JMMMU-Pro: leaderboard
Japanese multimodal understanding benchmark extending JMMMU with image-based multidisciplinary questions.
Metric: Accuracy (%). Source: huggingface.co. Status: saturation imminent. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro | 87.05 |
| 2 | GPT-5.2 | 83.33 |
| 3 | aya-vision-8B | 26.74 |
Interactive version: theaggregate.ai/benchmark?slug=jmmmu-pro · How It Works · Data refreshed daily, snapshot 2026-09-05.