like flat strokes with limited volumetric overlap." }, "toppings": { "score": 0.58, "reason": "Scored mushroom caps, tofu pores, scallion rings and leaf veins provide readable variety. However, leaves are smooth paddle shapes, nori resembles rigid gridded cards, and several vegetables have a plastic-like finish." }, "bowl": { "score": 0.78, "reason": "The elliptical rim, nested inner wall and dark supported base form a coherent vessel with restrained gold highlights. The exterior is very shallow and dark, limiting the sense of ceramic wall volume." }, "broth": { "score": 0.55, "reason": "Warm reddish shading and small surface highlights suggest soup, but the surface is unusually smooth and solid-looking, with most ingredients appearing placed on top rather than visibly immersed." }, "steam": { "score": 0.48, "reason": "Translucent columns change subtly across the normal frames and leave the food readable. Their broad, straight-sided ribbon shapes and similar spacing feel mechanical rather than irregular rising wisps." }, "composition": { "score": 0.84, "reason": "A clear central dish, diagonal chopsticks, warm gold-red palette and restrained dark background create a balanced focal arrangement without text or clutter. The mobile dish is somewhat small amid extensive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, leafy greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is apparent." } }, { "noodles": { "score": 0.64, "reason": "Lifted strands have continuous curves, warm highlights and readable separation. Strands in the soup are flatter, similarly thick strokes with limited submerged volume and somewhat repetitive bends." }, "toppings": { "score": 0.58, "reason": "Scored mushroom caps, seeded tofu, scallion rings and veined greens provide readable variety. However, greens are overly smooth oval forms, nori resembles rigid gridded cards, and corn and red vegetables look glossy and plastic." }, "bowl": { "score": 0.79, "reason": "The elliptical rim, inner wall, gold edging and shaded exterior form a coherent supported vessel. The shallow exterior and very dark lower body leave its depth and foot less clearly resolved." }, "broth": { "score": 0.57, "reason": "Warm reddish shading and small highlights suggest rich soup, but the surface lacks convincing liquid sheen and depth; many toppings and horizontal noodles appear laid on top rather than immersed." }, "steam": { "score": 0.53, "reason": "Translucent rising bands shift subtly across normal frames and leave the food visible. Their broad, nearly straight ribbon shapes and blunt upper haze feel more like light shafts than irregular spreading, fading vapor." }, "composition": { "score": 0.84, "reason": "The centered warm dish stands clearly against a restrained dark background, with diagonal chopsticks balancing the upright greens and nori. No text or clutter competes with the food; mobile presentation leaves somewhat excessive empty space." }, "vegan": { "score": 1, "reason": "Visible tofu, mushrooms, greens, seaweed, corn and noodles plausibly form vegan ramen; no visibly animal-derived ingredient is present." } } ], "diagnostics": { "vegan": 1 }, "veganFailed": false }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.6866666666666666, "toppings": 0.7133333333333333, "bowl": 0.69, "broth": 0.6066666666666667, "steam": 0.5166666666666667, "composition": 0.76 }, "reward": 0.6678333333333333, "sampleStdDev": 0.021096998206695965, "judgments": [ { "noodles": { "score": 0.7, "reason": "Lifted strands on chopsticks are supple, with a believable drape and curled tops. Submerged noodles twist and overlap with some volume. The lower noodles are somewhat flat and repetitive, and the strands are uniformly thin." }, "toppings": { "score": 0.74, "reason": "Tofu cubes with sesame, shiitake with a cut pattern, a bok choy with veins, corn, bamboo strips, scallion rings and a nori sheet are all readable. The nori is a rigid, grid-patterned card and the bok choy is somewhat blobby. The variety and detail are good." }, "bowl": { "score": 0.72, "reason": "The bowl has a coherent elliptical rim with a gold