Mathematical Reasoning on IMO-Bench
57.02AccuracyF-1
Evaluation Results
| Method | Links | |
|---|---|---|
| F-1Model=Gemini 2.5 Pro, Model Category=Proprietary2026.01 | 57.02 | |
| Zero-ShotModel=GPT-5, Model Category=Proprietary2026.01 | 56.26 | |
| F-1Model=GPT-5, Model Category=Proprietary2026.01 | 55.58 | |
| Zero-ShotModel=Gemini 2.5 Pro, Model Category=Proprietary2026.01 | 55.44 | |
| CoTModel=GPT-5, Model Category=Proprietary2026.01 | 54.71 | |
| CoTModel=Gemini 2.5 Pro, Model Category=Proprietary2026.01 | 53.91 | |
| CoTModel=Qwen3-30B, Model Category=Open Source2026.01 | 53.35 | |
| F-1Model=Qwen3-30B, Model Category=Open Source2026.01 | 51.82 | |
| Zero-ShotModel=Qwen3-30B, Model Category=Open Source2026.01 | 50.99 | |
| PoTModel=Gemini 2.5 Pro, Model Category=Proprietary2026.01 | 49.06 | |
| PoTModel=GPT-5, Model Category=Proprietary2026.01 | 47.73 | |
| SeirênesBackbone=Qwen3-30B-A3B, same budget=true2026.05 | 46.7 | |
| DAPOBackbone=Qwen3-30B-A3B, same budget=true2026.05 | 44.3 | |
| PoTModel=Qwen3-30B, Model Category=Open Source2026.01 | 42.61 | |
| SeirênesBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 41.9 | |
| Qwen3-30B-A3BBackbone=Qwen3-30B-A3B2026.05 | 40.5 | |
| InTBackbone=Qwen3-4B-Instruct2026.05 | 38.8 | |
| Dr.GRPOBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 38 | |
| DAPOBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 37.7 | |
| Qwen3-4B-InstructBackbone=Qwen3-4B-Instruct2026.05 | 34.1 | |
| Scaf-GRPOBackbone=Qwen3-4B-Instruct2026.05 | 33.9 | |
| SAGEBackbone=Qwen3-4B-Instruct2026.05 | 32.5 | |
| Zero-ShotModel=DeepSeek-V3.1, Model Category=Open Source2026.01 | 30.54 | |
| F-1Model=DeepSeek-V3.1, Model Category=Open Source2026.01 | 27.64 | |
| CoTModel=DeepSeek-V3.1, Model Category=Open Source2026.01 | 27.56 | |
| PoTModel=Qwen3-235B, Model Category=Open Source2026.01 | 23.94 | |
| Zero-ShotModel=Qwen3-235B, Model Category=Open Source2026.01 | 21.85 | |
| F-1Model=Qwen3-235B, Model Category=Open Source2026.01 | 21.84 | |
| CoTModel=Qwen3-235B, Model Category=Open Source2026.01 | 20.45 | |
| PoTModel=DeepSeek-V3.1, Model Category=Open Source2026.01 | 17.48 | |
| SeirênesBackbone=Qwen2.5-7B-Instruct, same budget=true2026.05 | 13.3 | |
| SCRLBase Model=Qwen3-14B-Base2026.05 | 11.9 | |
| SAGEBackbone=Qwen2.5-7B-Instruct2026.05 | 11.2 | |
| LUFFYBackbone=Qwen2.5-7B-Instruct2026.05 | 11 | |
| SFTBase Model=Qwen3-14B-Base2026.05 | 11 | |
| QuestABase Model=Qwen3-14B-Base2026.05 | 10.4 | |
| ExGRPOBackbone=Qwen2.5-7B-Instruct2026.05 | 9.9 | |
| GRPOBase Model=Qwen3-14B-Base2026.05 | 9.9 | |
| DAPOBase Model=Qwen3-14B-Base2026.05 | 9.9 | |
| DAPOBackbone=Qwen2.5-7B-Instruct, same budget=true2026.05 | 9.5 | |
| NuRLBase Model=Qwen3-14B-Base2026.05 | 9.4 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct2026.05 | 8.8 | |
| SCRLBase Model=Qwen3-4B-Base2026.05 | 8.7 | |
| DAPOBase Model=Qwen3-4B-Base2026.05 | 8.3 | |
| QuestABase Model=Qwen3-4B-Base2026.05 | 8.3 | |
| GRPOBase Model=Qwen3-4B-Base2026.05 | 7.8 | |
| Initial ModelBase Model=Qwen3-14B-Base2026.05 | 7.4 | |
| SFTBase Model=Qwen3-4B-Base2026.05 | 7.1 | |
| NuRLBase Model=Qwen3-4B-Base2026.05 | 7.1 | |
| SCRLBase Model=Llama3.2-3B-Instruct2026.05 | 6.8 | |
| QuestABase Model=Llama3.2-3B-Instruct2026.05 | 6.1 | |
| Initial ModelBase Model=Qwen3-4B-Base2026.05 | 6.1 | |
| NuRLBase Model=Llama3.2-3B-Instruct2026.05 | 5.9 | |
| DAPOBase Model=Llama3.2-3B-Instruct2026.05 | 5.8 | |
| GRPOBase Model=Llama3.2-3B-Instruct2026.05 | 5.4 | |
| Initial ModelBase Model=Llama3.2-3B-Instruct2026.05 | 4.5 | |
| SFTBase Model=Llama3.2-3B-Instruct2026.05 | 2.6 |