OpenVLM MMT-Bench - Multiple View Image Understanding: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL2-8B | 40 |
| 2 | Qwen 2 VL 7B | 40 |
| 3 | InternVL3-78B | 35 |
| 4 | InternVL3-38B | 30 |
| 5 | Aquila-VL-2B | 30 |
| 6 | Qwen 2 VL 2B | 25 |
| 7 | Grok 2 (1212) | 25 |
| 8 | InternVL3-8B | 20 |
| 9 | Gemini 1.5 Pro | 20 |
| 10 | GPT-4.1 (2025-04-14) | 20 |
| 11 | Llama 3.2 90B Vision Instruct | 20 |
| 12 | GPT-4.1 Mini | 15 |
| 13 | Gemma 3 27B | 15 |
| 14 | Gemini 1.5 Flash | 15 |
| 15 | Gemma 3 4B | 15 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-multiple-view-image-understanding · How It Works · Data refreshed daily, snapshot 2026-09-05.