BuilderBuilder
Claude Sonnet 5.5 · Snowboarding
The episode says the runs are playable and under two hours, and that the Claude builds flicker at the end of the map. Every ledger here is a partial or asset-judge failure. Those states stay. The selected clocks are all under two hours, which agrees with the spoken duration without turning it into a pass. Supplied source capture Unaided model Open this round
Measured harness ledgerPublic result
Claude Sonnet 5.5

Snowboarding — Claude Sonnet 5.5

Build a playable Babylon.js WebGPU snowboarding game with the supplied rider asset, carved terrain, tricks, obstacles, audio, and a complete downhill run.

High reasoningHeadline result
Workflow cost
$46.55
Wall-clock
1h 38m 07s
Processed tokens
Fresh input 1,540, cache creation 3,152,310, cache read 126,420,132, output 1,248,584. Thinking tokens 511,981 are recorded separately and are not added. No processed total is recorded.
Record state
additive_harness_projection
Public summary

Claude Sonnet 5.5 Snowboarding: 1h 38m 07s, $46.55 API-equivalent estimate, Fresh input 1,540, cache creation 3,152,310, cache read 126,420,132, output 1,248,584. Thinking tokens 511,981 are recorded separately and are not added. No processed total is recorded..

Run identity and stack
  • Result ID: carve-webgpu-snowboarding-sonnet-5.5-high-20261006
  • Technical model: claude-sonnet-5-5
  • Provider: Anthropic Claude Code
  • Client: Claude Code CLI
  • Stack: Babylon.js WebGPU
  • Stack: Supplied snowboarder fixture
  • Stack: Harness v1 CARVE prompt
Cost basis
  • Clock is measurements.clean_attempt_wall_clock_seconds. Cost is recorded_cost.clean_scored_usd. Actual subscription charge is null.
  • Source: measurements.clean_model_usage_tokens at harness commit 9b7c614dbb5103cc373fdf6275b5c4efc10d8f0e.
Validation evidence
  • Result: partial_gameplay_complete_asset_judge_fail_performance_variable
  • Verification: Recorded status is partial_gameplay_complete_asset_judge_fail_performance_variable. Pending, partial, stopped, qualified, and model-finished states stay unjudged. Additive projection from harness commit 9b7c614dbb5103cc373fdf6275b5c4efc10d8f0e. Broad harness sync was not run. Clock is measurements.clean_attempt_wall_clock_seconds. Cost is recorded_cost.clean_scored_usd. Actual subscription charge is null. Source: measurements.clean_model_usage_tokens at harness commit 9b7c614dbb5103cc373fdf6275b5c4efc10d8f0e.
Recorded caveats
  • Partial gameplay complete, asset judge fail, performance variable. Clean scored cost excludes a discarded contaminated amount.
  • Clock is measurements.clean_attempt_wall_clock_seconds. Cost is recorded_cost.clean_scored_usd. Actual subscription charge is null.
  • Source: measurements.clean_model_usage_tokens at harness commit 9b7c614dbb5103cc373fdf6275b5c4efc10d8f0e.
Visible evidence gaps
  • additive projection; broad harness sync was not run
  • source harness commit 9b7c614dbb5103cc373fdf6275b5c4efc10d8f0e
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.

Browse BuilderReturn to Claude Sonnet 5.5
RemakeBenchResearch console