edge, an inner wall and a dark base with a shadow. The perspective is consistent. The outer body is barely visible and the base is dark and vague, so the lacquer material reads thinly." }, "broth": { "score": 0.62, "reason": "The red-brown broth has depth and sheen, and the ingredients sit in it. The large red glossy blobs look like paint or chili oil patches and are a bit artificial. Concentric ring lines make the surface slightly flat." }, "steam": { "score": 0.55, "reason": "Faint translucent vertical wisps rise above the bowl and shift between frames. They are subtle and streaky, with a somewhat mechanical look. They do not hide the food." }, "composition": { "score": 0.78, "reason": "The bowl is centered as a clear focal point on a dark, restrained background. The warm colors are cohesive, the chopsticks add a diagonal, and the nori balances the bok choy. There is no clutter or text." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, bok choy, corn, bamboo, nori and noodles are all plant-based. No egg, meat or fish is visible." } }, { "noodles": { "score": 0.68, "reason": "Lifted noodle strands on chopsticks look supple and have natural bends and a convincing hanging volume. The noodles in the broth are thin, somewhat flat loops and a braided twist, a little repetitive and sparse. Overall competent but not fully convincing." }, "toppings": { "score": 0.7, "reason": "Variety is good and readable: tofu cubes with sesame, bok choy with veins, shiitake with a scored cap, bamboo shoots, corn kernels, scallion rings and nori. The nori looks like stiff gridded cards floating off the rim. The red blobs are ambiguous. Tofu and corn are a little glossy and generic." }, "bowl": { "score": 0.7, "reason": "The elliptical rim has a gold edge, an inner ring and a dark base with a shadow. The perspective is consistent. The bowl wall and foot are barely visible and dark, so the contact with the surface is vague and the material reads somewhat flat." }, "broth": { "score": 0.6, "reason": "The deep red-brown broth has a gradient and ring highlights, so it reads as liquid. The glossy red blobs look like pasted-on spots, and the surface is somewhat murky and flat. Ingredient immersion is only moderately convincing." }, "steam": { "score": 0.5, "reason": "Faint vertical wisps rise above the bowl and shift between frames, so they are translucent and not overpowering. They are subtle and streaky, with a somewhat mechanical, columnar look, and they barely spread." }, "composition": { "score": 0.75, "reason": "The bowl is centered and is a clear focal point against a dark, restrained background. The warm and green colors balance well. The chopsticks add a diagonal. The nori and chopsticks are cropped slightly awkwardly, and there is no clutter." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, bok choy, corn, bamboo, nori and noodles are all plant-based. No meat or egg is visible. The red blobs are slightly ambiguous but read as chili oil." } }, { "noodles": { "score": 0.68, "reason": "Lifted strands on chopsticks are supple and continuous with nice bends, and the twisted noodles in the broth overlap plausibly. The submerged noodles are somewhat flat and uniform, and the lifted bundle is a bit parallel and rigid." }, "toppings": { "score": 0.7, "reason": "Variety is good: tofu cubes with sesame, bok choy with veins, shiitake with scoring, corn, bamboo shoots, nori and scallion rings. The nori looks like stiff gridded cards, and the red blobs and scallion rings are somewhat symbolic." }, "bowl": { "score": 0.65, "reason": "The elliptical rim has a gold edge and highlights, and the inner wall reads with depth. The base is dark and ambiguous, with a faint offset ring that looks slightly detached, and the material is only moderately convincing." }, "broth": { "score": 0.6, "reason": "The broth is a rich red-brown with a sheen and reflected rings, and the ingredients sit in it. The red chili blobs look pasted on and glossy, and the surface is a bit flat and murky." }, "steam": { "score": 0.5, "reason": "Faint translucent vertical wisps drift and change across