OpenVLM MMT-Bench - VI: leaderboard
The "VI" subtask of OpenVLM MMT-Bench. OpenCompass OpenVLM evaluation of MMT-Bench: multimodal tasks covering visual recognition, localization, OCR, counting, relation reasoning, visual math, and hallucination detection.
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 4B | 57.2 | #1089 |
| 2 | Gemma 3 12B | 56.6 | #671 |
| 3 | Pixtral-12B | 56.2 | #808 |
| 4 | Gemma 3 27B | 56.1 | #607 |
| 5 | Qwen 2 VL 2B | 55.9 | #1220 |
| 6 | InternVL3-8B | 52.4 | #620 |
| 7 | InternVL3-78B | 51 | #352 |
| 8 | Claude 3 Haiku | 50.7 | #803 |
| 9 | Aquila-VL-2B | 49.7 | #1238 |
| 10 | InternVL3-38B | 46 | #405 |
| 11 | Claude 3.7 Sonnet | 45.5 | #249 |
| 12 | Llama 3.2 11B Instruct | 44.9 | #1126 |
| 13 | GPT-4.1 Nano | 44.6 | #731 |
| 14 | GPT-4.1 (2025-04-14) | 43.2 | #250 |
| 15 | MiniCPM-V-2.6 | 42 | #836 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-vi · How It Works · Data refreshed daily, snapshot 2026-10-20.