Visual Mathematical Reasoning on FlowVerse Vision Dense EI+OQ
28.5CoT-EQwen-VL-Plus
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen-VL-Plus2025.03 | 28.5 | 20.9 | |
| InfiMM-Math-7B2025.03 | 28.8 | 15.4 | |
| Gemini-Pro2025.03 | 32.5 | 23.3 | |
| VLM-R1-7B†2025.03 | 36.2 | 24.5 | |
| MathLLaVA-13B2025.03 | 36.2 | 18.6 | |
| InternVL2.5-8B2025.03 | 38.4 | 20.2 | |
| Qwen-VL-Max2025.03 | 38.6 | 15.2 | |
| Qwen2-VL-72B2025.03 | 40.8 | 25.3 | |
| GPT-4o-mini2025.03 | 41.1 | 26 | |
| MultiMath-7B2025.03 | 41.7 | 22.1 | |
| SVE-Math-Qwen2.5-7B2025.03 | 44.2 | 28.6 | |
| Qwen2.5-VL-7B2025.03 | 45 | 31 | |
| Claude-3.5-Sonnet2025.03 | 45 | 25.4 | |
| GPT-4o2025.03 | 45.2 | 30 | |
| GPT-4V2025.03 | 48.1 | 30.3 | |
| InternVL2.5-78B2025.03 | 48.7 | 34.3 | |
| MathFlowBackbone=Qwen2.5-VL-7B2025.03 | 49 | 39.1 | |
| Gemini-2.5-pro2025.03 | 52.1 | 37.1 | |
| GPT-52025.03 | 53.8 | 44.7 | |
| MathFlowBackbone=GPT-52025.03 | 58.2 | 54.1 |