{ "format": "proofrun-publication", "schemaVersion": 5, "contentHash": "sha256:a18cc0cd707db557113d4d181a8a036b9a7e3ad956909b06a742c3f606a2afb8", "preparedAt": "2026-08-20T15:47:59.338Z", "editorial": { "format": "results-brief", "slug": "glm-5-3-vs-gemini-3-7-flash-orbit-radio", "headline": "glm-5.3 vs gemini-3.7-flash: Orbit Radio", "dek": "A tactile sci-fi tuner that tests hierarchy, motion, and input handling.", "status": "draft", "publishedAt": null }, "experiment": { "protocol": "proofrun-v0.7", "mode": "duel", "runOutcome": "head_to_head", "runId": "PR-MT1JXMC5-7F", "startedAt": "2026-08-20T13:24:05.687Z", "identityState": "revealed", "evaluationState": "locked", "legacyProtocolException": false, "blindOrder": { "scheme": "sha256-128bit-assignment-commit-reveal-v1", "commitment": "sha256:2022d278fff081f1d7916fb2354c2b42e69f4d6ebadfaeb78ae13ec9c5ed9a30", "secret": "90109A5FC5670AF1C45994EBA4270B3C", "verified": true } }, "test": { "id": "orbit-radio", "title": "Orbit Radio", "category": "Interface design", "summary": "A tactile sci-fi tuner that tests hierarchy, motion, and input handling.", "prompt": "Create a one-screen browser instrument called ORBIT RADIO. A large central dial tunes between five fictional deep-space signals. Turning or dragging the dial must change the station name, frequency, color, and an animated signal visualization. Add compact previous/next controls and make the current station unmistakable. Use one self-contained HTML file with vanilla CSS and JavaScript, no external assets or libraries, and make it responsive.", "origin": "curated", "version": 3, "hash": "curated:orbit-radio:v3", "scope": "quick", "runtime": "7-12 min", "executionMode": "web-single-file-v2", "signals": [ "Hierarchy", "Input quality", "Motion", "State design" ], "integrity": "versioned" }, "settings": { "reasoningEffort": "low", "maxTokens": 32000, "temperature": 0.7, "htmlStartDeadlineMinutes": 20, "writingGraceMinutes": 10, "maxCostMicrousdPerAttempt": 5000000 }, "verdict": { "winner": "A", "winnerModel": "z-ai/glm-5.3", "confidence": "low", "rationale": "This one was very close but Contestant A takes the win here because of the slightly nicer UI, despite the drag and scroll text being slightly behind the previous button Contestant As central dial is smooth and has no issues while dragging it while Contestant Bs central dial seems to jump around. Contestant As visualizations are also better in my opinion.", "scores": { "A": 46, "B": 34 }, "evaluationVersion": 3, "scoreConstruct": "craft-score", "compliance": { "A": 100, "B": 100 }, "identityGuess": { "recordedAt": "2026-08-20T15:22:20.469Z", "mapping": { "A": "z-ai/glm-5.3", "B": "google/gemini-3.7-flash" }, "confidence": "medium", "abstained": false } }, "requirements": [ { "id": "requirement-1", "label": "The central dial responds to turning or dragging.", "A": "pass", "B": "pass" }, { "id": "requirement-2", "label": "Five distinct fictional stations are available.", "A": "pass", "B": "pass" }, { "id": "requirement-3", "label": "Station name, frequency, color, and visualization change together.", "A": "pass", "B": "pass" }, { "id": "requirement-4", "label": "Previous and next controls work consistently.", "A": "pass", "B": "pass" }, { "id": "requirement-5", "label": "The current station is visually unmistakable.", "A": "pass", "B": "pass" }, { "id": "requirement-6", "label": "The single-file instrument remains responsive and works offline.", "A": "pass", "B": "pass" } ], "criteria": [ { "id": "visual-design", "label": "Visual design", "description": "Hierarchy, composition, typography, color, polish, and coherent visual judgment.", "weight": 30, "anchors": { "low": "Confusing or unfinished", "mid": "Competent and readable", "high": "Distinctive and exceptional" }, "A": 5, "B": 4 }, { "id": "interaction", "label": "Interaction & usability", "description": "Discoverability, responsiveness, feedback, control quality, and interaction feel.", "weight": 30, "anchors": { "low": "Frustrating or broken", "mid": "Usable with rough edges", "high": "Immediate and delightful" }, "A": 5, "B": 4 }, { "id": "technical-robustness", "label": "Technical