VMMU - Physics: leaderboard
Metric: Accuracy (%) on the VMMU Physics questions (Vietnamese page-image multiple choice); higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 81.16 |
| 2 | GPT-5 (Medium) | 76.45 |
| 3 | O3 (Medium) | 68.98 |
| 4 | Claude Opus 4.6 (Thinking) | 67.31 |
| 5 | Gemini 2.5 Flash | 67.04 |
| 6 | Qwen 2.5 VL 72B Instruct | 49.86 |
| 7 | Qwen 2.5 VL 32B Instruct | 49.17 |
| 8 | Llama 4 Scout | 44.88 |
| 9 | GPT-4.1 | 43.21 |
| 10 | Claude Sonnet 4 | 41.08 |
| 11 | Mistral Medium 3 | 36.57 |
| 12 | Gemma 3 27B | 30.47 |
| 13 | Mistral Small 3.2 | 30.28 |
| 14 | Llama 4 Maverick | 29.36 |
| 15 | aya-vision-8B | 26.04 |
Interactive version: theaggregate.ai/benchmark?slug=vmmu-physics · How It Works · Data refreshed daily, snapshot 2026-09-29.