MMRareBench - Diagnosis: leaderboard
Metric: Diagnosis track (T1) score: open-ended primary diagnosis from a masked rare-disease case narrative and its medical images, graded 0-100 by a Qwen3-VL-235B judge with a strict track-specific rubric that awards credit only when a criterion is fully met (exact match weighted three times disease category and specificity, capped at 20 without an exact match); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 76.6 |
| 2 | GPT-5 | 75.9 |
| 3 | Gemini 2.5 Pro | 74 |
| 4 | Claude Sonnet 4.5 | 68.5 |
| 5 | GPT-4o | 62.5 |
| 6 | Gemini 2.5 Flash | 62.2 |
| 7 | Qwen 3 VL 235B A22B Instruct | 58.1 |
| 8 | Qwen 2.5 VL 72B Instruct | 52.6 |
| 9 | Lingshu-32B | 49.8 |
| 10 | Qwen 3 VL 30B A3B Instruct | 46.9 |
| 11 | Qwen 2.5 VL 32B Instruct | 46.9 |
| 12 | MedGemma-27B-IT | 43.7 |
| 13 | Qwen 3 VL 8B Instruct | 39.3 |
| 14 | Qwen 2.5 VL 7B Instruct | 33.7 |
| 15 | Claude Haiku 4.5 | 30.9 |
Interactive version: theaggregate.ai/benchmark?slug=mmrarebench-diagnosis · How It Works · Data refreshed daily, snapshot 2026-10-07.