MMBU - Ungrounded Classification: leaderboard
MMBU (Massive Multimodal Biomedical Understanding, 2026) probes the visual perception of vision-language models on biomedical images from 35 submodalities. This board is closed-ended (multiple-choice) classification of the whole image, scored as micro-averaged F1 times 100; higher is better.
Metric: Closed-ended (multiple-choice) ungrounded classification of the whole image: micro-averaged F1 x100 (0-100) over MMBU's biomedical images from 35 submodalities (95% bootstrap intervals not published); zero-shot, official checkpoints; exact match after normalization. Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 Mini | 53.9 | #779 |
| 2 | GPT-5.4 Mini | 53.3 | #389 |
| 3 | Qwen 3 VL 32B Instruct | 53 | #585 |
| 4 | Qwen 2.5 VL 32B Instruct | 52.6 | #1053 |
| 5 | Qwen 3 VL 8B Instruct | 47.2 | #981 |
| 6 | MedGemma-4B-IT | 43.9 | #2716 |
| 7 | Qwen 2.5 VL 7B Instruct | 36.1 | #1833 |
| 8 | Gemma 3 4B (IT) | 30.3 | #3831 |
| 9 | Qwen 3 VL 4B Instruct | 29.3 | #1279 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mmbu-ungrounded-classification · How It Works · Data refreshed daily, snapshot 2026-11-13.