OpenVLM MMT-Bench - AND: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-78B | 69 |
| 2 | InternVL3-38B | 69 |
| 3 | Claude 3.7 Sonnet | 67 |
| 4 | Claude 3.5 Sonnet | 67 |
| 5 | GPT-4.1 (2025-04-14) | 65 |
| 6 | Llama 3.2 90B Vision Instruct | 64 |
| 7 | Qwen 2 VL 72B | 63 |
| 8 | Gemini 1.5 Pro | 63 |
| 9 | Pixtral-12B | 61 |
| 10 | GPT-4.1 Mini | 60 |
| 11 | Aquila-VL-2B | 59 |
| 12 | Grok 2 (1212) | 59 |
| 13 | Llama 3.2 11B Instruct | 58 |
| 14 | Qwen 2 VL 7B | 57 |
| 15 | Gemini 1.5 Flash | 57 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-and · How It Works · Data refreshed daily, snapshot 2026-09-05.