frames, and they don't hide the food. They are subtle, streaky and fairly mechanical, with a hazy block at the top." }, "composition": { "score": 0.75, "reason": "The bowl is centered as a clear focal point on a dark, atmospheric background, with a pleasant warm palette and green accents. The chopsticks and nori add dynamism, and there is no clutter or text." }, "vegan": { "score": 0.9, "reason": "Tofu, mushrooms, bok choy, corn, bamboo shoots, nori and scallions are all plant-based, with no visible meat or egg. The red broth could be ambiguous." } } ], "diagnostics": { "vegan": 0.9 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.6666666666666666, "toppings": 0.65, "bowl": 0.8, "broth": 0.7166666666666667, "steam": 0.7666666666666667, "composition": 0.8166666666666667 }, "reward": 0.7241666666666667, "sampleStdDev": 0.03971250852481284, "judgments": [ { "noodles": { "score": 0.6, "reason": "The chopstick-lifted bundle reads well: golden wavy strands with sesame specks and believable bends. But in-broth noodles are sparse, uniformly thin squiggly lines lying flat on the surface, and the lifted strands taper to points rather than merging into the broth \u2014 stylized but somewhat wire-like." }, "toppings": { "score": 0.6, "reason": "Good variety: tofu cubes with speckles, corn kernels, bamboo strips, scallion rings, scored mushroom caps, bok choy half. Defects: two large glossy red plastic-like spheres, rigid grid-patterned rectangles standing in for nori sheets, a flat disc with a star/spiral pattern, and a glossy blob-like dark green leaf." }, "bowl": { "score": 0.8, "reason": "Coherent dark lacquer bowl with a consistent gold-rimmed ellipse, smooth wall gradients with a left highlight, a visible foot ring, and a grounded shadow beneath. Slightly flat/elliptical foot and a faint vertical seam on the left wall, but no broken geometry or perspective clash." }, "broth": { "score": 0.7, "reason": "Deep red broth with darker edge shading, faint sheen curves near the rim and soft shadows under tofu/corn that suggest sitting in liquid. Still fairly flat: ingredients mostly rest on top of the surface rather than appearing submerged, and the surface lacks richer texture/depth." }, "steam": { "score": 0.8, "reason": "Translucent gray wisps rise from the soup and clearly change shape between desktop-1/2/3 and the pointer frames, showing genuine animation; they fade gradually and never obscure the food. Slightly uniform, parallel vertical plumes keep it from scoring higher." }, "composition": { "score": 0.8, "reason": "Centered, balanced focal dish on a warm dark vignette; no text, UI or clutter; chopsticks and nori add asymmetry. Minor issues: a redundant second chopstick pair crossing at top right and mechanically vertical steam bands; the warm glow does not substitute for food detail." }, "vegan": { "score": 0.5, "reason": "Mostly plant-based appearance: tofu, bok choy, corn, bamboo, scallions, mushrooms, nori. However, the dark disc with a white asterisk/spiral pattern strongly resembles a naruto fish cake, which is ambiguous enough to lower compliance; no other animal ingredients are clearly visible." } }, { "noodles": { "score": 0.8, "reason": "Chopstick-lifted bundle shows supple continuous strands with gentle bends and coherent gathering into the broth; in-broth wavy strands with sesame read well in the stylized style, though the hang is fairly parallel/straight and in-broth strands are a bit flat." }, "toppings": { "score": 0.7, "reason": "Good variety: sesame-topped tofu cubes, striated bamboo shoots, cross-hatched mushroom caps, bok choy halves, corn, scallion rings and nori sheets, all readable. Defects: two glossy red orbs read as plastic spheres, and the red scribble around the scallions looks messy." }, "bowl": { "score": 0.8, "reason": "Coherent shallow lacquer bowl with consistent rim ellipse, gold trim, inner ring, highlighted exterior wall and a grounded foot/shadow; chopsticks rest believably on the rim. Minor oddity: nori sheets overlap the rim edge ambiguously at left." }, "broth": { "score": 0.75, "reason": "Dark mahogany liquid with