Visual Evidence - MMMU 650 Unmodified: leaderboard
This study measures accuracy on an unmodified 650-question subset of MMMU, before its visual-evidence interventions. Accuracy on common 650-item MMMU subset (%).
Metric: Accuracy on common 650-item MMMU subset (%). Source: arxiv.org. Saturation forecast: Around January 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-8B | 54 | #614 |
| 2 | Qwen 2.5 VL 7B | 46.77 | #638 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=visual-evidence-mmmu-650-unmodified · How It Works · Data refreshed daily, snapshot 2026-10-29.