Synthetic Hospital - Specialty Summarization: leaderboard
Metric: Specialty-relevance F1 (%; specialty-conditioned summary scored against graph-derived specialty-relevant findings, zero-shot; public split of 200 synthetic longitudinal patients (1,268 in all, built from USMLE-style questions with ontology-grounded labels); single-turn, one prompting strategy locked per task for every model; graph-derived deterministic scoring). Source: arxiv.org. Saturation forecast: Around May 2027. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 68 |
| 2 | GLM-5 | 61.4 |
| 3 | GPT-5.3 | 59.5 |
| 4 | Kimi K2.5 (Thinking) | 59 |
| 5 | Qwen 3.5 397B A17B | 54.8 |
| 6 | DeepSeek V3.2 | 52.3 |
| 7 | Gemma 3 27B | 41.8 |
| 8 | Llama 4 Scout | 36.5 |
Interactive version: theaggregate.ai/benchmark?slug=synthetic-hospital-specialty-summarization · How It Works · Data refreshed daily, snapshot 2026-09-26.