← Back to leaderboard  ·  Show all task instances

ninja-build/ninja

a small build system with a focus on speed

12,895 cpp medium
1,438
Generated Behavioral Tests
72.3%
Best Score

Results by Model

# Model Score help_outline Percentage of hidden behavioral tests passed. Cost help_outline Total API cost in USD for this task instance. Calls help_outline Number of LLM API calls for this task instance.
1 Claude Opus 4.6 Anthropic 72.3% $16.10 324
2 Claude Sonnet 4.6 Anthropic 67.5% $18.29 400
3 Gemini 3 Flash Google 42.3% $0.23 65
4 Gemini 3.1 Pro Google 37.1% $0.68 45
5 Claude Haiku 4.5 Anthropic 34.2% $0.91 129
6 GPT 5 mini OpenAI 5.6% $0.01 11
7 Claude Opus 4.7 Anthropic 4.8% $1.69 69
8 GPT 5.4 OpenAI 3.5% $0.20 9
9 GPT 5.4 mini OpenAI 0.0% $0.03 14

Click row to see model details