Multimodal Mathematical Reasoning on FlowVerse All
36.5CoT ErrorInfiMM-Math-7B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| InfiMM-Math-7BModel Type=Open-source MLLMs2025.03 | 36.5 | 28.8 | |
| Math-LLaVA-13BModel Type=Math-specialized MLLMs2025.03 | 38 | 29.9 | |
| Qwen-VL-MaxModel Type=Closed-source MLLMs2025.03 | 43 | 36.3 | |
| MultiMath-7BModel Type=Math-specialized MLLMs2025.03 | 44 | 34.2 | |
| InternVL2.5-8BModel Type=Open-source MLLMs2025.03 | 44.7 | 41 | |
| SVE-Math-Qwen2.5-7BModel Type=Math-specialized MLLMs2025.03 | 46 | 39.4 | |
| VLM-R1-7B†Model Type=Open-source MLLMs2025.03 | 50.7 | 41.2 | |
| GPT-4o-miniModel Type=Closed-source MLLMs2025.03 | 51.3 | 44.5 | |
| Qwen2-VL-72BModel Type=Open-source MLLMs2025.03 | 52.3 | 48.6 | |
| Qwen2.5-VL-7BModel Type=Open-source MLLMs2025.03 | 53.8 | 42.2 | |
| InternVL2.5-78BModel Type=Open-source MLLMs2025.03 | 54.7 | 50.1 | |
| GPT-4oModel Type=Closed-source MLLMs2025.03 | 55.1 | 47.8 | |
| GPT-4VModel Type=Closed-source MLLMs2025.03 | 56.2 | 53.4 | |
| Claude-3.5-SonnetModel Type=Closed-source MLLMs2025.03 | 56.9 | 49.6 | |
| MathFlowModel Type=Open-source MLLMs, Backbone=Qwen2.5-VL-7B2025.03 | 57 | 46 | |
| Gemini-2.5-proModel Type=Closed-source MLLMs2025.03 | 62 | 55.3 | |
| GPT-5Model Type=Closed-source MLLMs2025.03 | 65.8 | 60.1 | |
| MathFlowModel Type=Closed-source MLLMs, Backbone=GPT-52025.03 | 66.5 | 61.8 |