{ "name": "tank-game", "type": "tank-game", "archiveDate": "2026-11-10", "displayDate": "2026-11-10", "runDate": "2026-11-10", "model": "Six models · high reasoning", "appName": "tank game", "description": "Browser tank game benchmark: playable apps, mobile controls, destruction, execution time, token estimates, and checks.", "capturedAt": "2026-11-10T05:15:02.863Z", "notes": "One independent task per model, with fresh Boat workspaces and identical task and platform instructions. Model effort names do not imply equal compute budgets. Timings exclude setup, independent review, and publication; token-cost estimates exclude VM, proxy, tool, subscription, and tax charges. Apps retain the generated gameplay code. Publication adds the model name to each browser title. Browser checks use headless Chrome and SwiftShader software graphics on a shared CPU host. Host load was not controlled; frame-rate samples are not comparable across apps and do not establish real-device FPS. This single comparison does not establish a quality winner. The efficiency instruction arrived after launch. Each agent received the same follow-up in its original session. Execution time and cost include both turns; total time also includes the wait between turns.", "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "results": [ { "key": "openai-gpt-6.1-sol-codex", "harness": "codex", "harnessName": "Codex", "provider": "openai", "providerName": "OpenAI", "model": "gpt-6.1-sol", "name": "GPT-6.1 Sol", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 2028.157, "totalSeconds": 2043.234, "tokens": { "inputTokens": 187186, "outputTokens": 81311, "cacheReadTokens": 9258788, "cacheWriteTokens": 0 }, "cost": { "pricedSubtotalUsd": "2.113360800000000000000000000000", "completeTotalUsd": "2.113360800000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: Final software-rendered stress measurements are 11.9 FPS desktop and 27.3 FPS mobile emulation; smooth 30/60 FPS is not achieved consistently in this environment. Physical phones and hardware-accelerated GPUs were not tested. Adaptive quality may lower the 3D buffer to DPR .65, shadows to 1024 pixels/10 updates per second, decorative grass density, and HUD blur. Geometry detail and rough-surface shaders are simplified; gameplay damage and fire propagation are preserved.", "files": { "assets/font-0-CCxn3Epz.ttf": "cc9fb2539cf27df48255982e896d28f544483537cdb41f9fabd741af8992d719", "assets/font-1-RZZ47rCX.ttf": "0d85af813fc3ed87db0c6265515689b2eef5cbaf7aab17922528dfc95a2cd73f", "assets/font-10-DmWgiO8-.ttf": "754dfc9d50cf7aabfb6b108d2c2f7d20a3f1f2cc6f6c01640c6728091272cec0", "assets/font-2-CQj3XIHL.ttf": "7dde307fa887fc65ff5830cfada77a7decc5dae8d3c816c9d39ba3f1af1c4ed7", "assets/font-3-BdFBoAAT.ttf": "73bc9e17231e9073780d3c81574c7253cecd5645c93544df1a935054b39a364f", "assets/font-4-CB7WfZ8W.ttf": "b0ae3e89a7d3ef2b1838bb4b1ceb0b17cc3435af061eee809d76aa623ff116e9", "assets/font-5-DCQhiIS7.ttf": "376427b5322148d8de1e03f263bcca4935a2d60ba71d9f94f4947ab664c711ae", "assets/font-6-CYvJJk-g.ttf": "b79a2ea7a38f33fd96a282fadb9ce97626bb0724a44d21075859f0e50239993a", "assets/font-7-BQ0abr9c.ttf": "2a8e7f713ee79fbccdef3cef5c787aff47580299f7da60144961e7db2ec08211", "assets/font-8-ClCTSiDw.ttf": "a7fdc993f9f0387caedba4e42521136fa97e4c0837ae6239bf0413ccf8af00c2", "assets/font-9-DyRt3UMW.ttf": "4fc14a73ca53ba9d32fd759ae1ca1a3133326035d0dd337862b3ee1633cc156e", "assets/index-CH-iKJAj.js": "ae58771d9fce0359d10c9633e0a736e1414ec72efc1d395f6f87d0f898d76cbb", "assets/index-DdBO4Wcj.css": "0bfdaf7562311bb276d98fdbcb3d9e90e166d542dac1d1dbe67118f1afb85a83", "credits.txt": "9efe832bf6aeeed8e568a6304271626d7a48221d4ae5e0e6b91cd38dc945c094", "index.html": "e4a2b6eccd81db52fa1eb017cae57c3781d9db69e3bc551eb804f51b173cb4a6", "licenses/Barlow-OFL.txt": "186d750eb496a4c17a76385f82be6aea2ac1cf2de074a811d63786cf374ea73f", "licenses/DM-Sans-OFL.txt": "9af36190332437f5ecd09974de43c1f7c77a310a996cdd8ceb25628b458840e1", "licenses/IBM-Plex-OFL.txt": "7e6b2818edbd8f6a01ae80641cc8f16a51080d08fb4e532be3a0b6f74adb07da", "textures/grass.jpg": "57b8041bfe0d0f01430e4dbaad45e7ddddf0a9fc97317f90dbc51f7b0d9e1b5d" }, "screenshots": { "light": "openai-gpt-6.1-sol-codex.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/openai/gpt-6.1-sol/codex/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T03:15:12.672Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Phone portrait and landscape: simultaneous touch driving, steering and shooting, movement, and control release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "1920x1080, DPR 1, headless Chrome with forced SwiftShader software graphics; animation-frame cadence, not a typical-laptop FPS result.", "browserVersion": "156.0.8078.4", "loadMs": 568, "title": "IRONWOOD — Armored Field Operations", "desktopTextBefore": "IRONWOOD\nARMORED FIELD OPERATIONS\nPROVING GROUNDS\nSingle player · Tactical sandbox\nFIELD MANUAL\n?\nTRAINING SECTOR / 04\nLeave your mark.\n\nA living battlefield. Every shot changes the landscape.\n\nSYSTEMS ONLINE\nLOCAL SESSION 01\nLIVE FIELD\n/\nBLACK PINE VALLEY\nSECTOR 04\n48° 16′ N   16° 24′ E\nN\n000\n◇\nFUEL DEPOT\n49 M\nOPERATION ACTIVE\nExplore. Engage. Adapt.\nM1A2 ABRAMS\nMAIN BATTLE TANK\nHULL INTEGRITY\n100%\nREADY\n120 MM · APFSDS\n∞\n00\nKM/H\n1\n↑\n←\n↓\n→\nDrive & steer\nSPACE\nFire cannon\nMove cursor to aim\nP\nPause\n PHYSICS ACTIVE\nYOUR OPERATION\nOP—014\nOperation\nIronwood\n\nTake command. Clear the valley.\nMake your own way through.\n\nPRIMARY OBJECTIVE\nNeutralize hostile assets\nDestroy the marked fuel depots.\n0 / 6 DESTROYED\n0%\nTACTICAL MAP\n04\nN ↑\n— 100 M\nYOU\nHOSTILE\nWOODLAND\nFIELD CONDITIONS\n☀   Clear skies\n18°C\nWind · NW\n8 KM/H\nTerrain\nWOODLAND\nFIELD NOTE\n↗\n\nFire spreads. Fuel explodes.\nThe environment remembers.\n\nREAL-TIME DESTRUCTION\n/\nPERSISTENT FIRE\n/\nTRACKED VEHICLE PHYSICS\nBUILT FOR THE FIELD. v1.0", "canvases": [ { "width": 1089, "height": 528, "visible": true }, { "width": 440, "height": 300, "visible": true } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "frameIntervalsMs": [ 2966.5, 16.799999999999272, 1116.5, 1649.9000000000015, 33.39999999999782, 1100, 700, 1283.2000000000044, 700, 683.2999999999956, 1616.5999999999985, 583.3000000000029, 516.6999999999971, 433.3000000000029, 99.90000000000146, 533.4000000000015, 899.8999999999942, 666.7000000000044, 183.29999999999563, 500, 350, 516.6000000000058, 566.5999999999985 ], "performance": { "samples": 23, "animationFramesPerSecond": 1.3, "medianFrameMs": 583.3, "p95FrameMs": 1649.9, "framesOver50Ms": 21 }, "desktopTextAfter": "IRONWOOD\nARMORED FIELD OPERATIONS\nPROVING GROUNDS\nSingle player · Tactical sandbox\nFIELD MANUAL\n?\nTRAINING SECTOR / 04\nLeave your mark.\n\nA living battlefield. Every shot changes the landscape.\n\nSYSTEMS ONLINE\nLOCAL SESSION 01\nLIVE FIELD\n/\nBLACK PINE VALLEY\nSECTOR 04\n48° 16′ N   16° 24′ E\nN\n000\n◇\nFUEL DEPOT\n20 M\nOPERATION ACTIVE\nExplore. Engage. Adapt.\nM1A2 ABRAMS\nMAIN BATTLE TANK\nHULL INTEGRITY\n100%\nRELOADING\n120 MM · 1.5 S\n∞\n02\nKM/H\n1\n↑\n←\n↓\n→\nDrive & steer\nSPACE\nFire cannon\nMove cursor to aim\nP\nPause\n PHYSICS ACTIVE\nYOUR OPERATION\nOP—014\nOperation\nIronwood\n\nTake command. Clear the valley.\nMake your own way through.\n\nPRIMARY OBJECTIVE\nNeutralize hostile assets\nDestroy the marked fuel depots.\n0 / 6 DESTROYED\n0%\nTACTICAL MAP\n04\nN ↑\n— 100 M\nYOU\nHOSTILE\nWOODLAND\nFIELD CONDITIONS\n☀   Clear skies\n18°C\nWind · NW\n8 KM/H\nTerrain\nWOODLAND\nFIELD NOTE\n↗\n\nFire spreads. Fuel explodes.\nThe environment remembers.\n\nREAL-TIME DESTRUCTION\n/\nPERSISTENT FIRE\n/\nTRACKED VEHICLE PHYSICS\nBUILT FOR THE FIELD. v1.0", "afterResetText": "IRONWOOD\nARMORED FIELD OPERATIONS\nPROVING GROUNDS\nSingle player · Tactical sandbox\nFIELD MANUAL\n?\nTRAINING SECTOR / 04\nLeave your mark.\n\nA living battlefield. Every shot changes the landscape.\n\nSYSTEMS ONLINE\nLOCAL SESSION 01\nLIVE FIELD\n/\nBLACK PINE VALLEY\nSECTOR 04\n48° 16′ N   16° 24′ E\nN\n000\n◇\nFUEL DEPOT\n57 M\nHOSTILE ASSET NEUTRALIZED\n1 of 6 fuel depots destroyed\nM1A2 ABRAMS\nMAIN BATTLE TANK\nHULL INTEGRITY\n100%\nREADY\n120 MM · APFSDS\n∞\n00\nKM/H\n1\n↑\n←\n↓\n→\nDrive & steer\nSPACE\nFire cannon\nMove cursor to aim\nP\nPause\n PHYSICS ACTIVE\nYOUR OPERATION\nOP—014\nOperation\nIronwood\n\nTake command. Clear the valley.\nMake your own way through.\n\nPRIMARY OBJECTIVE\nNeutralize hostile assets\nDestroy the marked fuel depots.\n1 / 6 DESTROYED\n17%\nTACTICAL MAP\n04\nN ↑\n— 100 M\nYOU\nHOSTILE\nWOODLAND\nFIELD CONDITIONS\n☀   Clear skies\n18°C\nWind · NW\n8 KM/H\nTerrain\nWOODLAND\nFIELD NOTE\n↗\n\nFire spreads. Fuel explodes.\nThe environment remembers.\n\nREAL-TIME DESTRUCTION\n/\nPERSISTENT FIRE\n/\nTRACKED VEHICLE PHYSICS\nBUILT FOR THE FIELD. v1.0", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "IRONWOOD\nARMORED FIELD OPERATIONS\nFIELD MANUAL\nTRAINING SECTOR / 04\nLeave your mark.\n\nA living battlefield. Every shot changes the landscape.\n\nLIVE FIELD\n/\nBLACK PINE VALLEY\nSECTOR 04\n48° 16′ N   16° 24′ E\nN\n000\nOPERATION ACTIVE\nExplore. Engage. Adapt.\nM1A2 ABRAMS\nMAIN BATTLE TANK\nHULL INTEGRITY\n100%\nREADY\n120 MM · APFSDS\n00\nKM/H\n▲\n◀\n▼\n▶\nFIRE\n←\n↓\n→\nDrive & steer\nSPACE\nFire cannon\nYOUR OPERATION\nOperation\nIronwood\n\nTake command. Clear the valley.\nMake your own way through.\n\nPRIMARY OBJECTIVE\nNeutralize hostile assets\nDestroy the marked fuel depots.\n0 / 6 DESTROYED\n0%\nTACTICAL MAP\n04\nN ↑\n— 100 M\nYOU\nHOSTILE\nWOODLAND\nFIELD CONDITIONS\n☀   Clear skies\n18°C\nWind · NW\n8 KM/H\nTerrain\nWOODLAND\nFIELD NOTE\n↗\n\nFire spreads. Fuel explodes.\nThe environment remembers.\n\nREAL-TIME DESTRUCTION\n/\nPERSISTENT FIRE\n/\nTRACKED VEHICLE PHYSICS", "buttons": [ { "text": "FIELD MANUAL?", "aria": null, "title": null, "width": 110, "height": 21 }, { "text": "", "aria": "Enable sound", "title": "Enable sound", "width": 14, "height": 14 }, { "text": "", "aria": "Pause", "title": "Pause (P)", "width": 14, "height": 14 }, { "text": "", "aria": "Restart operation", "title": "Restart operation", "width": 14, "height": 14 }, { "text": "", "aria": "Fullscreen", "title": "Fullscreen", "width": 14, "height": 14 }, { "text": "▲", "aria": "Drive forward", "title": null, "width": 42, "height": 42 }, { "text": "◀", "aria": "Turn left", "title": null, "width": 42, "height": 42 }, { "text": "▼", "aria": "Reverse", "title": null, "width": 42, "height": 42 }, { "text": "▶", "aria": "Turn right", "title": null, "width": 42, "height": 42 }, { "text": "FIRE", "aria": "Fire cannon", "title": null, "width": 73, "height": 73 } ] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "IRONWOOD\nARMORED FIELD OPERATIONS\nPROVING GROUNDS\nFIELD MANUAL\n?\nTRAINING SECTOR / 04\nLeave your mark.\n\nA living battlefield. Every shot changes the landscape.\n\nLIVE FIELD\n/\nBLACK PINE VALLEY\nSECTOR 04\n48° 16′ N   16° 24′ E\nN\n000\n◇\nFUEL DEPOT\n49 M\nOPERATION ACTIVE\nExplore. Engage. Adapt.\nM1A2 ABRAMS\nMAIN BATTLE TANK\nHULL INTEGRITY\n100%\nREADY\n120 MM · APFSDS\n00\nKM/H\n▲\n◀\n▼\n▶\nFIRE\n←\n↓\n→\nDrive & steer\nSPACE\nFire cannon\nYOUR OPERATION\nOperation\nIronwood\n\nTake command. Clear the valley.\nMake your own way through.\n\nPRIMARY OBJECTIVE\nNeutralize hostile assets\nDestroy the marked fuel depots.\n0 / 6 DESTROYED\n0%\nTACTICAL MAP\n04\nN ↑\n— 100 M\nYOU\nHOSTILE\nWOODLAND\nFIELD CONDITIONS\n☀   Clear skies\n18°C\nWind · NW\n8 KM/H\nTerrain\nWOODLAND\nFIELD NOTE\n↗\n\nFire spreads. Fuel explodes.\nThe environment remembers.