Multimodal Mathematical Reasoning on FlowVerse Text Centric (Acc, CoT-E)
38.1CoT ErrorQwen-VL-Plus
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen-VL-Plus2025.03 | 38.1 | 24.6 | |
| InfiMM-Math-7BModel Type=Open-source MLLMs2025.03 | 43.8 | 38.1 | |
| InfiMM-Math-7B2025.03 | 43.8 | 38.1 | |
| Gemini-Pro2025.03 | 44.3 | 36.5 | |
| Math-LLaVA-13BModel Type=Math-specialized MLLMs2025.03 | 45.1 | 39.3 | |
| MathLLaVA-13B2025.03 | 45.1 | 39.3 | |
| InternVL2.5-8BModel Type=Open-source MLLMs2025.03 | 49.2 | 41.3 | |
| InternVL2.5-8B2025.03 | 49.2 | 41.3 | |
| Qwen-VL-MaxModel Type=Closed-source MLLMs2025.03 | 49.8 | 42.1 | |
| Qwen-VL-Max2025.03 | 49.8 | 42.1 | |
| MultiMath-7BModel Type=Math-specialized MLLMs2025.03 | 50.6 | 44.8 | |
| MultiMath-7B2025.03 | 50.6 | 44.8 | |
| SVE-Math-Qwen2.5-7BModel Type=Math-specialized MLLMs2025.03 | 53.1 | 47.3 | |
| SVE-Math-Qwen2.5-7B2025.03 | 53.1 | 47.3 | |
| GPT-4o-miniModel Type=Closed-source MLLMs2025.03 | 58.7 | 54.8 | |
| GPT-4o-mini2025.03 | 58.7 | 54.8 | |
| VLM-R1-7B†Model Type=Open-source MLLMs2025.03 | 59 | 54.2 | |
| VLM-R1-7B†2025.03 | 59 | 54.2 | |
| Qwen2-VL-72BModel Type=Open-source MLLMs2025.03 | 59.4 | 47.3 | |
| Qwen2-VL-72B2025.03 | 59.4 | 47.3 | |
| Qwen2.5-VL-7BModel Type=Open-source MLLMs2025.03 | 60.1 | 52.8 | |
| Qwen2.5-VL-7B2025.03 | 60.1 | 52.8 | |
| Claude-3.5-SonnetModel Type=Closed-source MLLMs2025.03 | 60.8 | 52.6 | |
| Claude-3.5-Sonnet2025.03 | 60.8 | 52.6 | |
| GPT-4oModel Type=Closed-source MLLMs2025.03 | 61 | 56.8 | |
| GPT-4o2025.03 | 61 | 56.8 | |
| MathFlowModel Type=Open-source MLLMs, Backbone=Qwen2.5-VL-7B2025.03 | 62 | 53.8 | |
| MathFlowBackbone=Qwen2.5-VL-7B2025.03 | 62 | 53.8 | |
| InternVL2.5-78BModel Type=Open-source MLLMs2025.03 | 66.1 | 62.7 | |
| InternVL2.5-78B2025.03 | 66.1 | 62.7 | |
| Gemini-2.5-proModel Type=Closed-source MLLMs2025.03 | 68.3 | 61.9 | |
| Gemini-2.5-pro2025.03 | 68.3 | 61.9 | |
| GPT-4VModel Type=Closed-source MLLMs2025.03 | 69.1 | 57.1 | |
| GPT-4V2025.03 | 69.1 | 57.1 | |
| GPT-5Model Type=Closed-source MLLMs2025.03 | 74.3 | 68.1 | |
| GPT-52025.03 | 74.3 | 68.1 | |
| MathFlowModel Type=Closed-source MLLMs, Backbone=GPT-52025.03 | 74.6 | 68.5 | |
| MathFlowBackbone=GPT-52025.03 | 74.6 | 68.5 |