Mathematical Reasoning on AIME 2025 (Accuracy, AVG)
81.5AccuracyBase
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaseModel=Qwen3-235B-A22B2026.01 | 81.5 | 83.6 | |
| T3S(R1+235B+QWQ)Model=Qwen3-32B2026.01 | 80 | 83.34 | |
| T3S(R1+235B+QWQ)Model=Qwen3-8B2026.01 | 76.67 | 80.42 | |
| T3S(R1+235B)Model=Qwen3-32B2026.01 | 76.67 | 80.42 | |
| T3S(R1+235B)Model=Qwen3-8B2026.01 | 73.33 | 78.33 | |
| T3S(R1)Model=Qwen3-32B2026.01 | 73.33 | 78.33 | |
| T3S(R1)Model=Qwen3-8B2026.01 | 73.13 | 76.57 | |
| BaseModel=Qwen3-32B2026.01 | 72.08 | 76.77 | |
| BaseModel=DeepSeek-R12026.01 | 70 | 74.9 | |
| SFT(R1+235B+QWQ)Model=Qwen3-32B2026.01 | 69.17 | 73.75 | |
| SFT(R1)Model=Qwen3-32B2026.01 | 67.5 | 72.09 | |
| BaseModel=Qwen3-8B2026.01 | 67.08 | 71.46 | |
| SFT(R1+235B)Model=Qwen3-32B2026.01 | 66.67 | 74.17 | |
| SFT(R1+235B+QWQ)Model=Qwen3-8B2026.01 | 65.83 | 69.58 | |
| SFT(R1+235B)Model=Qwen3-8B2026.01 | 64.17 | 69.59 | |
| SFT(R1)Model=Qwen3-8B2026.01 | 55.83 | 64.17 |