\n\nREAL-TIME DESTRUCTION\n/\nPERSISTENT FIRE\n/\nTRACKED VEHICLE PHYSICS", "buttons": [ { "text": "FIELD MANUAL?", "aria": null, "title": null, "width": 158, "height": 24 }, { "text": "", "aria": "Enable sound", "title": "Enable sound", "width": 14, "height": 14 }, { "text": "", "aria": "Pause", "title": "Pause (P)", "width": 14, "height": 14 }, { "text": "", "aria": "Restart operation", "title": "Restart operation", "width": 14, "height": 14 }, { "text": "", "aria": "Fullscreen", "title": "Fullscreen", "width": 14, "height": 14 }, { "text": "▲", "aria": "Drive forward", "title": null, "width": 42, "height": 42 }, { "text": "◀", "aria": "Turn left", "title": null, "width": 42, "height": 42 }, { "text": "▼", "aria": "Reverse", "title": null, "width": 42, "height": 42 }, { "text": "▶", "aria": "Turn right", "title": null, "width": 42, "height": 42 }, { "text": "FIRE", "aria": "Fire cannon", "title": null, "width": 73, "height": 73 } ] } ], "passed": true, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "initialControls": [ { "key": "ArrowUp", "y": 620.84375, "height": 42 }, { "key": "ArrowLeft", "y": 666.84375, "height": 42 }, { "key": "ArrowDown", "y": 666.84375, "height": 42 }, { "key": "ArrowRight", "y": 666.84375, "height": 42 }, { "key": "Space", "y": 628.34375, "height": 73 } ], "before": { "position": { "x": 0, "z": 27 }, "speed": 0, "reload": 0, "keys": [], "shells": 0 }, "firing": { "position": { "x": -0.000014421486317428549, "z": 27.00249993159478 }, "speed": -0.299398807979237, "reload": 1.5333333333333334, "keys": [ "ArrowUp", "ArrowRight", "Space" ], "shells": 1 }, "held": { "position": { "x": 18.114199382147653, "z": 36.78254497905418 }, "speed": 13, "reload": 1.1833333333333347, "keys": [ "ArrowUp", "ArrowRight", "Space" ], "shells": 1 }, "released": { "position": { "x": 12.496161970717463, "z": 41.642367398527234 }, "speed": 13, "reload": 0.6000000000000008, "keys": [], "shells": 0 }, "method": "Trusted simultaneous CDP touch driving, steering and firing. Existing telemetry observed without modifying simulation state." }, { "viewport": { "width": 844, "height": 390 }, "initialControls": [ { "key": "ArrowUp", "y": 649.984375, "height": 42 }, { "key": "ArrowLeft", "y": 695.984375, "height": 42 }, { "key": "ArrowDown", "y": 695.984375, "height": 42 }, { "key": "ArrowRight", "y": 695.984375, "height": 42 }, { "key": "Space", "y": 657.484375, "height": 73 } ], "before": { "position": { "x": 0, "z": 27 }, "speed": 0, "reload": 0, "keys": [], "shells": 0 }, "firing": { "position": { "x": 0.017629507371123176, "z": 26.782489211944377 }, "speed": 1.8975020636297897, "reload": 1.2833333333333343, "keys": [ "ArrowUp", "ArrowRight", "Space" ], "shells": 1 }, "held": { "position": { "x": 16.94408122999369, "z": 24.63456393598326 }, "speed": 13, "reload": 0.6166666666666675, "keys": [ "ArrowUp", "ArrowRight", "Space" ], "shells": 0 }, "released": { "position": { "x": 19.28403826033726, "z": 33.53495154173869 }, "speed": 12.993451299022519, "reload": 1.4500000000000004, "keys": [], "shells": 1 }, "method": "Trusted simultaneous CDP touch driving, steering and firing. Existing telemetry observed without modifying simulation state." } ] }, "recordedRun": { "createdAt": "2026-11-10T02:36:45.489Z", "executionStartedAt": "2026-11-10T02:36:49.347Z", "completedAt": "2026-11-10T03:10:48.723Z" }, "actualModels": [ { "model": "gpt-6.1-sol", "requests": 87, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "codex", "model": "gpt-6.1-sol", "reasoning": "high", "version": "codex-cli 0.167.0", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "codex", "model": "gpt-6.1-sol", "reasoning": "high", "version": "codex-cli 0.167.0", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "c86fdf10-4138-4bf1-86c8-b6a7847d6b2c", "bucket": 0, "model": "gpt-6.1-sol", "provider": "openai", "usageSource": "boat_workspace_delta", "priceSnapshotId": "4aef97c0-8283-4b09-b51d-e67323cc5255", "effectiveAt": "2026-11-06T01:02:42.703Z", "effectiveUntil": null, "usageTime": "2026-11-10T03:10:48.723Z", "costUsd": "2.113360800000000000000000000000", "categories": { "inputTokens": "0.374372000000000000000000000000", "outputTokens": "0.813110000000000000000000000000", "cacheReadTokens": "0.925878800000000000000000000000", "cacheWriteTokens": "0" }, "tokens": { "inputTokens": 187186, "outputTokens": 81311, "cacheReadTokens": 9258788, "cacheWriteTokens": 0 }, "sourceUrl": "https://developers.openai.com/api/docs/pricing.md", "approximate": true, "stale": false, "assumptions": [ "Standard processing and short context. Regional and tool charges are excluded.", "Cache writes use the 30-minute rate." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "Final software-rendered stress measurements are 11.9 FPS desktop and 27.3 FPS mobile emulation; smooth 30/60 FPS is not achieved consistently in this environment. Physical phones and hardware-accelerated GPUs were not tested.", "Adaptive quality may lower the 3D buffer to DPR .65, shadows to 1024 pixels/10 updates per second, decorative grass density, and HUD blur. Geometry detail and rough-surface shaders are simplified; gameplay damage and fire propagation are preserved.", "Physics use flat terrain, circular collision footprints, fixed 60 Hz approximate ballistics and scripted burning/collapse. Frame catch-up is capped at 250 ms, so severe stalls can slow simulated time.", "Visual storage is bounded: 960 particles, 650 track stamps, 100 craters, 768 rubble instances, 10 shells. Older visual marks are recycled; exhausted particle storage does not cancel blast damage or ignition.", "Profiles use individual 8-second windows plus repeated resets; whole-process memory/VRAM, long thermal/battery behavior and prolonged device soaks were not measured.", "One map/mission with no save across reload. WebGL is required. Production build passes with a chunk-size advisory." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 18333, "exitCode": 0, "executionStartedAtMs": 1794278209347, "completedAtMs": 1794279167141 }, { "name": "efficiency", "status": "finished", "pid": 33605, "exitCode": 0, "executionStartedAtMs": 1794279178360, "completedAtMs": 1794280248723 } ], "skillCount": 0 }, { "key": "openai-gpt-6-astra-codex", "harness": "codex", "harnessName": "Codex", "provider": "openai", "providerName": "OpenAI", "model": "gpt-6-astra", "name": "GPT-6 Astra", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 2054.285, "totalSeconds": 2066.54, "tokens": { "inputTokens": 159536, "outputTokens": 75747, "cacheReadTokens": 7308163, "cacheWriteTokens": 0 }, "cost": { "pricedSubtotalUsd": "12.690873000000000000000000000000", "completeTotalUsd": "12.690873000000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: Smooth 30 or 60 FPS was NOT achieved in the available environment; final stress results were 7.16 FPS desktop and 16.32 FPS mobile emulation. Further GPU-focused optimization and physical-device validation remain necessary. Mobile testing used touch emulation on the desktop Intel UHD 620 GPU. It does not establish performance, thermal behavior, battery use, or compatibility on physical phones.", "files": { "assets/font-0-D9VpUvlI.ttf": "a35caa4687f434d66453c1fe423d7074cfa9fd7ab80c19b0780f3862887d2fc9", "assets/font-1-Apw_3I1e.ttf": "91c841fdfa8e7b94ffedbb983a363947ba6ed720f3bbf0c71d48b618053655bc", "assets/font-10-DyRt3UMW.ttf": "4fc14a73ca53ba9d32fd759ae1ca1a3133326035d0dd337862b3ee1633cc156e", "assets/font-11-DmWgiO8-.ttf": "754dfc9d50cf7aabfb6b108d2c2f7d20a3f1f2cc6f6c01640c6728091272cec0", "assets/font-2-S-AH7y_S.ttf": "c15439e7a03af5714282ec1780ff7b0214ec6a7db96300b54928dbcd2569ca0c", "assets/font-3-CoiCOvr2.ttf": "74103fd6d41bc32c63aa2e6eb3db619ddb80ce9ca1b1420d5548a23b8ee5d154", "assets/font-4-VEi3Du1s.ttf": "3f25dda39ecf9f2519a703b217629fbdfc54fb5e5725bf4aa2b9234e78a2fc04", "assets/font-5-CCxn3Epz.ttf": "cc9fb2539cf27df48255982e896d28f544483537cdb41f9fabd741af8992d719", "assets/font-6-RZZ47rCX.ttf": "0d85af813fc3ed87db0c6265515689b2eef5cbaf7aab17922528dfc95a2cd73f", "assets/font-7-CQj3XIHL.ttf": "7dde307fa887fc65ff5830cfada77a7decc5dae8d3c816c9d39ba3f1af1c4ed7", "assets/font-8-BdFBoAAT.ttf": "73bc9e17231e9073780d3c81574c7253cecd5645c93544df1a935054b39a364f", "assets/font-9-ClCTSiDw.ttf": "a7fdc993f9f0387caedba4e42521136fa97e4c0837ae6239bf0413ccf8af00c2", "assets/index-BcsjDgJU.css": "7f04a5687a67c41072bf480b560d5d5c0f9dbeadcc4d27b55ac055088c943f47", "assets/index-D1yYB_l7.js": "3a3679eaf832ba1485724a8820fc6cf13e4921debc0269d0116658848ae76c4e", "assets/three-M4BZUMwk.js": "3deac5a71d6ec515c1fb7e86dd5b14dfc6a7258c208d04e9e8062e214d509bea", "favicon.svg": "a87a0cb6e07c52e1448e791672f9bc261ad47c7d0e97838f6fb512a8417ea687", "index.html": "e00d66fa72c951878d8f9299a195776830d5a7a296d171e1171c8da872498b6e" }, "screenshots": { "light": "openai-gpt-6-astra-codex.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/openai/gpt-6-astra/codex/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T03:17:06.449Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Phone portrait and landscape: simultaneous touch driving, steering and shooting, movement, and control release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "1920x1080, DPR 1, headless Chrome with forced SwiftShader software graphics; animation-frame cadence, not a typical-laptop FPS result.", "browserVersion": "156.0.8078.4", "loadMs": 580, "title": "IRONFIELD — Proving Grounds", "desktopTextBefore": "IRONFIELD\nTACTICAL SANDBOX\nPROVING GROUNDS\n01\nFIELD GUIDE\nSINGLE PLAYER\nFREE-ROAM SIMULATION\nLeave your mark.\n\nA living battlefield. Everything reacts.\n\n00:01\nPAUSE\nESC\nFIELD EXERCISE\n01\nScorched earth\n\nPut your firepower to the test.\nDestroy all marked targets.\n\nTARGETS DESTROYED\n0 / 8\nORANGE DIAMONDS MARK TARGETS\n24°C\nCLEAR\nWIND\n↗\nNE · 8 KM/H\nN\nW\n+\nE\nS\n01\n02\n04\n05\n06\n07\nYOU\n120 MM\nSystems online. Arrow keys to drive. Space to fire.\nYOUR VEHICLE\nM1A2 ABRAMS\nOPERATIONAL\nARMOR\n100%\nAMMO\n40 AP\n\n● GUN READY\n\nSPEED\n0 KM/H\n\nNEUTRAL\n\n+\nBLACK PINE TRAINING RANGE\n48° 31′ 12″ N   16° 22′ 08″ E\nSECTOR 07\nN ↑\nYOU\nTARGET\n200 M\n↑\n←\n↓\n→\nDRIVE & STEER\nSPACE\nFIRE MAIN GUN\nAIM TURRET\nREAL-TIME DESTRUCTION\nRESET FIELD", "canvases": [ { "width": 1920, "height": 939, "visible": true }, { "width": 240, "height": 150, "visible": true } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "frameIntervalsMs": [ 2716.600000000002, 16.700000000000728, 350, 3266.5, 33.299999999995634, 3816.5, 66.70000000000437, 10699.599999999999, 16.69999999999709, 33.200000000004366, 16.69999999999709, 16.700000000004366, 16.69999999999709, 499.90000000000146, 1416.699999999997, 16.599999999998545, 100, 183.40000000000146, 2016.5, 1916.5999999999985, 16.700000000004366, 100, 633.2999999999956 ], "performance": { "samples": 23, "animationFramesPerSecond": 0.8, "medianFrameMs": 100, "p95FrameMs": 3816.5, "framesOver50Ms": 14 }, "desktopTextAfter": "IRONFIELD\nTACTICAL SANDBOX\nPROVING GROUNDS\n01\nFIELD GUIDE\nSINGLE PLAYER\nFREE-ROAM SIMULATION\nLeave your mark.\n\nA living battlefield. Everything reacts.\n\n00:04\nPAUSE\nESC\nFIELD EXERCISE\n01\nScorched earth\n\nPut your firepower to the test.\nDestroy all marked targets.\n\nTARGETS DESTROYED\n2 / 8\nORANGE DIAMONDS MARK TARGETS\n24°C\nCLEAR\nWIND\n↗\nNE · 8 KM/H\nN\nW\n+\nE\nS\n03\n04\n05\n06\n07\n08\nYOU\n120 MM\nFuel depot detonated · 2 / 8 targets\nYOUR VEHICLE\nM1A2 ABRAMS\nOPERATIONAL\nARMOR\n100%\nAMMO\n38 AP\n\n● GUN READY\n\nSPEED\n42 KM/H\n\nFORWARD\n\n+\nBLACK PINE TRAINING RANGE\n48° 31′ 12″ N   16° 22′ 08″ E\nSECTOR 07\nN ↑\nYOU\nTARGET\n200 M\n↑\n←\n↓\n→\nDRIVE & STEER\nSPACE\nFIRE MAIN GUN\nAIM TURRET\nREAL-TIME DESTRUCTION\nRESET FIELD", "afterResetText": "IRONFIELD\nTACTICAL SANDBOX\nPROVING GROUNDS\n01\nFIELD GUIDE\nSINGLE PLAYER\nFREE-ROAM SIMULATION\nLeave your mark.