restrained sheen, inner-rim shadow depth, floating oil dots and believable ingredient contact/occlusion; somewhat murky in places, making the surface read less liquid than brighter broths." }, "steam": { "score": 0.8, "reason": "Normal frames at ~1s, ~3.4s, ~5.5s and pointer frames show translucent irregular wisps whose shapes and positions change over time (left-side streaks appear, shift and fade), rising without obscuring the food; reduced-motion frames sensibly omit it." }, "composition": { "score": 0.8, "reason": "Centered focal bowl on a warm dark vignette, balanced diagonal chopsticks, steam and soft glow; no clutter or text. On mobile the bowl is small with large empty dark margins, slightly weakening the presentation." }, "vegan": { "score": 0.9, "reason": "Visible items (tofu, mushrooms, bok choy, corn, scallion, chili, bamboo shoots, nori, noodles) are plausibly plant-based; the dark disc with a light asterisk marking is mildly ambiguous but reads more like a scored mushroom cap than a fish cake." } }, { "noodles": { "score": 0.6, "reason": "Lifted chopstick bundle hangs with mostly straight, parallel, uniformly thick strands that read slightly wire-like, and in-broth noodles are flat repetitive squiggles; still coherent with some curves and believable immersion." }, "toppings": { "score": 0.65, "reason": "Good variety: bok choy, scored mushrooms, tofu cubes, bamboo, corn, scallion/chili rings, sesame, nori. But nori reads as rigid grid-lined panels leaning on the rim and the glossy red sphere is plastic-like; shapes are clean yet simplified." }, "bowl": { "score": 0.8, "reason": "Coherent lacquer bowl with consistent elliptical rim, gold trim, interior depth, streaked body highlights and a grounded base/shadow; perspective is internally consistent, only mildly shallow for ramen." }, "broth": { "score": 0.7, "reason": "Glossy dark red broth with restrained sheen and rim-hugging ripple arcs; toppings sit with contact shadows, though the thin concentric rim lines and flat surface shading look slightly artificial." }, "steam": { "score": 0.7, "reason": "Translucent wisps clearly change shape and position across desktop-1/2/3, so motion is animated, but plumes are fairly uniform vertical bands with mechanical regularity rather than organic curling wisps." }, "composition": { "score": 0.85, "reason": "Polished dark scene: bowl is a clear focal point, chopsticks lead the eye upper-right, warm palette against moody background, no text or clutter; minor awkwardness from the leaning nori and hard-cropped chopstick ends." }, "vegan": { "score": 0.6, "reason": "Tofu, bok choy, mushrooms, bamboo, corn, scallion, chili and nori are plant-based; one dark disc with a white asterisk pattern is ambiguous (could read as a narutomaki-like fishcake), so not full credit." } } ], "diagnostics": { "vegan": 0.6666666666666666 }, "veganFailed": false } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.8555555555555555, "veganFailed": false }, "performance": { "frameIntervalMedianMs": 150.04999999999927, "frameIntervalP95Ms": 166.70000000000073, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } }, { "artifactSha256": "704cd34268c9d801f4ec5811b4172a2c0b7549db8485531ce8a2957875f476b5", "gradedAt": "2026-10-16T00:44:39.739313+00:00", "scores": { "standalone": 1.0, "procedural": 1.0, "noodles": 0.5422222222222223, "toppings": 0.6833333333333333, "bowl": 0.7833333333333333, "broth": 0.6222222222222222, "steam": 0.6411111111111111, "composition": 0.8077777777777778 }, "reward": 0.0, "runId": "google/gemini-4-argon/high", "rewardStdDev": 0.045049766719501255, "judgeResults": [ { "model": "openrouter/openai/gpt-6.1-sol", "effort": "low", "scores": { "noodles": 0.49333333333333335, "toppings": 0.6233333333333333, "bowl": 0.86, "broth": 0.5666666666666667, "steam": 0.5833333333333333, "composition": 0.8466666666666667 }, "reward": 0.6553333333333333, "sampleStdDev": 0.005008326400438942, "judgments": [ { "noodles": { "score": 0.48, "reason": "Visible golden strands overlap beneath the toppings, but mostly form broad, flat parallel ribbons with angular bends rather than supple, rounded noodles." }, "toppings": { "score": 0.64, "reason": "Scallion rings, corn, a scored mushroom and veined greens provide readable variety. Repeated crosshatched slices look stamped, nori resembles rigid rectangular cards, and many small garnishes are overly spherical and glossy." