Code Generation on LiveCodeBench Overall 2025-02-01 – present
80.9Pass@1 AccuracyFunPRM
Evaluation Results
| Method | Links | |
|---|---|---|
| FunPRMBase Model=O4-mini (High), Test-time Scaling Strategy=FunPRM2026.01 | 80.9 | |
| ORMBase Model=O4-mini (High), Test-time Scaling Strategy=Outcome Reward Models2026.01 | 79.4 | |
| Skywork-PRMBase Model=O4-mini (High), Test-time Scaling Strategy=Skywork-PRM2026.01 | 78.1 | |
| Self-CertaintyBase Model=O4-mini (High), Test-time Scaling Strategy=Self-Certainty2026.01 | 77.3 | |
| O4-miniReasoning Effort=High2026.01 | 77.1 | |
| Gemini-2.52026.01 | 72.5 | |
| O3Reasoning Effort=High2026.01 | 71.8 | |
| DeepSeek-R12026.01 | 68.7 |