OpenVLM MMT-Bench - Scene Graph Recognition — leaderboard
Metric: Score (%). Source: huggingface.co. 207 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-38B | 100 |
| 2 | InternVL3-78B | 95 |
| 3 | Qwen 2 VL 72B | 95 |
| 4 | InternVL3-8B | 90 |
| 5 | Qwen 2 VL 7B | 90 |
| 6 | MiniCPM-V-2.6 | 90 |
| 7 | InternVL3-14B | 85 |
| 8 | Ovis2-8B | 85 |
| 9 | Grok 2 (1212) | 85 |
| 10 | GPT-4.1 (2025-04-14) | 85 |
| 11 | GPT-4.1 Mini | 80 |
| 12 | InternVL2-8B | 80 |
| 13 | Aquila-VL-2B | 80 |
| 14 | Claude 3.7 Sonnet | 80 |
| 15 | Gemma 3 27B | 80 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmt-bench-scene-graph-recognition · How the rankings work · Data refreshed daily, snapshot 2026-07-22.