OpenVLM MMStar - Math: leaderboard
OpenCompass OpenVLM MMStar subcategory score for visual math questions.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 284 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 89.2 | #98 |
| 2 | GPT-5 Mini (2025-08-07) | 87.2 | #172 |
| 3 | GPT-5 Nano | 83.2 | #425 |
| 4 | InternVL3-78B | 83.2 | #352 |
| 5 | InternVL3-38B | 80.4 | #405 |
| 6 | InternVL2.5-78B | 80 | #431 |
| 7 | Gemini 2.0 Flash | 79.6 | #342 |
| 8 | Gemini 1.5 Pro (002) | 79.6 | #349 |
| 9 | GPT-4o ChatGPT | 78.8 | |
| 10 | GPT-4.5 | 78.4 | #274 |
| 11 | GPT-4.1 (2025-04-14) | 78 | #250 |
| 12 | InternVL3-14B | 78 | #505 |
| 13 | Gemini 2.5 Pro | 77.6 | #152 |
| 14 | InternVL3-8B | 76.4 | #620 |
| 15 | Grok 2 (1212) | 75.2 | #496 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmstar-math · How It Works · Data refreshed daily, snapshot 2026-10-20.