Mathematical Reasoning on MathVerse Text Lite (testmini)
21.2CoT ErrorQwen-VL-Plus
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Qwen-VL-Plus2025.03 | 21.2 | 11.1 | |
| SPHINX-MoE2025.03 | 21.9 | 16.4 | |
| InternLM-XC22025.03 | 28.3 | 17 | |
| MAVIS-7B2025.03 | 29.1 | — | |
| InfiMM-Math2025.03 | 32.4 | — | |
| Gemini-Pro2025.03 | 34.7 | 23.5 | |
| Qwen2-VL-72B2025.03 | 35.6 | 27.8 | |
| Qwen-VL-Max2025.03 | 37.7 | 26.1 | |
| MathFlowPerception Model=MathFlow-P-7B, Inference Model=Qwen2-VL-72B2025.03 | 37.7 | 31.7 | |
| GPT-4V2025.03 | 56.6 | 41.4 | |
| MathFlowPerception Model=MathFlow-P-7B, Inference Model=GPT-4V2025.03 | 58.9 | 46.4 |