Mathematical Reasoning on AIME 2025 (Average Accuracy @ 32 samples)
13.9Average Accuracy (32 samples)WIST
Evaluation results
| Method | Links | |
|---|---|---|
| WISTBackbone=Qwen3-8B-Base2026.03 | 13.9 | |
| SPICEBackbone=Qwen3-8B-Base2026.03 | 13.4 | |
| R-ZeroBackbone=Qwen3-8B-Base2026.03 | 12.8 | |
| Base ModelBackbone=Qwen3-8B-Base2026.03 | 11.3 | |
| SPICEBackbone=Qwen3-4B-Base2026.03 | 10.9 | |
| WISTBackbone=Qwen3-4B-Base2026.03 | 9.7 | |
| R-ZeroBackbone=Qwen3-4B-Base2026.03 | 7.1 | |
| Base ModelBackbone=Qwen3-4B-Base2026.03 | 6.4 | |
| R-ZeroBackbone=OctoThinker-8B-Hybrid-Base2026.03 | 1.5 | |
| WISTBackbone=OctoThinker-8B-Hybrid-Base2026.03 | 1.4 | |
| Base ModelBackbone=OctoThinker-8B-Hybrid-Base2026.03 | 1.1 | |
| SPICEBackbone=OctoThinker-8B-Hybrid-Base2026.03 | 0.9 | |
| SPICEBackbone=OctoThinker-3B-Hybrid-Base2026.03 | 0.8 | |
| Base ModelBackbone=OctoThinker-3B-Hybrid-Base2026.03 | 0.6 | |
| WISTBackbone=OctoThinker-3B-Hybrid-Base2026.03 | 0.6 | |
| R-ZeroBackbone=OctoThinker-3B-Hybrid-Base2026.03 | 0.4 |