robustness", "description": "Runtime correctness, edge-case handling, performance, responsiveness, and code reliability.", "weight": 30, "anchors": { "low": "Repeatable failures", "mid": "Mostly sound with risks", "high": "Stable under scrutiny" }, "A": 5, "B": 4 }, { "id": "originality", "label": "Originality", "description": "Useful creative choices that go beyond a generic first solution without harming clarity.", "weight": 10, "anchors": { "low": "Generic or derivative", "mid": "Some authored character", "high": "Memorable and purposeful" }, "A": 6, "B": 5 } ], "contestants": [ { "side": "A", "model": "z-ai/glm-5.3", "resolvedModel": "z-ai/glm-5.3", "provider": "Z.AI", "status": "complete", "artifactOutcome": "completed_submission", "selectedAttemptOrdinal": 1, "routing": { "mode": "auto", "endpointSlug": null, "allowFallbacks": true, "resolvedProvider": "Z.AI", "resolvedEndpoint": null }, "screenshot": { "path": "screenshots/contestant-a-standardized-signal-step-demonstration-3447b790.png", "capturedAt": "2026-08-20T13:25:37.861Z", "width": 1280, "height": 720, "byteLength": 429123, "method": "browser-paint-evidence-recipe-v1", "rendererVersion": null, "cameraProvenance": null, "camera": null, "recipeReceipt": { "schemaVersion": 1, "id": "ER-mt1jzl1j", "recipeId": "orbit-radio-standard-demonstration", "recipeVersion": 1, "recipeHash": "sha256:f81eabe774ab8258d9ff403badfe2440ba2ae6a5b18f9bd88d6039831ac3660f", "testId": "orbit-radio", "side": "A", "status": "succeeded", "startedAt": "2026-08-20T13:25:35.306Z", "completedAt": "2026-08-20T13:25:37.873Z", "viewport": { "width": 1280, "height": 720, "deviceScale": 1 }, "resetBeforeStart": true, "actionTranscript": [ { "id": "reset-before-start", "label": "Reset before start", "type": "reset", "status": "skipped", "detail": "The recipe does not require a reset control." }, { "id": "signal-next-01", "label": "Tune to next signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT control." }, { "id": "signal-next-02", "label": "Tune to following signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT control." }, { "id": "signal-previous-01", "label": "Return one signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching PREV control." } ], "stabilization": { "rule": "settled-window", "waitMs": 700, "animationFrames": 0, "status": "passed", "detail": "Waited 700ms and 0 animation frames before assertion and capture." }, "assertions": [ { "id": "next-present", "label": "Next control detected", "required": true, "status": "passed", "detail": "A matching NEXT control was detected." }, { "id": "previous-present", "label": "Previous control detected", "required": true, "status": "passed", "detail": "A matching PREV control was detected." }, { "id": "state-changed", "label": "Signal demonstration differs from the initial state", "required": true, "status": "passed", "detail": "The post-recipe visual fingerprint differs from the reset state." }, { "id": "runtime-clean", "label": "No new runtime errors during the recipe", "required": true, "status": "passed", "detail": "No new runtime errors were captured during the recipe." } ], "capture": { "state": "demonstration", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "publicationPrimaryEligible": true, "method": "browser-paint-evidence-recipe-v1", "capturedAt": "2026-08-20T13:25:37.861Z" }, "failure": null } }, "scene": null, "artifact": { "format": "proofrun-web-artifact", "schemaVersion": 1, "side": "A", "path": "artifacts/contestant-a.html", "contentHash": "sha256:774cd412774abaeb3e6bd2b351c84ddb5d4a0e74877e07f93439758aabb7efae", "byteLength": 15527, "mediaType": "text/html;charset=utf-8", "executionMode": "web-single-file-v2", "sandboxProfile": "proofrun-public-web-v1" }, "evidenceImages": [ { "id": "EV-d51f6a0a-ef8b-40ce-ba2d-c6d93447b790", "path": "screenshots/contestant-a-standardized-signal-step-demonstration-3447b790.png", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "provenance": "harness-recipe", "state": "demonstration", "primary": true, "capturedAt": "2026-08-20T13:25:37.861Z", "width": 1280, "height": 720, "byteLength": 