ObGynLongBench - History-Level EHR - L2 Multi-Source Integration: leaderboard
ObGynLongBench (arXiv 2609.07601v1, Table 2): 17 LLMs answer 1,500 rule-grounded obstetric multiple-choice decision points from real longitudinal EHRs of 976 pregnant patients; open-source and medical models run with thinking disabled. Full-history input, cases whose rule integrates evidence from multiple sources; accuracy percent, higher is better.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 71.5 | #103 |
| 2 | GPT-5.4 Mini | 68 | #228 |
| 3 | DeepSeek V4 Flash | 66.5 | #155 |
| 4 | Qwen 3.5 35B A3B (Non-reasoning) | 62.1 | #268 (Qwen 3.5 35B A3B) |
| 5 | Gemma 4 26B A4B (Non-reasoning) | 61.4 | #360 (Gemma 4 26B A4B) |
| 6 | Claude Haiku 4.5 | 61 | #287 |
| 7 | Qwen 3.5 9B (Non-reasoning) | 60.1 | #391 (Qwen 3.5 9B) |
| 8 | Qwen 3.5 4B (Non-reasoning) | 57.5 | #514 (Qwen 3.5 4B) |
| 9 | Gemma 4 E4B (Non-reasoning) | 53.2 | #762 (Gemma 4 E4B) |
| 10 | Phi-4 Mini Instruct | 47.8 | #1232 |
| 11 | MedGemma 1.5 4B | 43.5 | #788 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=obgynlongbench-history-level-ehr-l2-multi-source-integration · How It Works · Data refreshed daily, snapshot 2026-11-05.