VIPER (Veterinary Pathology): leaderboard
VIPER (Harvard Medical School and Mass General Brigham, 2026) is an expert-curated vision-language benchmark for veterinary toxicologic pathology: 1,251 questions written and validated by board-certified veterinary pathologists on 419 H&E-stained rat histology images from seven organ systems. The overall score in percent is the mean of the multiple-choice, KPrim and free-text scores; higher is better.
Metric: Overall score (%; mean of the MCQ, KPrim and free-text scores over 1,251 questions written and validated by board-certified veterinary pathologists on 419 H&E-stained rat histology images from seven organ systems). Source: arxiv.org. Saturation forecast: Around February 2027. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 56 | #123 |
| 2 | Gemma 4 31B (IT) | 54.4 | #342 |
| 3 | Qwen 3.5 27B | 52.4 | #359 |
| 4 | Claude Sonnet 4.6 | 48.2 | #161 |
| 5 | GPT-5.4 Mini | 48 | #389 |
| 6 | Gemini 2.5 Flash | 41 | #453 |
| 7 | GPT-5.4 Nano | 27.4 | #635 |
| 8 | MedGemma 1.5 4B | 24.8 | #2562 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=viper-veterinary-pathology · How It Works · Data refreshed daily, snapshot 2026-11-13.