OpenVLM MMBench V1.1 CN - Physical Property Reasoning: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 284 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | InternVL3-38B | 82.3 |
| 2 | InternVL3-78B | 81 |
| 3 | GPT-5 | 79.7 |
| 4 | GPT-4.1 (2025-04-14) | 78.5 |
| 5 | Qwen 2 VL 72B | 77.2 |
| 6 | GPT-4o ChatGPT | 77.2 |
| 7 | GPT-4.5 | 77.2 |
| 8 | InternVL2.5-78B | 75.9 |
| 9 | GPT-5 Mini (2025-08-07) | 75.9 |
| 10 | Gemini 2.5 Pro | 73.4 |
| 11 | Grok 2 (1212) | 73.4 |
| 12 | Gemma 3 27B | 70.9 |
| 13 | Claude 3.5 Sonnet (20241022) | 68.4 |
| 14 | GPT-5 Nano | 68.4 |
| 15 | InternVL3-8B | 67.1 |
Interactive version: theaggregate.ai/benchmark?slug=openvlm-mmbench-v1-1-cn-physical-property-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.