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, decorated inner wall, ribbed dark exterior and grounded foot create coherent depth and convincing lacquer-like highlights. No material construction defect is observed." }, "broth": { "score": 0.57, "reason": "Warm orange shading and scattered oil-like highlights suggest rich broth, but the surface is rather solid and matte; several ingredients and droplets appear suspended above it rather than immersed." }, "steam": { "score": 0.6, "reason": "Normal frames show changing translucent plumes rising above the soup while leaving the food readable. The numerous broad, sharply bounded ribbon shapes feel repetitive and sheet-like rather than softly dispersing wisps." }, "composition": { "score": 0.86, "reason": "Warm food colors stand out clearly against the dark wooden setting, with a balanced spoon-and-chopstick arrangement and no distracting text or UI. Mobile framing leaves somewhat excessive empty space above the dish." }, "vegan": { "score": 0, "reason": "Two unmistakable egg halves with white surrounds and orange yolks are visible among the toppings." } }, { "noodles": { "score": 0.5, "reason": "Visible golden strands have continuous bends and some overlap, but mostly read as flat, parallel ribbons beneath the toppings rather than a supple, voluminous noodle nest." }, "toppings": { "score": 0.62, "reason": "Scallion rings, corn, scored mushroom cap and veined greens are readable and varied. However, nori resembles rigid rectangular cards, the crosshatched slices are repetitive, and scattered glossy spheres look artificial." }, "bowl": { "score": 0.86, "reason": "A coherent elliptical rim, patterned inner wall, shaded ribbed exterior and grounded foot establish convincing depth and support. Layered edge highlights communicate thickness; no material construction defect is observed." }, "broth": { "score": 0.57, "reason": "Warm orange shading and small highlights suggest rich soup, but the surface is unusually uniform and solid-looking. Strong topping shadows and raised droplets make much of the garnish appear suspended above it rather than immersed." }, "steam": { "score": 0.57, "reason": "Normal frames show changing translucent plumes emerging above the soup. Their broad, repeated leaf-shaped ribbons and accumulated haze look more mechanical than irregular rising wisps." }, "composition": { "score": 0.84, "reason": "The warm focal bowl contrasts clearly with the dark wooden setting, with chopsticks and spoon providing balanced diagonals and no distracting UI or text. Mobile framing leaves considerable unused space above the dish." }, "vegan": { "score": 0, "reason": "Two unmistakable egg halves with orange yolks and white surrounds are visible among the toppings." } }, { "noodles": { "score": 0.5, "reason": "Visible golden strands have consistent thickness and some curved overlaps, but are mostly obscured and read as a flattened bundle of broad, stiff ribbons rather than supple noodles." }, "toppings": { "score": 0.61, "reason": "Scallion rings, corn, veined greens and a scored mushroom are readable and varied. However, nori resembles rigid rectangular cards, crosshatched slices are repetitive, and numerous glossy orange beads look artificial." }, "bowl": { "score": 0.86, "reason": "The continuous elliptical rim, patterned inner wall, shaded ribbed exterior and grounded foot form a coherent vessel with convincing depth. Minor refinement could improve the subdued ceramic highlights." }, "broth": { "score": 0.56, "reason": "Warm orange shading and scattered oil-like highlights suggest rich broth, but the surface is unusually flat and opaque, with several ingredients and beads appearing suspended above it rather than immersed." }, "steam": { "score": 0.58, "reason": "Normal frames show changing translucent plumes