GPAgentBench-2K - Treatment Score: leaderboard
Metric: Treatment score (%; prescribed treatment scored by a GPT-5.4 judge against the ground-truth treatment; zero-shot GP agent with six clinical actions over 2,428 physician-validated primary-care cases, GPT-4o-mini patient simulator, at most 10 turns). Source: arxiv.org. Saturation forecast: Around September 2028. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 | 53.2 |
| 2 | Kimi K2.6 | 49.2 |
| 3 | Claude Opus 4.7 | 46 |
| 4 | Gemini 3.1 Pro (Preview) | 41.6 |
| 5 | DeepSeek V4 Pro | 41.2 |
| 6 | Qwen 3.7 Max | 41.1 |
| 7 | DeepSeek V4 Flash | 40.6 |
| 8 | MiMo-V2.5-Pro | 38.3 |
| 9 | Claude Sonnet 4.6 | 36 |
| 10 | GLM-5.1 | 34.3 |
| 11 | MiniMax-M2.7 | 31.3 |
| 12 | MedGemma-27B | 29.7 |
| 13 | Lingshu-32B | 25.2 |
| 14 | Lingshu-7B | 23 |
Interactive version: theaggregate.ai/benchmark?slug=gpagentbench-2k-treatment-score · How It Works · Data refreshed daily, snapshot 2026-09-26.