OpenVLM MMMU - Humanities and Social Science — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 85 |
| 2 | GPT-5 (2025-08-07) | 85 |
| 3 | GPT-4o ChatGPT | 82.5 |
| 4 | GPT-4.1 (2025-04-14) | 81.7 |
| 5 | InternVL2.5-78B | 80.8 |
| 6 | GPT-5 Mini (2025-08-07) | 80.8 |
| 7 | InternVL3-78B | 80 |
| 8 | Qwen 2 VL 72B | 80 |
| 9 | Grok 2 (1212) | 80 |
| 10 | GPT-4.5 | 80 |
| 11 | InternVL3-38B | 79.2 |
| 12 | InternVL3-14B | 79.2 |
| 13 | Gemma 3 27B | 79.2 |
| 14 | Gemini 1.5 Flash (002) | 78.3 |
| 15 | InternVL3-8B | 77.5 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmmu-humanities-and-social-science · How the rankings work · Data refreshed daily, snapshot 2026-07-22.