DALPHIN (Independent) - Diagnosis: leaderboard
Naming the most likely diagnosis, by WHO tumour classification where possible, from the histopathology images of the 300 cases of the DALPHIN pathology copilot benchmark (Lems et al., 2026), with only the tissue type given. The extracted answer is scored by BioBERT semantic similarity to the reference diagnosis or its best synonym; higher is better. Independent scenario: every question is sent as a standalone query without earlier answers.
Metric: BioBERT semantic similarity (biomedical sentence-embedding similarity, at most 1) of the extracted free-response diagnosis to the reference diagnosis or its best-matching WHO or ICD synonym (300 cases, tissue type given; independent scenario: each question asked as a standalone query). Source: arxiv.org. Saturation forecast: Around February 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro (Medium) | 0.71 | #262 (Gemini 2.5 Pro) |
| 2 | GPT-5 (Medium) | 0.64 | #163 (GPT-5) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dalphin-independent-diagnosis · How It Works · Data refreshed daily, snapshot 2026-11-12.