Mathematics on AIME 2025 (Acc, throughput)
67.7AccuracyDense
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DenseModel=Phi 4 reasoning plus2025.07 | 67.7 | 168.12 | |
| ReasonCacheModel=QwQ 32B2025.07 | 63.3 | 82.3 | |
| DenseModel=QwQ 32B2025.07 | 63.3 | 48.72 | |
| ReasonCacheModel=Phi 4 reasoning plus2025.07 | 63.3 | 243.21 | |
| DenseModel=DeepSeek R1 Distill Qwen 32B2025.07 | 53.3 | 57.55 | |
| ReasonCacheModel=DeepSeek R1 Distill Qwen 32B2025.07 | 50 | 82.34 | |
| RandomModel=QwQ 32B2025.07 | 23.3 | 84.1 | |
| RandomModel=DeepSeek R1 Distill Qwen 32B2025.07 | 20 | 81.81 | |
| RandomModel=Phi 4 reasoning plus2025.07 | 13.3 | 239.41 |