\n\nA living battlefield. Everything reacts.\n\n00:10\nPAUSE\nESC\nFIELD EXERCISE\n01\nScorched earth\n\nPut your firepower to the test.\nDestroy all marked targets.\n\nTARGETS DESTROYED\n2 / 8\nORANGE DIAMONDS MARK TARGETS\n24°C\nCLEAR\nWIND\n↗\nNE · 8 KM/H\nN\nW\n+\nE\nS\n03\n04\n05\n06\n07\n08\nYOU\n120 MM\nFuel depot detonated · 2 / 8 targets\nYOUR VEHICLE\nM1A2 ABRAMS\nOPERATIONAL\nARMOR\n100%\nAMMO\n37 AP\n\n● GUN READY\n\nSPEED\n0 KM/H\n\nNEUTRAL\n\n+\nBLACK PINE TRAINING RANGE\n48° 31′ 12″ N   16° 22′ 08″ E\nSECTOR 07\nN ↑\nYOU\nTARGET\n200 M\n↑\n←\n↓\n→\nDRIVE & STEER\nSPACE\nFIRE MAIN GUN\nAIM TURRET\nREAL-TIME DESTRUCTION\nRESET FIELD", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "IRONFIELD\nTACTICAL SANDBOX\nFIELD GUIDE\nFREE-ROAM SIMULATION\nLeave your mark.\n\nA living battlefield. Everything reacts.\n\nFIELD EXERCISE\n01\nScorched earth\n\nPut your firepower to the test.\nDestroy all marked targets.\n\nTARGETS DESTROYED\n0 / 8\nORANGE DIAMONDS MARK TARGETS\nN\nW\n+\nE\nS\n01\n06\nYOU\n120 MM\nSystems online. Use touch controls. Tap field to aim.\nYOUR VEHICLE\nM1A2 ABRAMS\nOPERATIONAL\nARMOR\n100%\nAMMO\n40 AP\n\n● GUN READY\n\nSPEED\n0 KM/H\n\nNEUTRAL\n\nSECTOR 07\n↑\n←\n↓\n→\nFIRE\nRESET FIELD", "buttons": [ { "text": "FIELD GUIDE", "aria": null, "title": null, "width": 79, "height": 63 }, { "text": "", "aria": "Enable sound", "title": "Enable sound", "width": 32, "height": 32 }, { "text": "", "aria": "Settings", "title": "Settings", "width": 32, "height": 32 }, { "text": "PAUSEESC", "aria": null, "title": null, "width": 33, "height": 31 }, { "text": "", "aria": "Expand tactical map", "title": "Expand tactical map", "width": 12, "height": 12 }, { "text": "↑", "aria": "Drive forward", "title": null, "width": 42, "height": 42 }, { "text": "←", "aria": "Turn left", "title": null, "width": 42, "height": 42 }, { "text": "↓", "aria": "Reverse", "title": null, "width": 42, "height": 42 }, { "text": "→", "aria": "Turn right", "title": null, "width": 42, "height": 42 }, { "text": "FIRE", "aria": "Fire main gun", "title": null, "width": 64, "height": 64 }, { "text": "RESET FIELD", "aria": null, "title": null, "width": 84, "height": 23 } ] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "IRONFIELD\nTACTICAL SANDBOX\nPROVING GROUNDS\n01\nFIELD GUIDE\nFREE-ROAM SIMULATION\nLeave your mark.\n\nA living battlefield. Everything reacts.\n\n00:01\nPAUSE\nESC\nFIELD EXERCISE\n01\nScorched earth\n\nPut your firepower to the test.\nDestroy all marked targets.\n\nTARGETS DESTROYED\n0 / 8\nORANGE DIAMONDS MARK TARGETS\n24°C\nCLEAR\nWIND\n↗\nNE · 8 KM/H\nN\nW\n+\nE\nS\n01\n02\n04\n05\n06\n07\nYOU\n120 MM\nSystems online. Arrow keys to drive. Space to fire.\nYOUR VEHICLE\nM1A2 ABRAMS\nOPERATIONAL\nARMOR\n100%\nAMMO\n40 AP\n\n● GUN READY\n\nSPEED\n0 KM/H\n\nNEUTRAL\n\nSECTOR 07\nN ↑\nYOU\nTARGET\n200 M\n↑\n←\n↓\n→\nFIRE\n↑\n←\n↓\n→\nDRIVE & STEER\nSPACE\nFIRE MAIN GUN\nAIM TURRET\nRESET FIELD", "buttons": [ { "text": "PROVING GROUNDS 01", "aria": null, "title": null, "width": 153, "height": 67 }, { "text": "FIELD GUIDE", "aria": null, "title": null, "width": 117, "height": 67 }, { "text": "", "aria": "Enable sound", "title": "Enable sound", "width": 32, "height": 32 }, { "text": "", "aria": "Settings", "title": "Settings", "width": 42, "height": 32 }, { "text": "PAUSEESC", "aria": null, "title": null, "width": 104, "height": 34 }, { "text": "", "aria": "Expand tactical map", "title": "Expand tactical map", "width": 12, "height": 12 }, { "text": "↑", "aria": "Drive forward", "title": null, "width": 45, "height": 42 }, { "text": "←", "aria": "Turn left", "title": null, "width": 45, "height": 42 }, { "text": "↓", "aria": "Reverse", "title": null, "width": 45, "height": 42 }, { "text": "→", "aria": "Turn right", "title": null, "width": 45, "height": 42 }, { "text": "FIRE", "aria": "Fire main gun", "title": null, "width": 65, "height": 65 }, { "text": "RESET FIELD", "aria": null, "title": null, "width": 112, "height": 23 } ] } ], "passed": true, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "before": { "position": { "x": 0, "z": 19 }, "heading": 0, "turret": 0, "speed": 0, "health": 100, "ammo": 40, "shots": 0, "destroyed": 0, "burning": 0, "objectsDestroyed": 0, "paused": false, "elapsed": 0.49999999999999994, "targets": [ { "x": 0, "z": -13, "hp": 25, "dead": false, "screen": { "x": 195, "y": 207.6784599342133, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 461.18962432915924, "y": 74.65626083333838, "visible": false } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": -175.7641196013289, "y": -152.26396116227178, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 556.257347303859, "y": 371.99999999999994, "visible": false } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": -33.16253513927932, "y": -27.066597302624857, "visible": false } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 385.1354459493994, "y": 614.569892478066, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": -137.737030411449, "y": 371.99999999999994, "visible": false } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 660.8318425760286, "y": -120.96462019736002, "visible": false } } ], "performance": { "pixelRatio": 1, "frameAverageMs": 0, "droppedTime": 2.936100000095367, "particles": 0, "awakeDebris": 0, "geometries": 28, "textures": 22 }, "renderer": { "frame": 2, "calls": 24, "triangles": 71806, "points": 0, "lines": 0 } }, "firing": { "position": { "x": -0.0008245681602320754, "z": 19.023791562729016 }, "heading": -0.10085655480301989, "turret": -0.10085655480301989, "speed": 0.17813729614936413, "health": 100, "ammo": 39, "shots": 1, "destroyed": 0, "burning": 0, "objectsDestroyed": 0, "paused": false, "elapsed": 1.0833333333333344, "targets": [ { "x": 0, "z": -13, "hp": 25, "dead": false, "screen": { "x": 194.7881381848321, "y": 207.63487683614008, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 460.8037317529377, "y": 74.39031579941654, "visible": false } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": -176.44625674423824, "y": -151.94317938057753, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 556.2600652823426, "y": 371.62007630734115, "visible": false } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": -33.681106149347755, "y": -26.88462864150597, "visible": false } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 385.4553322489782, "y": 614.3171744239999, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": -137.7339107856611, "y": 372.2345653880029, "visible": false } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 660.1901334213242, "y": -121.38772943616232, "visible": false } } ], "performance": { "pixelRatio": 1, "frameAverageMs": 21.5361976973525, "droppedTime": 17.535500000095364, "particles": 26, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 17, "calls": 28, "triangles": 71898, "points": 0, "lines": 0 } }, "held": { "position": { "x": 4.782504606779872, "z": 14.942628152736757 }, "heading": -1.3214836545813224, "turret": -1.3214836545813224, "speed": 7.436244664577476, "health": 100, "ammo": 38, "shots": 2, "destroyed": 1, "burning": 5, "objectsDestroyed": 4, "paused": false, "elapsed": 2.499999999999996, "targets": [ { "x": 0, "z": -13, "hp": -51, "dead": true, "screen": { "x": 161.0887486013484, "y": 234.2053324797727, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 426.80973398371736, "y": 100.61066325246809, "visible": false } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": -210.93888767214113, "y": -124.68950199574172, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 522.9221153782355, "y": 397.5916871853692, "visible": false } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": -67.89787622390969, "y": 0.11102105258450745, "visible": false } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 352.6534758917238, "y": 640.4575191644869, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": -171.06935849639046, "y": 399.2433579599338, "visible": false } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 625.7635799013859, "y": -95.39558031252406, "visible": false } } ], "performance": { "pixelRatio": 0.85, "frameAverageMs": 137.66282507502905, "droppedTime": 19.052000000095358, "particles": 188, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 27, "calls": 26, "triangles": 69920, "points": 0, "lines": 0 } }, "released": { "position": { "x": 9.291254934696877, "z": 15.039204278456767 }, "heading": -1.830891795255209, "turret": -1.830891795255209, "speed": 9.50597247935356, "health": 100, "ammo": 38, "shots": 2, "destroyed": 1, "burning": 5, "objectsDestroyed": 4, "paused": false, "elapsed": 3.0333333333333274, "targets": [ { "x": 0, "z": -13, "hp": -51, "dead": true, "screen": { "x": 124.15797745197163, "y": 238.45730700351288, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 390.3432722674481, "y": 105.42765033645989, "visible": false } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": -246.61785691255074, "y": -121.48123667831582, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 485.42067273100713, "y": 402.7728435794659, "visible": false } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": -104.01219775515415, "y": 3.7144777595622, "visible": false } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 314.3066662593186, "y": 645.3494054349117, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": -208.57370473528832, "y": 402.7881456700935, "visible": false } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 589.9791236512559, "y": -90.19996835837196, "visible": false } } ], "performance": { "pixelRatio": 0.85, "frameAverageMs": 154.5755021487031, "droppedTime": 20.95190000009536, "particles": 181, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 31, "calls": 27, "triangles": 46530, "points": 0, "lines": 0 } }, "method": "Trusted simultaneous CDP touch driving, steering and firing. Existing telemetry observed without modifying simulation state." }, { "viewport": { "width": 844, "height": 390 }, "before": { "position": { "x": 0, "z": 19 }, "heading": 0, "turret": 0, "speed": 0, "health": 100, "ammo": 40, "shots": 0, "destroyed": 0, "burning": 0, "objectsDestroyed": 0, "paused": false, "elapsed": 0.49999999999999994, "targets": [ { "x": 0, "z": -13, "hp": 25, "dead": false, "screen": { "x": 422, "y": 131.7748348198319, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 552.7692307692307, "y": 66.42589157874349, "visible": true } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": 239.8571428571429, "y": -45.05171747958383, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 599.4725274725274, "y": 212.49999999999997, "visible": true } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": 309.91208791208794, "y": 16.45317027673467, "visible": true } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 515.4065934065934, "y": 331.6657200278671, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": 258.53846153846155, "y": 212.49999999999997, "visible": true } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 650.8461538461538, "y": -29.67549554050418, "visible": false } } ], "performance": { "pixelRatio": 1, "frameAverageMs": 816.5920000000001, "droppedTime": 1.7572999999523162, "particles": 0, "awakeDebris": 0, "geometries": 28, "textures": 22 }, "renderer": { "frame": 2, "calls": 24, "triangles": 101962, "points": 0, "lines": 0 } }, "firing": { "position": { "x": 