MATH-Vision
Multimodal · 2024-02-22
MATH-Vision evaluates visual mathematical reasoning on 3,040 problems drawn from real competitions across 16 disciplines and five difficulty levels. Scores are reported as answer accuracy.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Sol | 95.8 |
| Kimi K3 | 94.3 |
| GPT-5.5 | 92.2 |
| GPT-5.4 | 92.0 |
| Gemini 3 Flash Preview | 90.6 |
| Gemini 3.1 Pro Preview | 89.8 |
| Doubao Seed 2.0 Pro | 88.8 |
| Qwen3.5 397B A17B | 88.6 |
| Kimi K2.6 | 87.4 |
| Opus 4.8 | 86.7 |
| Gemini 3 Pro Preview | 86.6 |
| Qwen3.5 122B A10B | 86.2 |
| Qwen3.5-27B | 86.0 |
| Kimi K2.5 | 84.2 |
| Qwen3.5 35B A3B | 83.9 |
| GPT-5.2 | 83.0 |
| Opus 4.5 | 77.1 |
| Step3-VL-10B | 76.0 |
| Qwen3 VL 235B A22B Thinking | 74.6 |
| GPT-5 | 72.0 |
| GPT-5 Mini | 71.9 |
| Opus 4.6 | 71.2 |
| Sonnet 4.5 | 71.1 |
| Qwen3-VL-235B-A22B-Instruct | 66.0 |
| Opus 4.1 | 66.0 |
| GLM-4.5V | 65.6 |
| GLM-4.6V | 63.5 |
| o1 (source unspecified) | 60.3 |
| Qwen3 VL 8B Thinking | 59.6 |
| Claude 3.7 Sonnet | 58.6 |
| O4 Mini | 58.0 |
| Kimi-VL-A3B-Thinking-2506 | 56.9 |
| GLM-4.1V-9B-Thinking | 54.4 |
| GLM 4.6V Flash | 54.0 |
| InternVL3.5-8B | 52.0 |
| Gemini 2.0 Pro Experimental 02-05 | 48.1 |
| GPT-4.5 | 47.3 |
| Gemma 3 27B IT | 46.0 |
| InternVL3-78B | 43.1 |
| Gemini 2.0 Flash | 41.3 |