ClinTraceBench - Trend Classification: leaderboard
Metric: Accuracy (%; task T2, trend classification: direction and magnitude of a lab across two to five visits, 889 questions; full-context strategy: the whole multi-visit patient-clinician dialogue derived from MIMIC-IV records is given verbatim, models called through OpenRouter). Source: arxiv.org. Saturation forecast: Around September 2027. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 47.4 |
| 2 | DeepSeek V3.1 | 37.5 |
| 3 | Claude Haiku 4.5 | 34.9 |
| 4 | GPT-4o Mini | 25.5 |
Interactive version: theaggregate.ai/benchmark?slug=clintracebench-trend-classification · How It Works · Data refreshed daily, snapshot 2026-09-26.