DermoBench - OOD Diagnosis - SNU134: leaderboard

Metric: Accuracy (%; 4-way MCQA in the source dataset's own label space, out-of-distribution). Source: arxiv.org. Saturation forecast: Around December 2026. 18 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.5 (Thinking)68.75
2Qwen 2.5 VL 72B Instruct66.67
3Gemini 2.5 Flash65.42
4QVQ-72B-Preview62.92
5GPT-4o Mini59.17
6GLM-4.5V57.08
7Qwen 3 VL 8B Instruct55.42
8Nemotron Nano 12B v2 VL52.08
9Llama 3.2 90B49.58
10Lingshu-7B40

Interactive version: theaggregate.ai/benchmark?slug=dermobench-ood-diagnosis-snu134 · How It Works · Data refreshed daily, snapshot 2026-09-25.