429123, "method": "browser-paint-evidence-recipe-v1", "rendererVersion": null, "cameraProvenance": null, "camera": null, "recipeReceipt": { "schemaVersion": 1, "id": "ER-mt1jzl1j", "recipeId": "orbit-radio-standard-demonstration", "recipeVersion": 1, "recipeHash": "sha256:f81eabe774ab8258d9ff403badfe2440ba2ae6a5b18f9bd88d6039831ac3660f", "testId": "orbit-radio", "side": "A", "status": "succeeded", "startedAt": "2026-08-20T13:25:35.306Z", "completedAt": "2026-08-20T13:25:37.873Z", "viewport": { "width": 1280, "height": 720, "deviceScale": 1 }, "resetBeforeStart": true, "actionTranscript": [ { "id": "reset-before-start", "label": "Reset before start", "type": "reset", "status": "skipped", "detail": "The recipe does not require a reset control." }, { "id": "signal-next-01", "label": "Tune to next signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT control." }, { "id": "signal-next-02", "label": "Tune to following signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT control." }, { "id": "signal-previous-01", "label": "Return one signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching PREV control." } ], "stabilization": { "rule": "settled-window", "waitMs": 700, "animationFrames": 0, "status": "passed", "detail": "Waited 700ms and 0 animation frames before assertion and capture." }, "assertions": [ { "id": "next-present", "label": "Next control detected", "required": true, "status": "passed", "detail": "A matching NEXT control was detected." }, { "id": "previous-present", "label": "Previous control detected", "required": true, "status": "passed", "detail": "A matching PREV control was detected." }, { "id": "state-changed", "label": "Signal demonstration differs from the initial state", "required": true, "status": "passed", "detail": "The post-recipe visual fingerprint differs from the reset state." }, { "id": "runtime-clean", "label": "No new runtime errors during the recipe", "required": true, "status": "passed", "detail": "No new runtime errors were captured during the recipe." } ], "capture": { "state": "demonstration", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "publicationPrimaryEligible": true, "method": "browser-paint-evidence-recipe-v1", "capturedAt": "2026-08-20T13:25:37.861Z" }, "failure": null } } ], "generation": { "startedAt": "2026-08-20T13:24:05.783Z", "finishedAt": "2026-08-20T13:25:34.478Z", "wallTimeMs": 88695, "firstActivityMs": 4099, "htmlStartMs": 5519, "lastActivityMs": 88694, "activityChunks": 4956, "finishReason": "stop", "reasoningEffort": "low" }, "usage": { "inputTokens": 179, "visibleTokens": 4905, "reasoningTokens": 80, "cachedTokens": 0, "totalTokens": 5164, "cost": 0.0221846, "costSource": "openrouter-reported" }, "evidence": { "artifactIssues": [], "runtimeErrors": [], "previewAudit": { "textCharacters": 260, "visibleElements": 84, "bodyChildren": 2, "documentCharacters": 241691 }, "probe": { "status": "passed", "durationMs": 543, "viewport": { "width": 536, "height": 520 }, "actions": [ { "id": "resize", "label": "Narrow viewport", "status": "passed", "detail": "538 x 520 viewport exercised" }, { "id": "click", "label": "Single click", "status": "passed", "detail": "Dispatched one complete pointer and mouse click sequence" }, { "id": "drag", "label": "Short drag", "status": "passed", "detail": "Dispatched a bounded six-step diagonal drag" }, { "id": "burst", "label": "Five-point burst", "status": "passed", "detail": "Dispatched five bounded clicks across the interaction surface" }, { "id": "reset", "label": "Reset control", "status": "skipped", "detail": "No reset, clear, or restart control was detected" } ], "issues": [], "errors": [] } } }, { "side": "B", "model": "google/gemini-3.7-flash", "resolvedModel": "google/gemini-3.7-flash", "provider": "Google AI Studio", "status": "complete", "artifactOutcome": "completed_submission", "selectedAttemptOrdinal": 1, "routing": { "mode": "auto", "endpointSlug": null, "allowFallbacks": true, "resolvedProvider": "Google AI Studio", "resolvedEndpoint": null }, "screenshot": { "path": "screenshots/contestant-b-standardized-signal-step-demonstration-d4b605f1.png", "capturedAt": "2026-08-20T13:25:40.990Z", "width": 1280, "height": 