LongMedBench - Visit Sorting: leaderboard

Metric: Mean Kendall tau (-1 to 1; agreement between the ordering given by the model and the true chronological order, averaged over 765 items; visit sorting: order five shuffled visit summaries with timestamps removed, from clinical progression cues; MIMIC-IV patients with long multi-visit event streams; naive long-context agent, provider default configuration). Source: arxiv.org. Saturation forecast: Around December 2026. 4 models tracked.

Top models

#ModelScore
1DeepSeek V3.2 (Thinking)0.42
2GPT-5 Mini0.38
3DeepSeek V3.2 (Non-reasoning)0.26
4Qwen Turbo0.11

Interactive version: theaggregate.ai/benchmark?slug=longmedbench-visit-sorting · How It Works · Data refreshed daily, snapshot 2026-09-29.