0.000027672352629946932, "z": 18.99779565832023 }, "heading": -0.01255290836879131, "turret": -0.012552908368791308, "speed": -0.7677290780217253, "health": 100, "ammo": 39, "shots": 1, "destroyed": 0, "burning": 0, "objectsDestroyed": 0, "paused": false, "elapsed": 1.0000000000000013, "targets": [ { "x": 0, "z": -13, "hp": 25, "dead": false, "screen": { "x": 422.1748432696128, "y": 131.75742416643158, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 553.0854118184576, "y": 66.58598384296698, "visible": true } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": 240.41528130765062, "y": -45.37549950058347, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 599.4722382685559, "y": 212.76088892532343, "visible": true } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": 310.3368956507993, "y": 16.24578716090444, "visible": true } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 515.1483252437613, "y": 331.82953425580007, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": 258.53871413641025, "y": 212.26049852831088, "visible": true } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 651.370334319921, "y": -29.392672479755152, "visible": false } } ], "performance": { "pixelRatio": 0.85, "frameAverageMs": 407.2209812658317, "droppedTime": 12.923499999952316, "particles": 20, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 17, "calls": 28, "triangles": 102042, "points": 0, "lines": 0 } }, "held": { "position": { "x": 7.0485756255510195, "z": 14.703285714550486 }, "heading": -1.5888344829618635, "turret": -1.5888344829618635, "speed": 8.601494972412315, "health": 100, "ammo": 38, "shots": 2, "destroyed": 1, "burning": 5, "objectsDestroyed": 4, "paused": false, "elapsed": 2.8166666666666615, "targets": [ { "x": 0, "z": -13, "hp": -51, "dead": true, "screen": { "x": 396.24186663776464, "y": 146.87164425453508, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 527.0106954966312, "y": 81.52114025124759, "visible": true } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": 214.0979219796129, "y": -29.956898042306243, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 573.7148905834573, "y": 227.59732491842442, "visible": true } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": 284.1532453006615, "y": 31.54865396715687, "visible": true } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 489.6496894109151, "y": 346.7652477403216, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": 232.7808246537595, "y": 227.59874544772188, "visible": true } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 625.0870275301085, "y": -14.582149486424182, "visible": false } } ], "performance": { "pixelRatio": 0.7, "frameAverageMs": 362.3176655894598, "droppedTime": 14.973299999952316, "particles": 181, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 29, "calls": 29, "triangles": 74776, "points": 0, "lines": 0 } }, "released": { "position": { "x": 11.581254097643681, "z": 15.988993663058487 }, "heading": -2.078042575312545, "turret": -2.078042575312545, "speed": 10.308172064810737, "health": 100, "ammo": 38, "shots": 2, "destroyed": 1, "burning": 5, "objectsDestroyed": 4, "paused": false, "elapsed": 3.3166666666666598, "targets": [ { "x": 0, "z": -13, "hp": -51, "dead": true, "screen": { "x": 376.89898366992657, "y": 145.04529173331684, "visible": true } }, { "x": 28, "z": -30, "hp": 25, "dead": false, "screen": { "x": 507.6684695622771, "y": 79.69668012713163, "visible": true } }, { "x": -39, "z": -59, "hp": 25, "dead": false, "screen": { "x": 194.75681686298424, "y": -31.781780797452722, "visible": false } }, { "x": 38, "z": 8, "hp": 125, "dead": false, "screen": { "x": 554.3711959886147, "y": 225.7709438711044, "visible": true } }, { "x": -24, "z": -43, "hp": 125, "dead": false, "screen": { "x": 264.81152180105477, "y": 29.723305631620978, "visible": true } }, { "x": 20, "z": 39, "hp": 150, "dead": false, "screen": { "x": 470.304796697371, "y": 344.93646750210996, "visible": true } }, { "x": -35, "z": 8, "hp": 150, "dead": false, "screen": { "x": 213.43713005630892, "y": 225.77004216064054, "visible": true } }, { "x": 49, "z": -55, "hp": 150, "dead": false, "screen": { "x": 605.7457678207452, "y": -16.404468517491363, "visible": false } } ], "performance": { "pixelRatio": 0.7, "frameAverageMs": 363.36178230998684, "droppedTime": 15.973199999952316, "particles": 178, "awakeDebris": 0, "geometries": 29, "textures": 22 }, "renderer": { "frame": 31, "calls": 29, "triangles": 73946, "points": 0, "lines": 0 } }, "method": "Trusted simultaneous CDP touch driving, steering and firing. Existing telemetry observed without modifying simulation state." } ] }, "recordedRun": { "createdAt": "2026-11-10T02:36:49.260Z", "executionStartedAt": "2026-11-10T02:36:53.164Z", "completedAt": "2026-11-10T03:11:15.800Z" }, "actualModels": [ { "model": "gpt-6-astra", "requests": 80, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "codex", "model": "gpt-6-astra", "reasoning": "high", "version": "codex-cli 0.167.0", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "codex", "model": "gpt-6-astra", "reasoning": "high", "version": "codex-cli 0.167.0", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "69823ece-283c-457a-abd6-ae09236134b4", "bucket": 0, "model": "gpt-6-astra", "provider": "openai", "usageSource": "boat_workspace_delta", "priceSnapshotId": "d694c53e-23d8-4769-baba-55680769f1ab", "effectiveAt": "2026-11-06T01:02:42.703Z", "effectiveUntil": null, "usageTime": "2026-11-10T03:11:15.800Z", "costUsd": "12.690873000000000000000000000000", "categories": { "inputTokens": "1.595360000000000000000000000000", "outputTokens": "3.787350000000000000000000000000", "cacheReadTokens": "7.308163000000000000000000000000", "cacheWriteTokens": "0" }, "tokens": { "inputTokens": 159536, "outputTokens": 75747, "cacheReadTokens": 7308163, "cacheWriteTokens": 0 }, "sourceUrl": "https://developers.openai.com/api/docs/pricing.md", "approximate": true, "stale": false, "assumptions": [ "Standard processing and short context. Regional and tool charges are excluded.", "Cache writes use the 30-minute rate." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "Smooth 30 or 60 FPS was NOT achieved in the available environment; final stress results were 7.16 FPS desktop and 16.32 FPS mobile emulation. Further GPU-focused optimization and physical-device validation remain necessary.", "Mobile testing used touch emulation on the desktop Intel UHD 620 GPU. It does not establish performance, thermal behavior, battery use, or compatibility on physical phones.", "Adaptive 3D resolution can fall to 0.6 desktop / 0.7 mobile pixel ratio. Distant foliage uses fewer polygons and slow-device grass density is reduced. These preserve scene composition, materials, lighting and gameplay but reduce world-image sharpness/detail.", "Performance results are single sequential runs on a shared host with variable scheduling. They are not statistically stable hardware guarantees. Geometry/texture counts are resource counts, not measurements of total GPU memory bytes.", "The fixed-step simulation caps catch-up at 15 steps / 250 ms; sustained extreme stalls can drop simulated time to avoid a backlog. The final stress runs advanced approximately 12 simulation seconds in 12 wall seconds.", "Physics remains a game approximation with circular collision bounds, material damage thresholds, projectile gravity and bouncing debris, rather than a complete structural or combustion simulation.", "One development profiling attempt timed out because a Vite cache-busting query prevented instrumentation; the URL matcher was fixed. A subsequent development mobile profiling attempt ended unexpectedly with a closed browser, cause undetermined. All four final production comparison runs completed; final production browser checks reported no errors.", "Vite reports a 504.38 kB vendor chunk size advisory; the production build succeeds." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 18365, "exitCode": 0, "executionStartedAtMs": 1794278213164, "completedAtMs": 1794279171268 }, { "name": "efficiency", "status": "finished", "pid": 32029, "exitCode": 0, "executionStartedAtMs": 1794279179619, "completedAtMs": 1794280275800 } ], "skillCount": 0 }, { "key": "anthropic-claude-opus-5-5-claude", "harness": "claude", "harnessName": "Claude Code", "provider": "anthropic", "providerName": "Anthropic", "model": "claude-opus-5-5", "name": "Claude Opus 5.5", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 4859.87, "totalSeconds": 4869.187, "tokens": { "inputTokens": 224, "outputTokens": 263620, "cacheReadTokens": 21527279, "cacheWriteTokens": 994201 }, "cost": { "pricedSubtotalUsd": "14.549756800000000000000000000000", "completeTotalUsd": "14.549756800000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: There is no real GPU in this sandbox and no physical phone. All fps figures come from headless Chrome on software-backed GL paths and from mobile emulation with CPU throttling, so they are proxies. Real hardware will differ, and on GPUs these paths are typically much faster at canvas fill. The heaviest scenes do not reach a steady 60 fps on the slow default GL path: about 18-30 fps adaptive on desktop and 20-33 fps on emulated mobile. On the SwiftShader path, desktop adaptive holds 56-60 fps, but full-quality inferno is ~29 fps and emulated mobile is 26-38 fps.", "files": { "assets/index-CsX5D1z1.css": "e9be8136441451564a3aa39581c4826aebcd09c3e27ac48ba886ad70a83b4b79", "assets/index-u_cJ7yeE.js": "e202527a73f0651fce806ff3fdbd640970a44a5cf38d0b350b0d0161962ddf98", "index.html": "4cd6a28cfca0cc62353a38ac931cf5d05ffba693d859bf4809dfb476a0094f32" }, "screenshots": { "light": "anthropic-claude-opus-5-5-claude.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/anthropic/claude-opus-5-5/claude/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T04:00:30.402Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Phone portrait and landscape: trusted simultaneous joystick movement and firing, observed movement/shots, and input release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "1920x1080, DPR 1, headless Chrome with forced SwiftShader software graphics; animation-frame cadence, not a typical-laptop FPS result.", "browserVersion": "156.0.8078.4", "loadMs": 634, "title": "Ironfield — Top-down Tank Sandbox", "desktopTextBefore": "HULL\nGUN\nREADY\n0 km/h\nScore 0\nTrees 0 · Bldg 0 · Veh 0 · Fires 0\n Wind 13 km/h\n?\nDestroy anything. Watch the wind.\n↑ ↓ drive · ← → steer · Space fire · Mouse aim (shell bursts at cursor) · click fire · Q/E turret · C centre turret · +/− zoom · M mute", "canvases": [ { "width": 1920, "height": 1080, "visible": true }, { "width": 168, "height": 168, "visible": true } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "frameIntervalsMs": [ 1750, 16.599999999998545, 16.69999999999709, 16.599999999998545, 16.700000000004366, 116.69999999999709, 16.599999999998545, 33.400000000001455, 16.599999999998545, 16.700000000004366, 16.69999999999709, 16.599999999998545, 16.700000000004366, 16.599999999998545, 16.69999999999709, 16.700000000004366, 16.69999999999709, 16.599999999998545, 16.700000000004366, 16.69999999999709, 16.599999999998545, 16.700000000004366, 16.599999999998545, 16.69999999999709, 16.700000000004366, 16.69999999999709, 16.599999999998545, 2216.600000000006, 16.599999999998545, 16.69999999999709, 33.400000000001455, 16.69999999999709, 16.60000000000582, 16.69999999999709, 16.599999999998545, 16.700000000004366, 2799.899999999994, 16.60000000000582, 66.69999999999709, 50, 16.69999999999709, 16.60000000000582, 16.69999999999709, 16.69999999999709, 2149.800000000003, 16.69999999999709, 16.700000000004366, 16.69999999999709, 16.599999999998545, 16.700000000004366, 16.69999999999709, 16.700000000004366, 16.599999999998545, 2166.5999999999985, 16.69999999999709, 16.60000000000582, 33.299999999995634, 50.099999999998545, 1966.5, 16.700000000004366, 16.599999999998545, 16.80000000000291, 16.599999999998545, 16.69999999999709, 16.700000000004366, 16.69999999999709, 16.5, 16.69999999999709, 16.700000000004366, 16.69999999999709, 1949.9000000000015, 16.599999999998545, 16.700000000004366, 16.69999999999709, 16.69999999999709, 16.60000000000582, 16.69999999999709, 16.69999999999709, 16.60000000000582, 16.69999999999709 ], "performance": { "samples": 80, "animationFramesPerSecond": 4.9, "medianFrameMs": 16.7, "p95FrameMs": 1966.5, "framesOver50Ms": 10 }, "desktopTextAfter": "HULL\nGUN\nLOADING\n16 km/h\nScore 0\nTrees 0 · Bldg 0 · Veh 0 · Fires 0\n Wind 13 km/h\n?