Competitive Programming on LiveCodeBench v6 (Pass@1 Difficulty Breakdown)
72.5Pass@1 (All)Qwen3-235B-A22B-2507
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Qwen3-235B-A22B-2507Model Size=235B, Checkpoint=A22B-25072026.04 | 72.5 | 99.71 | 81.73 | 51.88 | |
| RefineRL-4B-2507 + Skeptical@16Model Size=4B, Checkpoint=2507, Test-time Scaling Strategy=Skeptical@162026.04 | 67.82 | 100 | 78.85 | 43.36 | |
| OR-Nemotron-32BModel Size=32B2026.04 | 66.07 | 99.71 | 73.68 | 43.05 | |
| Qwen3-30B-A3B-2507Model Size=30B, Checkpoint=A3B-25072026.04 | 64.61 | 99.85 | 75.48 | 38.59 | |
| OCR-Nemotron-1.1-32BModel Size=32B, Version=1.12026.04 | 62.25 | 98.84 | 71.15 | 36.8 | |
| RefineRL-4B-2507Model Size=4B, Checkpoint=25072026.04 | 61.57 | 98.98 | 72.72 | 34.22 | |
| Qwen3-235B-A22BModel Size=235B, Checkpoint=A22B2026.04 | 59.93 | 99.13 | 70.67 | 31.88 | |
| Qwen3-32BModel Size=32B2026.04 | 59.25 | 98.98 | 70.07 | 30.86 | |
| RefineRL-4B + Skeptical@16Model Size=4B, Test-time Scaling Strategy=Skeptical@162026.04 | 56.54 | 97.97 | 67.19 | 27.34 | |
| Qwen3-30B-A3BModel Size=30B, Checkpoint=A3B2026.04 | 56 | 98.4 | 67.79 | 25.55 | |
| Qwen3-4B-2507Model Size=4B, Checkpoint=25072026.04 | 55.21 | 98.84 | 63.7 | 26.25 | |
| DeepSeek-R1-0528-Qwen3-8BModel Size=8B, Base Model=Qwen3-8B2026.04 | 51.84 | 97.63 | 56.13 | 24.31 | |
| Qwen3-8BModel Size=8B2026.04 | 49.32 | 95.49 | 57.09 | 19.45 | |
| RefineRL-4BModel Size=4B2026.04 | 48.79 | 97.24 | 55.17 | 18.59 | |
| X-Coder-RL-Qwen3-8BModel Size=8B, Base Model=Qwen3-8B2026.04 | 48.39 | 85.61 | 56.85 | 22.89 | |
| Qwen3-4BModel Size=4B2026.04 | 48.29 | 95.35 | 55.05 | 18.59 |