ObGynLongBench - History-Level EHR - L2 Multi-Source Integration: leaderboard
Metric: Accuracy (%). Source: arxiv.org. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 71.5 |
| 2 | GPT-5.4 Mini | 68 |
| 3 | DeepSeek V4 Flash | 66.5 |
| 4 | Qwen 3.5 35B A3B (Non-reasoning) | 62.1 |
| 5 | Claude Haiku 4.5 | 61 |
| 6 | Qwen 3.5 9B (Non-reasoning) | 60.1 |
| 7 | Lingshu-7B | 59.5 |
| 8 | Qwen 3 VL 4B | 58.2 |
| 9 | Qwen 3.5 4B (Non-reasoning) | 57.5 |
| 10 | Gemma 4 E4B (Non-reasoning) | 53.2 |
| 11 | Phi-4 Mini Instruct | 47.8 |
| 12 | MedGemma 1.5 4B | 43.5 |
Interactive version: theaggregate.ai/benchmark?slug=obgynlongbench-history-level-ehr-l2-multi-source-integration · How It Works · Data refreshed daily, snapshot 2026-09-19.