EHRStruct (Synthea) - Filtering (Less Than): leaderboard

Metric: Accuracy (%; D-U2: 100 zero-shot questions on synthetic Synthea patient tables given as plain text, where each item asks to list the IDs of the patients meeting a condition on a categorical field and a less-than condition on a numeric field; an answer counts when it matches the reference exactly; the paper's check mark, a perfect score, reads 100, and its cross, no valid output, reads as missing). Source: arxiv.org. Saturation forecast: Around December 2026. 18 models tracked.

Top models

#ModelScore
1Gemini 2.5 Flash58
2GPT-4.151
3Gemini 2.0 Flash43
4DeepSeek V341
5DeepSeek V2.541
6GPT-3.5 Turbo15

Interactive version: theaggregate.ai/benchmark?slug=ehrstruct-synthea-filtering-less-than · How It Works · Data refreshed daily, snapshot 2026-09-29.