Measured harness ledgerPublic result
Grok 4.7CARVE — Grok 4.7 xhigh
Build a playable Babylon.js WebGPU snowboarding game with the supplied rider asset, carved terrain, tricks, obstacles, audio, and a complete downhill run.
xhigh reasoningHeadline result
- Workflow cost
- ≥$10.22
- Wall-clock
- Not recorded
- Processed tokens
- Not recorded
- Record state
- partial_asset_quality_and_metrics_prompt_unavailable
Public summary
Grok 4.7 xhigh partial_asset_quality_and_metrics_prompt_unavailable ledger: Not recorded wall-clock, Not recorded, and ≥$10.22 Provider-recorded usage estimate, not an itemized subscription cash charge (lower bound).
Run identity and stack
- Result ID: carve-webgpu-snowboarding-grok-4.7-xhigh
- Technical model: Grok 4.7
- Provider: Provider not separately recorded
- Client: Grok Build CLI 1.0.40
- Stack: Grok Build CLI 1.0.40
- Stack: Technical model/configuration: Grok 4.7
- Stack: Babylon.js WebGPU
- Stack: Supplied snowboarder fixture
- Stack: Harness v1 CARVE prompt
Cost basis
- Provider-recorded usage is partial, so this figure is a lower bound.
Primary artifact integrity
- Kind: manifest-verified-artifact
- Path: artifacts/carve-webgpu-snowboarding-grok-4.7-xhigh/source/src/main.js
- SHA-256: f2072515f9cfc1ea1713e931d5371b85af819eae114dce1c62772756739c43d6
Recorded caveats
- The final boulder asset and required clay-detail images failed the independent asset-quality gate; this is not a fully passing CARVE result.
- The source rider GLB was a symlink to the read-only supplied fixture during generation; the archive materializes identical bytes as a regular file.
- The exact post-generation operator prompt was attempted but never executed; cost and timing are sanitized generation-ledger fallbacks.
- Prompt-to-completion wall (4,712s) and Grok model-turn elapsed (3,811.527s) are distinct.
- The final gameplay capture had one favicon-only 404.
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.