rising above the soup. Their broad, repeated leaf-shaped bands and dense overlap resemble layered gauze more than irregular dissipating wisps." }, "composition": { "score": 0.84, "reason": "Warm food colors stand clearly against a restrained dark wood setting, with chopsticks and spoon framing a single focal bowl and no distracting UI. Mobile preserves the arrangement but leaves substantial empty space above." }, "vegan": { "score": 0, "reason": "Two unmistakable egg halves with white surrounds and orange yolks are visible among the toppings." } } ], "diagnostics": { "vegan": 0 }, "veganFailed": true }, { "model": "openrouter/anthropic/claude-sonnet-5.5", "effort": "low", "scores": { "noodles": 0.48333333333333334, "toppings": 0.6933333333333334, "bowl": 0.7066666666666667, "broth": 0.65, "steam": 0.6066666666666667, "composition": 0.76 }, "reward": 0.6433333333333333, "sampleStdDev": 0.013012814197295434, "judgments": [ { "noodles": { "score": 0.45, "reason": "Noodles are mostly hidden under toppings. The visible strands near the center are small, thin and tangled, so the noodle mass reads as indistinct and lacks volume." }, "toppings": { "score": 0.68, "reason": "Toppings are varied and well drawn: a shiitake with a star pattern, bok choy with white stems, corn, menma, scallions, nori sheets and chili threads. The grilled slices look meat-like. Some pieces are slightly flat and illustrative, but there are no plastic blobs." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a patterned red band, a foot ring and a plate. The perspective is consistent and there are rim highlights. The base is dark and somewhat murky, and the bowl's support contact is a little ambiguous." }, "broth": { "score": 0.65, "reason": "The orange chili broth has a glossy gradient and floating oil droplets, and toppings sit in it with shadows. It is a bit flat and uniformly saturated, and the red droplets look like beads." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise and change shape across the frames. The plumes are somewhat repetitive and mechanical, with overlapping vertical leaf shapes, but they do not hide the food." }, "composition": { "score": 0.75, "reason": "The bowl is a central focal point in a dark, moody scene with warm lighting. The chopsticks and plate are balanced and there is no clutter. The bowl's lower half is slightly cramped near the frame bottom." }, "vegan": { "score": 0.2, "reason": "Halved soft-boiled eggs with orange yolks are clearly visible, and the grilled slices look meat-like. This reads as non-vegan." } }, { "noodles": { "score": 0.5, "reason": "Noodles are mostly hidden under toppings; the small visible patch has thin wavy strands but is indistinct and limited in volume." }, "toppings": { "score": 0.68, "reason": "Shiitake with a star pattern, bok choy with translucent stems, corn, nori sheets and scallions are readable and varied. The grilled slices look like meat and the menma is crude." }, "bowl": { "score": 0.72, "reason": "Coherent wide rim with a patterned red band, a foot, a saucer and a believable perspective. The thickness and highlights are decent. The base is a bit dark and the contact is vague." }, "broth": { "score": 0.62, "reason": "Warm orange broth with chili oil droplets and a gradient has depth. It looks somewhat flat and opaque, and ingredient immersion is limited." }, "steam": { "score": 0.62, "reason": "Translucent wisps rise and change between frames. They look like repeated overlapping lens shapes, and they are somewhat mechanical and hazy over the toppings." }, "composition": { "score": 0.75, "reason": "The bowl is a clear focal point in a dark, warm scene. The chopsticks and the spoon are balanced, and the colors are appetizing. The steam slightly obscures the bowl's left side." }, "vegan": { "score": 0.1, "reason": "Soft-boiled egg halves are clearly visible, and the grilled slices look like meat, so the bowl does not look vegan." } }, { "noodles": { "score": 0.5, "reason": "Noodles are mostly hidden in a small central cluster under toppings. Thin wavy strands are visible but flat and indistinct, with