OpenVLM MMT-Bench - Relation Hallucination: leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Pixtral-12B | 90 |
| 2 | InternVL3-8B | 85 |
| 3 | Qwen 2 VL 72B | 85 |
| 4 | Gemini 1.5 Flash | 85 |
| 5 | InternVL3-78B | 80 |
| 6 | InternVL3-38B | 80 |
| 7 | Claude 3.7 Sonnet | 80 |
| 8 | Grok 2 (1212) | 80 |
| 9 | GPT-4.1 (2025-04-14) | 80 |
| 10 | Llama 3.2 90B Vision Instruct | 80 |
| 11 | Qwen 2 VL 7B | 75 |
| 12 | Aquila-VL-2B | 75 |
| 13 | Gemini 1.5 Pro | 75 |
| 14 | Claude 3.5 Sonnet | 75 |
| 15 | GPT-4.1 Mini | 70 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-relation-hallucination · How It Works · Data refreshed daily, snapshot 2026-09-05.