Measured harness ledgerPublic result
Grok 4.6Off-road Mud Game — Grok 4.6 xhigh · Attempt 2
Build a one-request procedural off-road driving game with vehicle physics, independent suspension, streamed terrain, mud feedback, and multiple cameras.
xhigh reasoningHeadline result
- Workflow cost
- Not recorded
- Wall-clock
- Not recorded
- Processed tokens
- Not recorded
- Record state
- partial_runtime_functional_metrics_fail_visual_fail
Public summary
Grok 4.6 xhigh partial_runtime_functional_metrics_fail_visual_fail ledger: Not recorded wall-clock, Not recorded, and Not recorded.
Run identity and stack
- Result ID: off-road-driving-game-threejs-grok-4.6-xhigh-cli-attempt-2
- Technical model: Grok 4.6
- Provider: Grok CLI subscription
- Client: Grok Build CLI
- Attempt: 2
- Stack: Grok CLI subscription
- Stack: Grok Build CLI
- Stack: Technical model/configuration: Grok 4.6
- Stack: Attempt 2
- Stack: Three.js vehicle workflow
- Stack: Harness v1 off-road prompt
- Stack: Requested tool profile: bypassPermissions + alwaysApprove
Cost basis
- The canonical metrics extractor failed closed before isolating a usage window; no cost was inferred.
Primary artifact integrity
- Kind: interactive-threejs-off-road-driving-game
- Path: artifacts/off-road-driving-game-threejs-grok-4.6-xhigh-cli-attempt-2/source/src/main.js
- SHA-256: 7b7034dc07e73d2c4d2f6545bb1820c00de3acee8a48744f8c414f42e75c8301
Recorded caveats
- This fresh CLI run is distinct from the earlier Grok Build attempt-2 record.
- The model-authored receipt reports a long drive, but independent acceptance is partial because the browser replay exposed runtime and visual defects.
- Headless/software versus Metal/GPU FPS readings are not directly comparable; both are labeled by source.
- The canonical metrics extractor refused to infer token, cost, session, or turn-count fields after rewind/compaction ambiguity.
- No blind evaluation record is included.
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.