little volume and no strands lifting from the broth." }, "toppings": { "score": 0.72, "reason": "Variety is good: a shiitake with a star pattern, nori sheets, bok choy with white stems, corn, scallion rings, menma and chili threads. Textures are fairly polished. The grilled-looking slices are ambiguous and resemble meat, and the nori reads as stiff cards." }, "bowl": { "score": 0.7, "reason": "The bowl has a coherent elliptical rim, a patterned red inner band, a stepped base and a plate with a reflection. Thickness and ceramic highlights are convincing. The rim is a little heavy and the base slightly narrow, but there is no major distortion." }, "broth": { "score": 0.68, "reason": "Orange broth has a gradient, droplets of chili oil and some depth. The surface is slightly flat and opaque, and the toppings sit on it with soft contact shadows. It is appetizing but somewhat uniform." }, "steam": { "score": 0.6, "reason": "Translucent wisps rise and shift between frames. The plumes are repetitive, overlapping and mechanical, and a bit broad, partly veiling the bowl rim. Motion is evident but not very organic." }, "composition": { "score": 0.78, "reason": "The bowl is a clear focal point in a dark, moody scene. The warm orange palette is balanced against a dark background, and the chopsticks and spoon add interest. The bowl is slightly low and off-center, and the steam fills the top." }, "vegan": { "score": 0.2, "reason": "Halved soft-boiled eggs with orange yolks are clearly visible, and the grilled slices look like meat. Both make the bowl non-vegan in appearance." } } ], "diagnostics": { "vegan": 0.16666666666666669 }, "veganFailed": false }, { "model": "openrouter/z-ai/glm-5.3-flash", "effort": "default", "scores": { "noodles": 0.65, "toppings": 0.7333333333333333, "bowl": 0.7833333333333333, "broth": 0.65, "steam": 0.7333333333333333, "composition": 0.8166666666666667 }, "reward": 0.7266666666666667, "sampleStdDev": 0.04502314219746698, "judgments": [ { "noodles": { "score": 0.65, "reason": "Yellow strands are continuous with plausible waviness and coherent layering under the narutomaki and eggs, but the noodle mass is sparse and reads as flat painted ribbons spread on the surface rather than a supple, voluminous mound." }, "toppings": { "score": 0.7, "reason": "Good variety and mostly readable: halved eggs with orange yolks, a star-patterned narutomaki, bamboo shoots, corn, bok choy, scallion rings, chili threads, sesame and standing nori sheets with texture. Defects: the three grilled slices are repetitive identical flat discs with stamped crosshatch marks, and the eggs look slightly glossy/plastic." }, "bowl": { "score": 0.8, "reason": "Coherent high-angle bowl: dark exterior, decorative red inner band, elliptical rim with a left-edge highlight, offset broth ellipse consistent with the viewpoint, and a centered footed base resting on a woven mat. Only minor shortfall is the slightly narrow foot and soft wall definition at the front." }, "broth": { "score": 0.6, "reason": "The orange-red surface has gradient tone, darker smudges, glossy chili-oil beads and scattered sesame, but reads fairly matte and solid with little liquid sheen or depth, and toppings show weak occlusion/immersion shadows." }, "steam": { "score": 0.8, "reason": "Translucent irregular wisps rise from the soup and clearly change shape and position across desktop-1/2/3 and pointer frames, showing real motion without hiding the food. Slight mechanical quality from several similar parallel plumes forming a curtain." }, "composition": { "score": 0.8, "reason": "Balanced warm scene: bowl as clear focal point on a placemat with chopsticks, steam filling the upper frame, subtle wood-grain background and restrained bokeh, no text or clutter; mobile crop keeps the dish readable. Background glow does not substitute for the slightly flat broth but presentation is cohesive." }, "vegan": { "score": 0.0, "reason": "Clearly animal-derived toppings are visible: two halved boiled eggs with yolks and a narutomaki fish cake (the grilled disc slices also read as meat-based chashu), so compliance fails