\nDestroy anything. Watch the wind.\n↑ ↓ drive · ← → steer · Space fire · Mouse aim (shell bursts at cursor) · click fire · Q/E turret · C centre turret · +/− zoom · M mute", "afterResetText": "HULL\nGUN\nREADY\n0 km/h\nScore 0\nTrees 0 · Bldg 0 · Veh 0 · Fires 0\n Wind 13 km/h\n?\nDestroy anything. Watch the wind.\n↑ ↓ drive · ← → steer · Space fire · Mouse aim (shell bursts at cursor) · click fire · Q/E turret · C centre turret · +/− zoom · M mute", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "HULL\nGUN\nREADY\n0 km/h\nScore 0\n Wind 11 km/h\n?\nDestroy anything. Watch the wind.\nAIM\nFIRE", "buttons": [ { "text": "", "aria": null, "title": "Mute (M)", "width": 32, "height": 32 }, { "text": "?", "aria": null, "title": "Controls", "width": 32, "height": 32 }, { "text": "", "aria": null, "title": "Restart map", "width": 32, "height": 32 } ] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "HULL\nGUN\nREADY\n0 km/h\nScore 0\n Wind 11 km/h\n?\nAIM\nFIRE", "buttons": [ { "text": "", "aria": null, "title": "Mute (M)", "width": 32, "height": 32 }, { "text": "?", "aria": null, "title": "Controls", "width": 32, "height": 32 }, { "text": "", "aria": null, "title": "Restart map", "width": 32, "height": 32 } ] } ], "passed": true, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "before": { "position": { "x": 128, "y": 142 }, "reload": 0, "shots": 0, "move": { "x": 0, "y": 0, "active": false }, "fire": false }, "firing": { "position": { "x": 127.9557262067555, "y": 142.00899613412955 }, "reload": 1.8000000000000003, "shots": 1, "move": { "x": 0, "y": -0.7258064516129032, "active": true }, "fire": true }, "held": { "position": { "x": 129.75520059606734, "y": 134.89803631767157 }, "reload": 1.2167999999999999, "shots": 2, "move": { "x": 0, "y": -0.7258064516129032, "active": true }, "fire": true }, "released": { "position": { "x": 129.76910798182087, "y": 134.486827175243 }, "reload": 1.1501000000000003, "shots": 2, "move": { "x": 0, "y": 0, "active": false }, "fire": false }, "method": "Trusted simultaneous CDP floating-joystick drag and fire button; existing telemetry observed only." }, { "viewport": { "width": 844, "height": 390 }, "before": { "position": { "x": 128, "y": 142 }, "reload": 0, "shots": 0, "move": { "x": 0, "y": 0, "active": false }, "fire": false }, "firing": { "position": { "x": 127.9557262067555, "y": 142.00899613412955 }, "reload": 1.8000000000000003, "shots": 1, "move": { "x": 0, "y": -0.7258064516129032, "active": true }, "fire": true }, "held": { "position": { "x": 129.7479031311446, "y": 135.01480445784873 }, "reload": 1.2000000000000002, "shots": 2, "move": { "x": 0, "y": -0.7258064516129032, "active": true }, "fire": true }, "released": { "position": { "x": 129.75865985211806, "y": 134.7078877526039 }, "reload": 1.1500000000000001, "shots": 2, "move": { "x": 0, "y": 0, "active": false }, "fire": false }, "method": "Trusted simultaneous CDP floating-joystick drag and fire button; existing telemetry observed only." } ] }, "recordedRun": { "createdAt": "2026-11-10T02:36:53.071Z", "executionStartedAt": "2026-11-10T02:36:56.678Z", "completedAt": "2026-11-10T03:58:02.258Z" }, "actualModels": [ { "model": "claude-opus-5-5", "requests": 109, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "claude", "model": "claude-opus-5-5", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "claude", "model": "claude-opus-5-5", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "5e6e0218-d36f-4abc-ab4f-571b636dd947", "bucket": 0, "model": "claude-opus-5-5", "provider": "anthropic", "usageSource": "boat_workspace_delta", "priceSnapshotId": "88f7b0a9-e0c1-4505-b15f-a2d44355c36f", "effectiveAt": "2026-11-06T01:02:42.711Z", "effectiveUntil": null, "usageTime": "2026-11-10T03:58:02.258Z", "costUsd": "14.549756800000000000000000000000", "categories": { "inputTokens": "0.000896000000000000000000000000", "outputTokens": "5.272400000000000000000000000000", "cacheReadTokens": "4.305455800000000000000000000000", "cacheWriteTokens": "4.971005000000000000000000000000" }, "tokens": { "inputTokens": 224, "outputTokens": 263620, "cacheReadTokens": 21527279, "cacheWriteTokens": 994201 }, "sourceUrl": "https://platform.claude.com/docs/en/about-claude/pricing", "approximate": true, "stale": false, "assumptions": [ "Standard processing and short context. Regional and tool charges are excluded.", "Cache writes use the five-minute rate." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "There is no real GPU in this sandbox and no physical phone. All fps figures come from headless Chrome on software-backed GL paths and from mobile emulation with CPU throttling, so they are proxies. Real hardware will differ, and on GPUs these paths are typically much faster at canvas fill.", "The heaviest scenes do not reach a steady 60 fps on the slow default GL path: about 18-30 fps adaptive on desktop and 20-33 fps on emulated mobile. On the SwiftShader path, desktop adaptive holds 56-60 fps, but full-quality inferno is ~29 fps and emulated mobile is 26-38 fps.", "To keep frame rate, adaptive mode trades visual quality: fewer smoke and flame particles, render resolution as low as 0.5x devicePixelRatio (visibly softer image on weak GPUs), and no vignette at reduced resolution. Smoke, flames and shadows are always rendered at reduced resolution. These are soft effects by design, but fine detail at their edges is lost.", "Measurements in this environment are noisy, roughly +/-30% between identical runs (e.g. emulated-mobile cruise measured 20-31 fps across repeats), so treat the numbers as ranges.", "Under extreme load (below ~20 fps) physics is capped at 3 substeps per frame, so the game slows down rather than stuttering or exploding numerically.", "Gameplay limitations from the original build still apply: 2D top-down physics with visual 3D approximations, slowed visible shells (190 m/s), buildings damaged as single bodies, synthesized audio not verified in headless runs, fixed seeded map with no enemies, and a window.__game debug hook used by the verification scripts." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 17569, "exitCode": 0, "executionStartedAtMs": 1794278216678, "completedAtMs": 1794280279531 }, { "name": "efficiency", "status": "finished", "pid": 46572, "exitCode": 0, "executionStartedAtMs": 1794280285241, "completedAtMs": 1794283082258 } ], "skillCount": 0 }, { "key": "anthropic-claude-fable-5-1-claude", "harness": "claude", "harnessName": "Claude Code", "provider": "anthropic", "providerName": "Anthropic", "model": "claude-fable-5-1", "name": "Claude Fable 5.1", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 3083.265, "totalSeconds": 3094.436, "tokens": { "inputTokens": 168, "outputTokens": 183515, "cacheReadTokens": 10248305, "cacheWriteTokens": 227066 }, "cost": { "pricedSubtotalUsd": "14.577831250000000000000000000000", "completeTotalUsd": "14.577831250000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: Performance was measured only in headless Chrome with SwiftShader (CPU rasteriser) and CDP CPU throttling; real phones/GPUs will differ (likely better raster, possibly worse JS). No real device was available. Under heavy load the game trades visual density for frame rate: low quality means fewer/shorter-lived particles, no fire glow, fewer debris bodies and 1x DPR on phones. The first ~1.5 s of a sudden large blast on a slow device can dip (measured 24-33 fps) before the controller reacts.", "files": { "assets/index-8WX05R8e.js": "084daec4817bb84074367554bc22080d4f246537320b083e5c2889d269ad50fe", "assets/index-DJoWhx25.css": "ffb95c0f53958e7201ac399468c5defdae4fda32d0f0038a6d212d34e4dc3a0b", "index.html": "358e2fe4e5961214fd9916a6d3087d12b2dacab13b1e4e8aa3653e9821ba98dd" }, "screenshots": { "light": "anthropic-claude-fable-5-1-claude.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/anthropic/claude-fable-5-1/claude/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T03:38:12.087Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Phone portrait and landscape: trusted simultaneous touch movement and firing, observed movement/reload, and input release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "1920x1080, DPR 1, headless Chrome with forced SwiftShader software graphics; animation-frame cadence, not a typical-laptop FPS result.", "browserVersion": "156.0.8078.4", "loadMs": 556, "title": "Field of Fire – Tank Game", "desktopTextBefore": "ARMOUR\nGUN\nSCORE\n0\nENEMIES\n3\nWAVE\n1\nFIELD OF FIRE\n\nArrow keys / WASD: drive · Mouse or Q/E: aim turret · Space / click: fire\n\nDestroy enemy armour. Fuel barrels and ammo crates explode, wood and hay burn, fire spreads.\n\nPress any key or click to start", "canvases": [ { "width": 1920, "height": 1080, "visible": true }, { "width": 150, "height": 110, "visible": true } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "frameIntervalsMs": [ 16.699999999999818, 66.69999999999982, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.699999999999818, 16.699999999999818, 33.20000000000073, 16.699999999999818, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.600000000000364, 16.599999999999454, 16.800000000000182, 16.600000000000364, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.799999999999272, 16.5, 16.800000000000182, 16.600000000000364, 16.699999999999818, 16.600000000000364, 16.799999999999272, 16.600000000000364, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.699999999999818, 16.599999999999454, 16.800000000000182, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.599999999999454, 16.800000000000182, 16.600000000000364, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.599999999999454, 16.800000000000182, 16.600000000000364, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.599999999999454, 16.699999999999818, 16.800000000000182, 16.600000000000364, 16.599999999999454, 16.700000000000728, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.699999999999818, 16.599999999999454, 66.70000000000073, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.699999999999818, 16.699999999999818, 16.600000000000364, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.599999999999454, 16.699999999999818, 16.700000000000728, 16.699999999999818, 16.599999999999454, 16.700000000000728, 16.599999999999454, 16.699999999999818, 16.600000000000364, 16.800000000000182, 16.599999999999454, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.600000000000364, 16.799999999999272, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.799999999999272, 16.5, 16.80000000000109, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.80000000000109, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.80000000000109, 