OpenVLM MMT-Bench - DU: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-78B | 92.9 |
| 2 | InternVL3-38B | 88.6 |
| 3 | Qwen 2 VL 72B | 86.4 |
| 4 | Gemma 3 27B | 86.4 |
| 5 | InternVL3-8B | 85 |
| 6 | GPT-4.1 (2025-04-14) | 85 |
| 7 | GPT-4.1 Mini | 81.4 |
| 8 | Gemini 1.5 Flash | 81.4 |
| 9 | GPT-4.1 Nano | 79.3 |
| 10 | Qwen 2 VL 7B | 75.7 |
| 11 | Gemma 3 12B | 73.6 |
| 12 | Llama 3.2 90B Vision Instruct | 73.6 |
| 13 | InternVL2-8B | 72.9 |
| 14 | Grok 2 (1212) | 72.9 |
| 15 | Claude 3.7 Sonnet | 70.7 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-du · How It Works · Data refreshed daily, snapshot 2026-09-05.