regardless of quality." } }, { "noodles": { "score": 0.6, "reason": "Pale wavy strands are visible around the eggs and center, with plausible curvature, but they are thin, sparse and largely buried under toppings; much of the noodle body reads as indistinct pale mass rather than coherent strands." }, "toppings": { "score": 0.65, "reason": "Good variety: crosshatched chashu-like slices, a scored dark cap, bamboo shoots, greens, nori sheets, scallion rings, corn and chili threads. Defects: the egg halves are glossy plastic-like domes, the chashu discs read as flat cookies, and scattered red oil beads look like polished spheres." }, "bowl": { "score": 0.75, "reason": "Coherent wide bowl with a patterned red rim band, visible interior wall, dark foot resting on a saucer with shadow; ellipse perspective is consistent at the high viewing angle. Minor flatness in the foot/saucer join." }, "broth": { "score": 0.65, "reason": "Orange broth with edge shading, oil droplets and a mild sheen; eggs and greens show contact shadows. Surface reads slightly flat and toppings sit somewhat pasted-on in places, but immersion is broadly believable." }, "steam": { "score": 0.7, "reason": "Translucent gray wisps rise from the soup and clearly change shape across desktop-1/2/3 and pointer frames, fading without hiding the food; the plumes are somewhat parallel and mechanically repeated in form." }, "composition": { "score": 0.8, "reason": "Centered bowl as clear focal point on a warm dark atmospheric ground, chopsticks and spoon anchored below, no clutter or UI. Mobile frame leaves generous empty headroom but keeps the dish readable." }, "vegan": { "score": 0, "reason": "Two unmistakable soft-boiled egg halves with orange yolks sit in the bowl, an animal-derived topping, so vegan compliance fails." } }, { "noodles": { "score": 0.7, "reason": "Pale yellow strands are visible between the eggs and greens with gentle bends and coherent flow, but they are partially buried, read a bit flat/ribbon-like, and the noodle mass is sparse relative to the bowl." }, "toppings": { "score": 0.85, "reason": "Rich, readable variety: grill-marked chashu-style slices, shiitake with star gill pattern, leafy greens with vein detail, corn, scallion rings, chili threads, bamboo shoots, patterned nori sheets and halved eggs with glossy yolks; only minor weakness is the scattered glossy red beads reading slightly like plastic droplets." }, "bowl": { "score": 0.8, "reason": "Wide donburi with coherent elliptical rim, decorative red inner band, dark footed pedestal and consistent high-angle perspective; the resting ladle contacts the rim plausibly, with only minor stiffness at the rim-to-foot transition." }, "broth": { "score": 0.7, "reason": "Appetizing orange-red surface with gradient depth, sheen and scattered chili-oil droplets, and toppings cast contact shadows into the liquid; however the surface is fairly flat and uniform, with items reading more pasted-on than immersed." }, "steam": { "score": 0.7, "reason": "Translucent wisps clearly change shape and position across the time-ordered desktop frames, rising and fading without hiding food, but the many near-identical parallel plume shapes look somewhat mechanical and repetitive." }, "composition": { "score": 0.85, "reason": "Balanced dark atmospheric scene with warm rim light, centered focal dish, chopsticks and ladle arranged cleanly, no clutter or text; mobile framing leaves a large empty dark region above the bowl, a minor imbalance." }, "vegan": { "score": 0, "reason": "Two unmistakable soft-boiled egg halves with round yolks are prominent toppings, clearly animal-derived, so vegan compliance fails despite otherwise plant-forward ingredients." } } ], "diagnostics": { "vegan": 0.0 }, "veganFailed": true } ], "standaloneViolations": [], "complianceViolations": [], "diagnostics": { "vegan": 0.05555555555555556, "veganFailed": true }, "performance": { "frameIntervalMedianMs": 116.79999999999927, "frameIntervalP95Ms": 133.3000000000011, "desktopHorizontalOverflow": false, "mobileHorizontalOverflow": false, "reducedHorizontalOverflow": false } } ] }