16.599999999998545, 16.600000000000364, 16.80000000000109, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 183.39999999999964, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.700000000000728, 583.2999999999993, 16.600000000000364, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.700000000000728, 599.8999999999996, 16.700000000000728, 16.69999999999891, 16.700000000000728, 16.600000000000364, 16.69999999999891, 566.6000000000004, 16.700000000000728, 16.69999999999891, 16.600000000000364, 16.600000000000364, 16.799999999999272, 566.6000000000004, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 483.40000000000146, 16.599999999998545, 16.700000000000728, 16.600000000000364, 16.799999999999272, 16.600000000000364, 483.2999999999993, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 16.600000000000364, 1233.300000000001, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 566.7000000000007, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.599999999998545, 16.700000000000728, 633.3999999999978, 16.600000000002183, 16.599999999998545, 16.700000000000728, 16.700000000000728, 16.700000000000728, 750, 16.5, 16.69999999999709, 16.700000000000728, 16.700000000000728, 583.2999999999993, 16.600000000002183, 16.69999999999709, 16.700000000000728, 16.700000000000728, 16.599999999998545, 550, 16.700000000000728, 16.600000000002183, 16.69999999999709, 16.600000000002183, 16.799999999999272 ], "performance": { "samples": 416, "animationFramesPerSecond": 28.5, "medianFrameMs": 16.7, "p95FrameMs": 16.8, "framesOver50Ms": 15 }, "desktopTextAfter": "ARMOUR\nGUN\nSCORE\n15\nENEMIES\n3\nWAVE\n1", "afterResetText": "ARMOUR\nGUN\nSCORE\n15\nENEMIES\n3\nWAVE\n1", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "ARMOUR\nGUN\nSCORE\n0\nENEMIES\n3\nWAVE\n1\nFIELD OF FIRE\n\nLeft stick: drive · Right stick: aim turret · FIRE button: shoot\n\nDestroy enemy armour. Fuel barrels and ammo crates explode, wood and hay burn, fire spreads.\n\nTap to start\n\nDRIVE\nAIM\nFIRE", "buttons": [] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "ARMOUR\nGUN\nSCORE\n0\nENEMIES\n3\nWAVE\n1\nFIELD OF FIRE\n\nLeft stick: drive · Right stick: aim turret · FIRE button: shoot\n\nDestroy enemy armour. Fuel barrels and ammo crates explode, wood and hay burn, fire spreads.\n\nTap to start\n\nDRIVE\nAIM\nFIRE", "buttons": [] } ], "passed": true, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "before": { "started": false, "position": { "x": 18, "y": 56 }, "reload": 0, "move": { "id": -1, "dx": 0, "dy": 0 }, "fire": { "id": -1, "down": false } }, "firing": { "started": true, "position": { "x": 18.52791499673648, "y": 56 }, "reload": 1.6, "move": { "id": 2, "dx": 0, "dy": -0.7222222222222222 }, "fire": { "id": 3, "down": true } }, "held": { "started": true, "position": { "x": 34.59021521949744, "y": 56 }, "reload": 0.6833333333333346, "move": { "id": 2, "dx": 0, "dy": -0.7222222222222222 }, "fire": { "id": 3, "down": true } }, "released": { "started": true, "position": { "x": 34.82985963723617, "y": 56 }, "reload": 0.6500000000000011, "move": { "id": -1, "dx": 0, "dy": 0 }, "fire": { "id": -1, "down": false } }, "method": "Trusted simultaneous CDP touch movement stick and fire button; existing telemetry observed only." }, { "viewport": { "width": 844, "height": 390 }, "before": { "started": false, "position": { "x": 18, "y": 56 }, "reload": 0, "move": { "id": -1, "dx": 0, "dy": 0 }, "fire": { "id": -1, "down": false } }, "firing": { "started": true, "position": { "x": 18.532733848439786, "y": 56 }, "reload": 1.6, "move": { "id": 2, "dx": 0, "dy": -0.7291666666666666 }, "fire": { "id": 3, "down": true } }, "held": { "started": true, "position": { "x": 34.29963376076913, "y": 56 }, "reload": 0.7333333333333347, "move": { "id": 2, "dx": 0, "dy": -0.7291666666666666 }, "fire": { "id": 3, "down": true } }, "released": { "started": true, "position": { "x": 34.65940409601423, "y": 56 }, "reload": 0.6833333333333346, "move": { "id": -1, "dx": 0, "dy": 0 }, "fire": { "id": -1, "down": false } }, "method": "Trusted simultaneous CDP touch movement stick and fire button; existing telemetry observed only." } ] }, "recordedRun": { "createdAt": "2026-11-10T02:36:56.735Z", "executionStartedAt": "2026-11-10T02:37:00.303Z", "completedAt": "2026-11-10T03:28:31.171Z" }, "actualModels": [ { "model": "claude-fable-5-1", "requests": 81, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "claude", "model": "claude-fable-5-1", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "claude", "model": "claude-fable-5-1", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "c1cbb5f1-a9c0-4201-ab10-a9ec31712c76", "bucket": 0, "model": "claude-fable-5-1", "provider": "anthropic", "usageSource": "boat_workspace_delta", "priceSnapshotId": "97285416-4b41-4605-a9c1-3db4d3a11e12", "effectiveAt": "2026-11-06T01:02:42.711Z", "effectiveUntil": null, "usageTime": "2026-11-10T03:28:31.171Z", "costUsd": "14.577831250000000000000000000000", "categories": { "inputTokens": "0.001680000000000000000000000000", "outputTokens": "9.175750000000000000000000000000", "cacheReadTokens": "2.562076250000000000000000000000", "cacheWriteTokens": "2.838325000000000000000000000000" }, "tokens": { "inputTokens": 168, "outputTokens": 183515, "cacheReadTokens": 10248305, "cacheWriteTokens": 227066 }, "sourceUrl": "https://platform.claude.com/docs/en/about-claude/pricing", "approximate": true, "stale": false, "assumptions": [ "Standard processing and short context. Regional and tool charges are excluded.", "Cache writes use the five-minute rate." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "Performance was measured only in headless Chrome with SwiftShader (CPU rasteriser) and CDP CPU throttling; real phones/GPUs will differ (likely better raster, possibly worse JS). No real device was available.", "Under heavy load the game trades visual density for frame rate: low quality means fewer/shorter-lived particles, no fire glow, fewer debris bodies and 1x DPR on phones. The first ~1.5 s of a sudden large blast on a slow device can dip (measured 24-33 fps) before the controller reacts.", "On devices that cap rAF at 30 Hz (power-saving) the frame-interval heuristic will settle on low quality even if headroom exists.", "Headless harness used player teleportation and scripted ignition to build worst-case scenes; this is not extended human playtesting.", "Graphics are procedural vector drawings (no textures); 'realistic' is interpreted as physically plausible behaviour rather than photorealism.", "Physics is a 2D top-down approximation: flat shell ballistics with a maximum range, no terrain elevation.", "Fire model is heat-accumulation based and hand-tuned; enemy AI is simple (patrol, feeler-ray avoidance, lead aiming); balance only lightly tuned.", "Audio is synthesised WebAudio and starts after the first user interaction; it was not audibly verified.", "The preview is served by a minimal Node static server (no compression)." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 18450, "exitCode": 0, "executionStartedAtMs": 1794278220303, "completedAtMs": 1794280013964 }, { "name": "efficiency", "status": "finished", "pid": 42897, "exitCode": 0, "executionStartedAtMs": 1794280021567, "completedAtMs": 1794281311171 } ], "skillCount": 0 }, { "key": "anthropic-claude-fable-5-2-claude", "harness": "claude", "harnessName": "Claude Code", "provider": "anthropic", "providerName": "Anthropic", "model": "claude-fable-5-2", "name": "Claude Fable 5.2", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 9335.881, "totalSeconds": 9345.391, "tokens": { "inputTokens": 290, "outputTokens": 446602, "cacheReadTokens": 47954387, "cacheWriteTokens": 1158031 }, "cost": { "pricedSubtotalUsd": "39.037587400000000000000000000000", "completeTotalUsd": "39.037587400000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: Frame rates on real GPUs and phones are still unmeasured. The software-rendering figures show the optimised build is roughly 1.3-2.8x faster than the baseline in the same environment, not that it reaches 60 fps on any particular device. Under software GL the heavy scenes remain at 10-16 fps at 1440x900. A resolution test there (rendering 44% fewer pixels) did not make frames 10% faster, so that mode appears limited by per-draw overhead rather than pixel fill; heavy scenes still issue about 1,600-1,750 canvas draws per frame on desktop and about 920-950 on touch devices.", "files": { "assets/index-CaEJNsXD.js": "0f60751fb2b0fd828dd446b42c50597c5ec5877cefd9edf895744f379b347bf7", "index.html": "57d15b782ad8e4ff048363386ad7d672899b7e38291dfd3934f57db92453ae18" }, "screenshots": { "light": "anthropic-claude-fable-5-2-claude.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/anthropic/claude-fable-5-2/claude/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T05:13:55.829Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Desktop keyboard movement, steering and firing observed through position/shot counts; key release clears controls", "Phone portrait and landscape: trusted simultaneous touch movement, steering and firing; observed movement/shots and input release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "No independent gameplay FPS result. Initial generic cadence sample was from menu; actual playing keyboard and touch actions verified separately.", "browserVersion": "156.0.8078.4", "loadMs": 560, "title": "Ironfield — Tank Combat", "desktopTextBefore": "IRONFIELD\nARMOURED RECON · SINGLE PLAYER\n\nFive enemy tanks hold the valley and its fuel depot. Hunt them down. Everything out there reacts: trees topple and burn, fuel cooks off, dry grass and wheat carry fire downwind.\n\nKEYBOARD\n↑ ↓ drive forward / reverse\n← → steer the hull\nSpace fire main gun\nQ E traverse turret · C centre · click to mouse-aim\nP pause · M sound\nSTART", "canvases": [ { "width": 1920, "height": 1080, "visible": true }, { "width": 188, "height": 141, "visible": false } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "desktopTextAfter": "IRONFIELD\nARMOURED RECON · SINGLE PLAYER\n\nFive enemy tanks hold the valley and its fuel depot. Hunt them down. Everything out there reacts: trees topple and burn, fuel cooks off, dry grass and wheat carry fire downwind.\n\nKEYBOARD\n↑ ↓ drive forward / reverse\n← → steer the hull\nSpace fire main gun\nQ E traverse turret · C centre · click to mouse-aim\nP pause · M sound\nSTART", "afterResetText": "IRONFIELD\nARMOURED RECON · SINGLE PLAYER\n\nFive enemy tanks hold the valley and its fuel depot. Hunt them down. Everything out there reacts: trees topple and burn, fuel cooks off, dry grass and wheat carry fire downwind.