DermoBench - Diagnosis (4-Way): leaderboard
Four-option multiple-choice skin-disease diagnosis from DermoBench (DermoGPT paper, Ru et al., 2026) on 2,000 held-out images from the benchmark's 14 source datasets, with distractors drawn from sibling diagnoses in a 325-class ontology. Accuracy in percent; higher is better.
Metric: Accuracy (%; 4-way leaf-diagnosis MCQA with ontology-sibling distractors, in-distribution). Source: arxiv.org. Saturation forecast: Around February 2027. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 72.6 | #452 |
| 2 | Qwen 3 VL 8B Instruct | 67.2 | #982 |
| 3 | QVQ-72B-Preview | 64.65 | #1173 |
| 4 | GLM-4.5V | 63.65 | #739 |
| 5 | Qwen 2.5 VL 72B Instruct | 61.5 | #845 |
| 6 | GPT-4o Mini | 59.5 | #1627 |
| 7 | Claude Sonnet 4.5 (Thinking) | 55.35 | #244 (Claude Sonnet 4.5) |
| 8 | Llama 3.2 90B | 47.85 | #1740 |
| 9 | Nemotron Nano 12B v2 VL | 47.25 | #2248 |
| 10 | Llama 3.2 11B | 29.25 | #6434 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dermobench-diagnosis-4-way · How It Works · Data refreshed daily, snapshot 2026-11-12.