MathVision — leaderboard
Metric: Overall Accuracy (%). Source: mathllm.github.io. 164 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 | 96.1 |
| 2 | Kimi K2.6 | 93.2 |
| 3 | Gemini 3 Flash | 90.6 |
| 4 | Gemini 3.1 Pro (Preview) (Thinking) | 89.8 |
| 5 | Qwen 3.5 397B A17B | 88.6 |
| 6 | Gemini 3 Pro | 86.6 |
| 7 | Qwen 3.5 122B A10B | 86.2 |
| 8 | Qwen 3.5 27B | 86 |
| 9 | Gemma 4 31B | 85.6 |
| 10 | Kimi K2.5 (Thinking) | 85 |
| 11 | Kimi K2.5 | 84.2 |
| 12 | Qwen 3.5 35B A3B | 83.9 |
| 13 | GPT-5.2 | 83 |
| 14 | Gemma 4 26B A4B | 82.4 |
| 15 | Human Expert | 82 |
Interactive version: theaggregate.ai/benchmark?slug=mathvision · How the rankings work · Data refreshed daily, snapshot 2026-07-22.