Measured harness ledgerPublic result
DeepSeek V4 FlashExplorable space-flight game — DeepSeek V4 Flash Highest
Build a responsive browser space-flight game with flight controls, a coherent star-system environment, lighting, assets, and a playable game loop.
Highest reasoningHeadline result
- Workflow cost
- $0.36
- Wall-clock
- 1561.4s wall-clock
- Processed tokens
- 15.44M processed
- Record state
- PASS
Public summary
DeepSeek V4 Flash Highest PASS ledger: 1561.4s wall-clock, 15.44M processed, and $0.36 Provider-recorded OpenRouter cost.
Run identity and stack
- Result ID: space-flight-game-threejs-deepseek-v4-flash-vision-exp-openrouter-2026-09-15
- Technical model: deepseek-v4-flash-vision-exp-openrouter
- Provider: OpenRouter
- Client: OpenCode 1.18.23
- Variant: max
- Stack: OpenRouter
- Stack: OpenCode 1.18.23
- Stack: OpenCode → OpenRouter
- Stack: Technical model/configuration: deepseek-v4-flash-vision-exp-openrouter
- Stack: Requested variant: max
- Stack: Three.js / Vite game
- Stack: Generated spaceship assets
- Stack: Harness v1 space-flight prompt
Primary artifact integrity
- Kind: manifest-verified-artifact
- Path: artifacts/space-flight-game-threejs-deepseek-v4-flash-vision-exp-openrouter-2026-09-15/source/src/main.js
- SHA-256: 87821f84e1ba0bf3025c49b8054dd853c76aead2026a087be6baa81ad1a48e0d
Validation evidence
- Result: PASS
- Path: artifacts/space-flight-game-threejs-deepseek-v4-flash-vision-exp-openrouter-2026-09-15/verification/summary.public.json
- SHA-256: 6d8e0747b8b56834e7ef1b5cd92d6c31e139433556f308b36aeac22751bd9307
Recorded caveats
- No blind visual quality score is attached.
- Wall-clock time includes model inference and tool execution.
- The metrics operator prompt expected the OpenCode Go stack and therefore labeled the observed OpenRouter stack PARTIAL; this run intentionally used OpenCode with OpenRouter as requested.
- The session-level variant became default on the later metrics turn, but all 150 assistant/API messages in the bounded benchmark window record variant max.
Public result only
This result keeps its public summary and evidence, but it does not currently have a matching Builder test with prompts, projects, Harness workflows or skills.