\n\nKEYBOARD\n↑ ↓ drive forward / reverse\n← → steer the hull\nSpace fire main gun\nQ E traverse turret · C centre · click to mouse-aim\nP pause · M sound\nSTART", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "HULL\n100\nGUN\nREADY\nENEMY ARMOUR0/5\nSCORE0\nTURRET\nFWD\nFIRE", "buttons": [ { "text": "", "aria": "Pause", "title": null, "width": 40, "height": 36 }, { "text": "", "aria": "Toggle sound", "title": null, "width": 40, "height": 36 }, { "text": "TURRETFWD", "aria": null, "title": null, "width": 64, "height": 64 }, { "text": "FIRE", "aria": null, "title": null, "width": 104, "height": 104 } ] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "HULL\n100\nGUN\nREADY\nENEMY ARMOUR0/5\nSCORE0\nWIND\nTURRET\nFWD\nFIRE", "buttons": [ { "text": "", "aria": "Pause", "title": null, "width": 40, "height": 36 }, { "text": "", "aria": "Toggle sound", "title": null, "width": 40, "height": 36 }, { "text": "TURRETFWD", "aria": null, "title": null, "width": 64, "height": 64 }, { "text": "FIRE", "aria": null, "title": null, "width": 104, "height": 104 } ] } ], "passed": true, "desktopControls": { "before": { "mode": "playing", "position": { "x": 24, "y": 153 }, "shots": 0, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 0 }, "held": { "mode": "playing", "position": { "x": 25.317816737786025, "y": 160.56226338456176 }, "shots": 2, "control": { "throttle": 1, "steer": 1, "fire": true, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 5.935475882374666 }, "released": { "mode": "playing", "position": { "x": 23.750524922451653, "y": 161.89554654133485 }, "shots": 2, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 6.236031892607915 }, "method": "Trusted keyboard arrows/Space; existing telemetry observed only." }, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "before": { "mode": "playing", "position": { "x": 24, "y": 153 }, "shots": 0, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 0 }, "held": { "mode": "playing", "position": { "x": 32.021628354409906, "y": 156.14379212742944 }, "shots": 2, "control": { "throttle": 1, "steer": 0.3666495864188765, "fire": true, "aim": null, "turretRel": 0 }, "input": { "jx": 0.3055413220157304, "jy": -0.8917438367789096, "fire": true }, "speed": 6.478950416574289 }, "released": { "mode": "playing", "position": { "x": 32.78835239029943, "y": 156.78456049175367 }, "shots": 2, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 6.777303028945917 }, "method": "Trusted simultaneous CDP touch floating stick movement/steering and fire; read-only existing telemetry confirms movement, shots and input release." }, { "viewport": { "width": 844, "height": 390 }, "before": { "mode": "playing", "position": { "x": 24, "y": 153 }, "shots": 0, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 0 }, "held": { "mode": "playing", "position": { "x": 31.762844729697086, "y": 156.0129492427519 }, "shots": 2, "control": { "throttle": 1, "steer": 0.3666495864188765, "fire": true, "aim": null, "turretRel": 0 }, "input": { "jx": 0.3055413220157304, "jy": -0.8917438367789096, "fire": true }, "speed": 6.354576697127826 }, "released": { "mode": "playing", "position": { "x": 32.61009050026765, "y": 156.70799749963928 }, "shots": 2, "control": { "throttle": 0, "steer": 0, "fire": false, "aim": null, "turretRel": 0 }, "input": { "jx": 0, "jy": 0, "fire": false }, "speed": 6.725293092734643 }, "method": "Trusted simultaneous CDP touch floating stick movement/steering and fire; read-only existing telemetry confirms movement, shots and input release." } ], "menuFrameSample": { "scope": "Initial generic check sampled the menu before START. These samples are not gameplay performance.", "statistics": { "samples": 904, "animationFramesPerSecond": 59.9, "medianFrameMs": 16.7, "p95FrameMs": 16.7, "framesOver50Ms": 0 } } }, "recordedRun": { "createdAt": "2026-11-10T02:37:00.334Z", "executionStartedAt": "2026-11-10T02:37:05.299Z", "completedAt": "2026-11-10T05:12:45.725Z" }, "actualModels": [ { "model": "claude-fable-5-2", "requests": 139, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "claude", "model": "claude-fable-5-2", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "claude", "model": "claude-fable-5-2", "reasoning": "high", "version": "2.1.322 (Claude Code)", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "59dee348-5380-4e33-93d3-81de5eb410a0", "bucket": 0, "model": "claude-fable-5-2", "provider": "anthropic", "usageSource": "boat_workspace_delta", "priceSnapshotId": "e793560b-a4ad-4441-a498-29218bc970f5", "effectiveAt": "2026-11-10T01:05:20.988Z", "effectiveUntil": null, "usageTime": "2026-11-10T05:12:45.725Z", "costUsd": "39.037587400000000000000000000000", "categories": { "inputTokens": "0.002320000000000000000000000000", "outputTokens": "17.864080000000000000000000000000", "cacheReadTokens": "9.590877400000000000000000000000", "cacheWriteTokens": "11.580310000000000000000000000000" }, "tokens": { "inputTokens": 290, "outputTokens": 446602, "cacheReadTokens": 47954387, "cacheWriteTokens": 1158031 }, "sourceUrl": "https://platform.claude.com/docs/en/about-claude/pricing", "approximate": true, "stale": false, "assumptions": [ "Standard processing and short context. Regional and tool charges are excluded.", "Cache writes use the five-minute rate." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "Frame rates on real GPUs and phones are still unmeasured. The software-rendering figures show the optimised build is roughly 1.3-2.8x faster than the baseline in the same environment, not that it reaches 60 fps on any particular device.", "Under software GL the heavy scenes remain at 10-16 fps at 1440x900. A resolution test there (rendering 44% fewer pixels) did not make frames 10% faster, so that mode appears limited by per-draw overhead rather than pixel fill; heavy scenes still issue about 1,600-1,750 canvas draws per frame on desktop and about 920-950 on touch devices.", "Main-thread render time in heavy scenes did not go down (about 3 ms on the test CPU, 6-7 ms at 4x throttle) because compositing the low-res effect canvas is synchronous in the software modes measured; whether that cost exists on a hardware GPU is unknown.", "Tree canopies are still drawn individually every frame (about 150-250 sprites in view). Caching them in screen-resolution chunks was considered and not done.", "Visual trade-offs made for speed: smoke and fire glow are rendered at 1/3 and 1/6 resolution (softer); flames are fewer and larger; particle caps are lower than before, more so on touch devices; fire glow now draws over canopies and vehicles; light-coloured debris lying in a baked shadow is not darkened; debris from off-screen explosions is no longer spawned or stamped.", "Steady-state canvas memory is about 5 MB higher on desktop and 3 MB higher on touch devices than the baseline, in exchange for much lower peaks.", "Adaptive quality reduces effect density (to 75%, then 50%) on devices that stay below about 40 fps after a resolution drop fails to help; in the software-GL test it settled at 75% effects and full resolution with little fps gain. Append ?hq to disable it.", "Tested only in Chrome 151 (headless, desktop and mobile emulation). Firefox, Safari and iOS were not tested; the shadow removal relies on the color-dodge blend mode behaving as specified.", "Unchanged from before: physics is game-scaled (straight-line shells at 150 m/s, radial blast damage), keyboard-only play keeps the gun hull-locked unless Q/E are used, enemy AI is simple, one map, no save, window.__tank debug handle left in the build, preview is a detached process rather than a supervised service." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 18380, "exitCode": 0, "executionStartedAtMs": 1794278225299, "completedAtMs": 1794281382753 }, { "name": "efficiency", "status": "finished", "pid": 60827, "exitCode": 0, "executionStartedAtMs": 1794281387298, "completedAtMs": 1794287565725 } ], "skillCount": 0 }, { "key": "deepseek-deepseek-deepseek-v4.1-flash-deepseek", "harness": "deepseek", "harnessName": "DeepSeek Harness", "provider": "deepseek", "providerName": "DeepSeek", "model": "deepseek-v4.1-flash", "name": "DeepSeek V4.1 Flash", "backend": "Boat", "reasoning": "high", "status": "complete", "executionSeconds": 6285.555, "totalSeconds": 6399.302, "tokens": { "inputTokens": 503112, "outputTokens": 352238, "cacheReadTokens": 54846592, "cacheWriteTokens": 0 }, "cost": { "pricedSubtotalUsd": "1.212616314000000000000000000000", "completeTotalUsd": "1.212616314000000000000000000000", "partial": false, "provisional": false, "approximate": true }, "originalPreview": "Published", "publicationCheck": "Browser checks passed", "notes": "Agent-reported limits: All browser verification ran on headless Chrome with SwiftShader (software rasteriser): wall-clock frame time there is dominated by a fixed present/composite cost (a 480x300 canvas still costs ~130 ms/frame while render submission is 0.6 ms), so the frame-time numbers are not a proxy for real GPU hardware. The hardware-independent signals reported are per-subsystem CPU ms, per-call micro-benchmarks, draw calls / triangles / point sprites and heap after GC. GPU raster/fill cost could not be isolated in this environment; the fill-rate work (fewer and smaller point sprites, ~29% fewer triangles, per-tier pixel-ratio cap, shadows on a cadence) is justified by counters and the resolution-scaling result, not by a measured GPU frame time.", "files": { "assets/index-CqDb6SSw.css": "853d6d76f2ae3758860d3adac3d357468534c3582536a036dd36610d0c7d81cc", "assets/index-D5_92lWo.js": "fe19df5d4147cc0253b45ad5a47e5e37e8ce410987e3ace183ac00ffea171d26", "index.html": "eb729a9451a6e1b29ff8390dfb69b66b1249d55703ecf0aedad5a3e38251b440" }, "screenshots": { "light": "deepseek-deepseek-deepseek-v4.1-flash-deepseek.webp" }, "appUrl": "https://benchmarks.zzz.link/benchmark/2026-11-10/tank-game/deepseek/deepseek-v4.1-flash/deepseek/", "publicationTransforms": [ "Static artifact copied without application-code changes. Published title gains provider/model prefix." ], "publication": { "checkedAt": "2026-11-10T04:25:56.727Z", "checks": [ "Loads from a nested URL path and shows a game canvas", "Arrow-key movement and Space shooting exercised", "Phone portrait and landscape fit without horizontal overflow", "Phone portrait and landscape: trusted simultaneous touch movement and firing, observed movement/reload, and input release passed" ], "errors": [], "failedRequests": [], "performanceMethod": "1920x1080, DPR 1, headless Chrome with forced SwiftShader software graphics; animation-frame cadence, not a typical-laptop FPS result.", "browserVersion": "156.0.8078.4", "loadMs": 586, "title": "Steel Fields — Aerial Tank Combat", "desktopTextBefore": "HULL\nENGINE\nSCORE\n4\nKILLS\n0\nSHELLS\n∞\nYOU\nHOSTILE\nFIRE\nARROWS drive · MOUSE aim · SPACE / CLICK fire · WHEEL zoom · V view · R horn · M mute\nGRAPHICS AUTO-ADJUSTED\nBATTLE START\nSTEEL FIELDS\n\nTop-down aerial tank combat · destructible terrain & spreading fire\n\nKEYBOARD\n↑↓ throttle / reverse\n←→ steer hull\nSpace fire main gun\nMouse aims the turret\nW A S D also drive\nR horn · V change view\nTOUCH\nLeft stick drives & steers\nRight stick aims the turret\nFIRE button shoots, HORN honks\nPinch to zoom\n\nWooden crates and fences burn, fuel barrels detonate, trees topple and grass fires spread with the wind. Hunt the hostile armour.\n\nDEPLOY\nBUILDING BATTLEFIELD…", "canvases": [ { "width": 1920, "height": 1080, "visible": true }, { "width": 176, "height": 176, "visible": true } ], "renderer": "ANGLE (Google, Vulkan 1.3.0 (SwiftShader Device (Subzero) (0x0000C0DE)), SwiftShader driver)", "frameIntervalsMs": [ 1816.6000000000004, 16.69999999999891, 433.3000000000011, 116.59999999999854, 516.7000000000007, 216.70000000000073, 16.599999999998545, 533.2999999999993, 516.7000000000007, 583.2999999999993, 1166.6000000000022, 583.2999999999993, 450, 516.7000000000007, 83.29999999999927, 16.700000000000728, 483.2999999999993, 1049.9000000000015, 833.3999999999978, 49.900000000001455, 50.099999999998545, 766.6000000000022, 366.59999999999854, 33.29999999999927, 583.4000000000015, 2199.899999999998, 150, 16.700000000000728, 33.20000000000073, 416.7000000000007, 616.7000000000007, 1249.8999999999978, 400, 50, 366.7000000000007, 50, 833.2999999999993, 499.90000000000146 ], "performance": { "samples": 38, "animationFramesPerSecond": 2, "medianFrameMs": 433.3, "p95FrameMs": 1816.6, "framesOver50Ms": 29 }, "desktopTextAfter": "HULL\nENGINE\nSCORE\n4\nKILLS\n0\nSHELLS\n∞\nYOU\nHOSTILE\nFIRE\nARROWS drive · MOUSE aim · SPACE / CLICK fire · WHEEL zoom · V view · R horn · M mute\nARROWS DRIVE · MOUSE AIMS · SPACE FIRES\nBATTLE START\nSTEEL FIELDS\n\nTop-down aerial tank combat · destructible terrain & spreading fire\n\nKEYBOARD\n↑↓ throttle / reverse\n←→ steer hull\nSpace fire main gun\nMouse aims the turret\nW A S D also drive\nR horn · V change view\nTOUCH\nLeft stick drives & steers\nRight stick aims the turret\nFIRE button shoots, HORN honks\nPinch to zoom\n\nWooden crates and fences burn, fuel barrels detonate, trees topple and grass fires spread with the wind. Hunt the hostile armour.