720, "byteLength": 427997, "method": "browser-paint-evidence-recipe-v1", "rendererVersion": null, "cameraProvenance": null, "camera": null, "recipeReceipt": { "schemaVersion": 1, "id": "ER-mt1jznlf", "recipeId": "orbit-radio-standard-demonstration", "recipeVersion": 1, "recipeHash": "sha256:f81eabe774ab8258d9ff403badfe2440ba2ae6a5b18f9bd88d6039831ac3660f", "testId": "orbit-radio", "side": "B", "status": "succeeded", "startedAt": "2026-08-20T13:25:38.529Z", "completedAt": "2026-08-20T13:25:41.000Z", "viewport": { "width": 1280, "height": 720, "deviceScale": 1 }, "resetBeforeStart": true, "actionTranscript": [ { "id": "reset-before-start", "label": "Reset before start", "type": "reset", "status": "skipped", "detail": "The recipe does not require a reset control." }, { "id": "signal-next-01", "label": "Tune to next signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT ▶ control." }, { "id": "signal-next-02", "label": "Tune to following signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT ▶ control." }, { "id": "signal-previous-01", "label": "Return one signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching ◀ PREV control." } ], "stabilization": { "rule": "settled-window", "waitMs": 700, "animationFrames": 0, "status": "passed", "detail": "Waited 700ms and 0 animation frames before assertion and capture." }, "assertions": [ { "id": "next-present", "label": "Next control detected", "required": true, "status": "passed", "detail": "A matching NEXT ▶ control was detected." }, { "id": "previous-present", "label": "Previous control detected", "required": true, "status": "passed", "detail": "A matching ◀ PREV control was detected." }, { "id": "state-changed", "label": "Signal demonstration differs from the initial state", "required": true, "status": "passed", "detail": "The post-recipe visual fingerprint differs from the reset state." }, { "id": "runtime-clean", "label": "No new runtime errors during the recipe", "required": true, "status": "passed", "detail": "No new runtime errors were captured during the recipe." } ], "capture": { "state": "demonstration", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "publicationPrimaryEligible": true, "method": "browser-paint-evidence-recipe-v1", "capturedAt": "2026-08-20T13:25:40.990Z" }, "failure": null } }, "scene": null, "artifact": { "format": "proofrun-web-artifact", "schemaVersion": 1, "side": "B", "path": "artifacts/contestant-b.html", "contentHash": "sha256:3b0cc808fe699a561c6d3fc86095f4e2c49a4b05fd262dd372e58f2a681d3480", "byteLength": 19273, "mediaType": "text/html;charset=utf-8", "executionMode": "web-single-file-v2", "sandboxProfile": "proofrun-public-web-v1" }, "evidenceImages": [ { "id": "EV-65a41603-80f8-49b4-a75b-835ad4b605f1", "path": "screenshots/contestant-b-standardized-signal-step-demonstration-d4b605f1.png", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "provenance": "harness-recipe", "state": "demonstration", "primary": true, "capturedAt": "2026-08-20T13:25:40.990Z", "width": 1280, "height": 720, "byteLength": 427997, "method": "browser-paint-evidence-recipe-v1", "rendererVersion": null, "cameraProvenance": null, "camera": null, "recipeReceipt": { "schemaVersion": 1, "id": "ER-mt1jznlf", "recipeId": "orbit-radio-standard-demonstration", "recipeVersion": 1, "recipeHash": "sha256:f81eabe774ab8258d9ff403badfe2440ba2ae6a5b18f9bd88d6039831ac3660f", "testId": "orbit-radio", "side": "B", "status": "succeeded", "startedAt": "2026-08-20T13:25:38.529Z", "completedAt": "2026-08-20T13:25:41.000Z", "viewport": { "width": 1280, "height": 720, "deviceScale": 1 }, "resetBeforeStart": true, "actionTranscript": [ { "id": "reset-before-start", "label": "Reset before start", "type": "reset", "status": "skipped", "detail": "The recipe does not require a reset control." }, { "id": "signal-next-01", "label": "Tune to next signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT ▶ control." }, { "id": "signal-next-02", "label": "Tune to following signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching NEXT ▶ control." }, { "id": "signal-previous-01", "label": "Return one signal", "type": "activate-control", "status": "passed", "detail": "Activated