MMBU - Grounded Classification (Boxes, Open-Ended): leaderboard
Metric: Open-ended classification of a region marked by a bounding box: micro-averaged F1 x100 (0-100) over MMBU's biomedical images from 35 submodalities (95% bootstrap intervals not published); zero-shot, official checkpoints; Qwen3-32B judges strict biomedical equivalence. Source: arxiv.org. Saturation forecast: Around May 2028. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Lingshu-32B | 15 |
| 2 | Qwen 3 VL 32B Instruct | 14.5 |
| 3 | GPT-5.4 Mini | 13.2 |
| 4 | InternVL3.5-8B | 12.2 |
| 5 | Qwen 3 VL 8B Instruct | 9.8 |
| 6 | Qwen 3 VL 4B Instruct | 9.3 |
| 7 | GPT-4.1 Mini | 8.7 |
| 8 | Qwen 2.5 VL 32B Instruct | 7.9 |
| 9 | MedGemma-4B-IT | 6.6 |
| 10 | Lingshu-7B | 4.7 |
| 11 | Qwen 2.5 VL 7B Instruct | 3 |
| 12 | Gemma 3 4B (IT) | 2.6 |
Interactive version: theaggregate.ai/benchmark?slug=mmbu-grounded-classification-boxes-open-ended · How It Works · Data refreshed daily, snapshot 2026-09-29.