\n\nDEPLOY\nBUILDING BATTLEFIELD…", "afterResetText": "HULL\nENGINE\nSCORE\n6\nKILLS\n0\nSHELLS\n∞\nYOU\nHOSTILE\nFIRE\nARROWS drive · MOUSE aim · SPACE / CLICK fire · WHEEL zoom · V view · R horn · M mute\nHORN\nBATTLE START\nSTEEL FIELDS\n\nTop-down aerial tank combat · destructible terrain & spreading fire\n\nKEYBOARD\n↑↓ throttle / reverse\n←→ steer hull\nSpace fire main gun\nMouse aims the turret\nW A S D also drive\nR horn · V change view\nTOUCH\nLeft stick drives & steers\nRight stick aims the turret\nFIRE button shoots, HORN honks\nPinch to zoom\n\nWooden crates and fences burn, fuel barrels detonate, trees topple and grass fires spread with the wind. Hunt the hostile armour.\n\nDEPLOY\nBUILDING BATTLEFIELD…", "mobile": [ { "viewport": { "width": 390, "height": 844 }, "horizontalOverflow": false, "text": "HULL\nENGINE\nSCORE\n0\nKILLS\n0\nSHELLS\n∞\nYOU\nHOSTILE\nFIRE\nARROWS DRIVE · MOUSE AIMS · SPACE FIRES\nBATTLE START\nDRIVE\nAIM\nFIRE\nVIEW\nHORN\nSTEEL FIELDS\n\nTop-down aerial tank combat · destructible terrain & spreading fire\n\nKEYBOARD\n↑↓ throttle / reverse\n←→ steer hull\nSpace fire main gun\nMouse aims the turret\nW A S D also drive\nR horn · V change view\nTOUCH\nLeft stick drives & steers\nRight stick aims the turret\nFIRE button shoots, HORN honks\nPinch to zoom\n\nWooden crates and fences burn, fuel barrels detonate, trees topple and grass fires spread with the wind. Hunt the hostile armour.\n\nDEPLOY\nBUILDING BATTLEFIELD…", "buttons": [ { "text": "FIRE", "aria": null, "title": null, "width": 84, "height": 84 }, { "text": "VIEW", "aria": null, "title": null, "width": 58, "height": 58 }, { "text": "HORN", "aria": null, "title": null, "width": 58, "height": 58 }, { "text": "DEPLOY", "aria": null, "title": null, "width": 217, "height": 50 } ] }, { "viewport": { "width": 844, "height": 390 }, "horizontalOverflow": false, "text": "HULL\nENGINE\nSCORE\n0\nKILLS\n0\nSHELLS\n∞\nYOU\nHOSTILE\nFIRE\nARROWS DRIVE · MOUSE AIMS · SPACE FIRES\nBATTLE START\nDRIVE\nAIM\nFIRE\nVIEW\nHORN\nSTEEL FIELDS\n\nTop-down aerial tank combat · destructible terrain & spreading fire\n\nKEYBOARD\n↑↓ throttle / reverse\n←→ steer hull\nSpace fire main gun\nMouse aims the turret\nW A S D also drive\nR horn · V change view\nTOUCH\nLeft stick drives & steers\nRight stick aims the turret\nFIRE button shoots, HORN honks\nPinch to zoom\n\nWooden crates and fences burn, fuel barrels detonate, trees topple and grass fires spread with the wind. Hunt the hostile armour.\n\nDEPLOY\nBUILDING BATTLEFIELD…", "buttons": [ { "text": "FIRE", "aria": null, "title": null, "width": 84, "height": 84 }, { "text": "VIEW", "aria": null, "title": null, "width": 58, "height": 58 }, { "text": "HORN", "aria": null, "title": null, "width": 58, "height": 58 }, { "text": "DEPLOY", "aria": null, "title": null, "width": 217, "height": 50 } ] } ], "passed": true, "mobileTouch": [ { "viewport": { "width": 390, "height": 844 }, "before": { "started": true, "position": { "x": -10.271713678071649, "y": 71.49607410448972 }, "reload": 0, "shots": 0, "move": { "x": 0, "y": 0 }, "fire": false }, "firing": { "started": true, "position": { "x": -9.70143300291127, "y": 67.99039645245702 }, "reload": 0.8266400000000007, "shots": 3, "move": { "x": 0, "y": -1 }, "fire": true }, "held": { "started": true, "position": { "x": -6.144543089978819, "y": 49.16637747370246 }, "reload": 0.11680000000000101, "shots": 4, "move": { "x": 0, "y": -1 }, "fire": true }, "released": { "started": true, "position": { "x": -5.992928133308132, "y": 49.16191470422717 }, "reload": 0.8917499999999994, "shots": 5, "move": { "x": 0, "y": 0 }, "fire": false }, "method": "Trusted simultaneous CDP touch movement stick and fire button; existing telemetry observed only." }, { "viewport": { "width": 844, "height": 390 }, "before": { "started": true, "position": { "x": 89.16384720085617, "y": -50.59160159550574 }, "reload": 0, "shots": 0, "move": { "x": 0, "y": 0 }, "fire": false }, "firing": { "started": true, "position": { "x": 87.39005359056358, "y": -49.585150410840896 }, "reload": 0.9966800000000005, "shots": 2, "move": { "x": 0, "y": -1 }, "fire": true }, "held": { "started": true, "position": { "x": 80.96579808063869, "y": -45.94002533482546 }, "reload": 0.01338000000000139, "shots": 3, "move": { "x": 0, "y": -1 }, "fire": true }, "released": { "started": true, "position": { "x": 72.31873080306612, "y": -41.03367599694847 }, "reload": 0.4499999999999992, "shots": 4, "move": { "x": 0, "y": 0 }, "fire": false }, "method": "Trusted simultaneous CDP touch movement stick and fire button; existing telemetry observed only." } ] }, "recordedRun": { "createdAt": "2026-11-10T02:37:05.320Z", "executionStartedAt": "2026-11-10T02:38:53.299Z", "completedAt": "2026-11-10T04:23:44.622Z" }, "actualModels": [ { "model": "deepseek/deepseek-v4.1-flash", "requests": 248, "errors": 0 } ], "extraModels": [], "nativeConfigurations": [ { "phase": "initial", "harness": "deepseek", "model": "deepseek/deepseek-v4.1-flash", "reasoning": "high", "version": "@deepseek-ai/dsh@0.2.1-rc.5", "standardSpeed": true, "noFallbackConfigured": true }, { "phase": "efficiency", "harness": "deepseek", "model": "deepseek/deepseek-v4.1-flash", "reasoning": "high", "version": "@deepseek-ai/dsh@0.2.1-rc.5", "standardSpeed": true, "noFallbackConfigured": true } ], "usageDetails": [ { "measurementId": "40591d38-dd08-41f9-a265-35cb032da802", "bucket": 0, "model": "deepseek/deepseek-v4.1-flash", "provider": "openrouter", "usageSource": "boat_workspace_delta", "priceSnapshotId": "6a75d603-fdd1-4de6-ac30-d6b55166c6b0", "effectiveAt": "2026-11-07T01:03:40.544Z", "effectiveUntil": null, "usageTime": "2026-11-10T04:23:44.622Z", "costUsd": "1.212616314000000000000000000000", "categories": { "inputTokens": "0.025658712000000000000000000000", "outputTokens": "0.199718946000000000000000000000", "cacheReadTokens": "0.987238656000000000000000000000", "cacheWriteTokens": "0" }, "tokens": { "inputTokens": 503112, "outputTokens": 352238, "cacheReadTokens": 54846592, "cacheWriteTokens": 0 }, "sourceUrl": "https://openrouter.ai/api/v1/models", "approximate": true, "stale": false, "assumptions": [ "OpenRouter listed token rates. Routing, context tiers, tool charges, and discounts can change billing." ] } ], "taskHash": "9adb61fed924e5c02dc8b45b25ca0d9b67f542ff83ac211bd967fdf9bdb01fb8", "reportedLimitations": [ "All browser verification ran on headless Chrome with SwiftShader (software rasteriser): wall-clock frame time there is dominated by a fixed present/composite cost (a 480x300 canvas still costs ~130 ms/frame while render submission is 0.6 ms), so the frame-time numbers are not a proxy for real GPU hardware. The hardware-independent signals reported are per-subsystem CPU ms, per-call micro-benchmarks, draw calls / triangles / point sprites and heap after GC.", "GPU raster/fill cost could not be isolated in this environment; the fill-rate work (fewer and smaller point sprites, ~29% fewer triangles, per-tier pixel-ratio cap, shadows on a cadence) is justified by counters and the resolution-scaling result, not by a measured GPU frame time.", "The agent cannot view images, so visual quality was checked programmatically (lit, green-dominant, non-uniform framebuffer; ground median luminance ~160/255; draw-call and triangle counts) rather than by eye. The optimisations do change detail: terrain is 140 segments instead of 200, ground clutter is 2,120 instances instead of 3,020, point sprites are capped at 120 px, and tier 0 hides ground clutter entirely.", "Fixed quality tiers were used for like-for-like comparison (tier 2, full effects); the adaptive governor was measured separately and settles at tier 0 in the software-rendered harness, which is intended behaviour but means the adaptive settings themselves were not exercised on real GPU hardware.", "Mobile controls were exercised with synthetic PointerEvents at 390x844 in headless Chrome; real multi-touch hardware, iOS/Android browser quirks and real phone thermals were not tested.", "Physics is a purpose-built lightweight engine (vertical circle/AABB colliders, no rotational rigid-body dynamics), so barrels and crates slide and are thrown by blasts but never tip over or fracture; destruction is material-profile based rather than mesh-accurate fracture.", "Ground fire uses a coarse 3 m fuel grid with a hard cap on simultaneous cells (170/130/90 by tier) and stochastic spread biased by wind and uphill slope; it is a believable approximation, not a fluid/thermal simulation. The cap means a very large wildfire stops taking new cells instead of growing without bound.", "Slow-burn systems (burning props, fire spread, ground fire, particles, decals) now run once per frame with the frame delta instead of once per physics sub-step; at 60 fps this is identical, but at very low frame rates fire advances in larger, less granular steps.", "Audio is fully procedural WebAudio and only starts after a user gesture; it was verified by code path and absence of errors, not by listening.", "vite preview applies SPA fallback for unknown paths, so the preview URL is not itself proof of arbitrary-path hosting; the arbitrary-prefix requirement was verified separately by serving dist with a plain static server at /deep/prefix/play/ and booting the game there.", "The published URL is a tunnel to the local preview process; it stays alive only while that workspace-owned preview job runs.", "The production bundle still contains the window.__tankGame debug hooks (stats/fireAt/detonateAt) used by the verification harness; they are not exposed in the game UI.", "Single-player only (no networking), no save/persistence, and the player receives 3.5 s of damage immunity after respawning.", "Performance numbers come from a 4 vCPU container and include run-to-run variance of roughly +/-6% on frame times; micro-benchmarks of sub-0.1 ms operations are at Chrome's timer resolution limit and are reported as medians with p95." ], "followup": { "requestedAt": "2026-11-10T02:42:43.193Z", "prompt": "Make the tank game efficient so that it runs smoothly on desktop and mobile, including during explosions, spreading fire, and object destruction. Profile performance where possible and optimize physics, rendering, effects, and memory use while preserving the requested gameplay, convincing physical behavior, and visual quality. Test demanding scenes and report measured performance and limitations honestly. Rebuild the final app, update the retained public preview, and update benchmark-result.json. Continue the same task and session; do not delegate or change model or reasoning settings." }, "phases": [ { "name": "initial", "status": "finished", "pid": 17889, "exitCode": 0, "executionStartedAtMs": 1794278333299, "completedAtMs": 1794282557326 }, { "name": "efficiency", "status": "finished", "pid": 73834, "exitCode": 0, "executionStartedAtMs": 1794282563094, "completedAtMs": 1794284624622 } ], "skillCount": 0 } ] }