the matching ◀ PREV control." } ], "stabilization": { "rule": "settled-window", "waitMs": 700, "animationFrames": 0, "status": "passed", "detail": "Waited 700ms and 0 animation frames before assertion and capture." }, "assertions": [ { "id": "next-present", "label": "Next control detected", "required": true, "status": "passed", "detail": "A matching NEXT ▶ control was detected." }, { "id": "previous-present", "label": "Previous control detected", "required": true, "status": "passed", "detail": "A matching ◀ PREV control was detected." }, { "id": "state-changed", "label": "Signal demonstration differs from the initial state", "required": true, "status": "passed", "detail": "The post-recipe visual fingerprint differs from the reset state." }, { "id": "runtime-clean", "label": "No new runtime errors during the recipe", "required": true, "status": "passed", "detail": "No new runtime errors were captured during the recipe." } ], "capture": { "state": "demonstration", "label": "Standardized signal-step demonstration", "caption": "Next was activated twice and Previous once from a fresh Orbit Radio load, then the resulting signal state was captured after the standardized stabilization window.", "publicationPrimaryEligible": true, "method": "browser-paint-evidence-recipe-v1", "capturedAt": "2026-08-20T13:25:40.990Z" }, "failure": null } } ], "generation": { "startedAt": "2026-08-20T13:24:05.783Z", "finishedAt": "2026-08-20T13:24:21.716Z", "wallTimeMs": 15933, "firstActivityMs": 1596, "htmlStartMs": 1596, "lastActivityMs": 15849, "activityChunks": 237, "finishReason": "stop", "reasoningEffort": "low" }, "usage": { "inputTokens": 174, "visibleTokens": 6804, "reasoningTokens": 0, "cachedTokens": 0, "totalTokens": 6978, "cost": 0.0256455, "costSource": "openrouter-reported" }, "evidence": { "artifactIssues": [], "runtimeErrors": [], "previewAudit": { "textCharacters": 232, "visibleElements": 60, "bodyChildren": 2, "documentCharacters": 247579 }, "probe": { "status": "passed", "durationMs": 535, "viewport": { "width": 536, "height": 520 }, "actions": [ { "id": "resize", "label": "Narrow viewport", "status": "passed", "detail": "538 x 520 viewport exercised" }, { "id": "click", "label": "Single click", "status": "passed", "detail": "Dispatched one complete pointer and mouse click sequence" }, { "id": "drag", "label": "Short drag", "status": "passed", "detail": "Dispatched a bounded six-step diagonal drag" }, { "id": "burst", "label": "Five-point burst", "status": "passed", "detail": "Dispatched five bounded clicks across the interaction surface" }, { "id": "reset", "label": "Reset control", "status": "skipped", "detail": "No reset, clear, or restart control was detected" } ], "issues": [], "errors": [] } } } ], "attempts": [ { "side": "A", "ordinal": 1, "retryOfOrdinal": null, "lifecycle": "terminal", "artifactOutcome": "completed_submission", "terminalCause": "normal", "requestedModel": "z-ai/glm-5.3", "routingMode": "auto", "requestedEndpoint": null, "resolvedProvider": "Z.AI", "resolvedEndpoint": null, "resolvedModel": "z-ai/glm-5.3", "failure": null, "startedAt": "2026-08-20T13:24:05.783Z", "finishedAt": "2026-08-20T13:25:34.478Z" }, { "side": "B", "ordinal": 1, "retryOfOrdinal": null, "lifecycle": "terminal", "artifactOutcome": "completed_submission", "terminalCause": "normal", "requestedModel": "google/gemini-3.7-flash", "routingMode": "auto", "requestedEndpoint": null, "resolvedProvider": "Google AI Studio", "resolvedEndpoint": null, "resolvedModel": "google/gemini-3.7-flash", "failure": null, "startedAt": "2026-08-20T13:24:05.783Z", "finishedAt": "2026-08-20T13:24:21.716Z" } ], "protocolExceptions": [], "reviewAmendments": [], "disclosure": { "rawModelResponsesIncluded": false, "contestantSourceIncluded": true, "privateReviewerNotesIncluded": false, "credentialsIncluded": false, "generatedHtmlExecutedByPublicRecord": false, "generatedHtmlExecutedByJournal": false, "generatedHtmlAvailableInIsolatedSandbox": true, "rawSceneScriptIncluded": false, "compiledVoxelSceneIncluded": false, "editorialAssistance": { "evidenceAnalysis": null, "articleDrafting": null } } }