OpenVLM MMMU - Math: leaderboard
OpenCompass OpenVLM MMMU validation subcategory score for visual college-level questions in math.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around January 2027. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini (2025-08-07) | 86.7 | #172 |
| 2 | GPT-5 | 76.7 | #98 |
| 3 | GPT-5 Nano | 76.7 | #425 |
| 4 | Gemini 2.5 Pro | 73.3 | #152 |
| 5 | Gemini 1.5 Pro (002) | 66.7 | #349 |
| 6 | InternVL3-78B | 66.7 | #352 |
| 7 | InternVL3-8B | 63.3 | #620 |
| 8 | InternVL3-38B | 63.3 | #405 |
| 9 | Gemma 3 27B | 63.3 | #607 |
| 10 | Qwen 2.5 VL 7B | 63.3 | #643 |
| 11 | InternVL2.5-78B | 63.3 | #431 |
| 12 | GPT-4.1 Nano | 60 | #731 |
| 13 | Gemini 2.0 Flash | 60 | #342 |
| 14 | Gemini 1.5 Flash (002) | 60 | #583 |
| 15 | GPT-4o ChatGPT | 60 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmmu-math · How It Works · Data refreshed daily, snapshot 2026-10-20.