EviPathBench - Diagnostic Reasoning: leaderboard
EviPathBench (2026) evaluates vision-language models on whole-slide pathology using diagnostic paths that pathologists annotated at three magnifications on TCGA slides; each question has four options. This board gives the findings along a diagnostic route and asks for the triage class, diagnosis or refinement; accuracy in percent, chance is 25 and higher is better.
Metric: Accuracy (%) pooled over all questions (the All part of each cell) on Task 4: choosing the triage class, integrated diagnosis or refinement from the organ and the pathologist's findings along a diagnostic route at three magnifications (text-only reference DeepSeek-V4-Flash included); EviPathBench: pathologist-authored diagnostic paths on 1,822 TCGA whole-slide images (16 organs), findings at 2.5x, 10x and 40x magnification; four options per question with distractors mined from the same organ and magnification; parsed choice scored by exact match; chance 25; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 91.18 | #164 |
| 2 | Qwen 3.5 Flash | 90.98 | #435 |
| 3 | DeepSeek V4 Flash | 90.49 | #258 |
| 4 | GPT-5.2 | 89.2 | #176 |
| 5 | MedGemma-4B | 89.06 | #2444 |
| 6 | Qwen 2.5 VL 7B Instruct | 88.26 | #1833 |
| 7 | MedGemma 1.5 4B | 78.13 | #2562 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=evipathbench-diagnostic-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-13.