Measured harness ledgerPublic result
GPT-5.6 SolCARVE — GPT-5.6 Sol Max
Build a playable Babylon.js WebGPU snowboarding game with the supplied rider asset, carved terrain, tricks, obstacles, audio, and a complete downhill run.
Max reasoningHeadline result
- Workflow cost
- $27.40
- Wall-clock
- 2h 19m 55s wall-clock
- Processed tokens
- 37.02M processed
- Record state
- partial_two_user_turn_token_timing_source_build_ledger
Public summary
GPT-5.6 Sol Max partial_two_user_turn_token_timing_source_build_ledger ledger: 2h 19m 55s wall-clock, 37.02M processed, and $27.40 Token-only standard API-equivalent estimate, not the actual Codex subscription charge.
Run identity and stack
- Result ID: carve-webgpu-snowboarding-gpt-5.6-sol-max
- Technical model: gpt-5.6-sol
- Provider: OpenAI Codex
- Stack: OpenAI Codex
- Stack: Technical model/configuration: gpt-5.6-sol
- Stack: Babylon.js WebGPU
- Stack: Supplied snowboarder fixture
- Stack: Harness v1 CARVE prompt
Cost basis
- Prompts with more than 272,000 input tokens use the published long-context rate for the whole request. The receipt reports zero calls over that threshold.
- No separately priced tool or non-token cost is included because no verifiable usage counts were supplied.
Primary artifact integrity
- Kind: interactive-webgpu-snowboarding-game
- Path: artifacts/carve-webgpu-snowboarding-gpt-5.6-sol-max/source/src/main.js
- SHA-256: 27d8a14a063715f864d94df57cc9dc90d5e1203f822be4dc40e1b04859488d1d
Validation evidence
- Path: artifacts/carve-webgpu-snowboarding-gpt-5.6-sol-max/validation/build-and-source-check.public.json
- SHA-256: ec393a1784baa8f0e2fe6db7f22d6875494da3122e8e25453383e158f0e4d409
Recorded caveats
- The root-task receipt records two user turns where the frozen benchmark protocol expects one; it is disclosed rather than treated as a clean one-shot result.
- Wall-clock is end-to-end workflow latency including tools and idle gaps, not model-only compute time.
- Output tokens include hidden reasoning, visible prose/code, and tool-call JSON.
- Cache-read tokens are discounted, so total processed tokens substantially overstate cost.
- Cache creation is recorded as zero because the transcript schema did not expose a cache-write field; no estimate was inferred.
- The cost is an API-list-price equivalent, not the actual subscription-backed Codex charge, and separately priced tools are excluded.
- The source notices identify the player GLB as user-provided and do not assert redistribution rights. The source archives the asset because it is a required supplied input, without relicensing it.
- The production build passed, but no independent browser acceptance run, local FPS measurement, hardware/viewport receipt, or blind evaluation was supplied.
- The supplied root-task receipt reports two user turns. This result is not presented as a clean one-shot run.
Visible evidence gaps
- clean one-user-turn rerun
- independent WebGPU browser acceptance run
- local FPS, browser, viewport, and hardware receipt
- independent final-capture receipt
- blind-evaluation record
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.