GPAgentBench-2K - Diagnosis Accuracy: leaderboard
Metric: Diagnosis accuracy (%; final diagnosis judged correct by a GPT-5.4 judge against the ground truth; 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 April 2028. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 70 |
| 2 | Gemini 3.1 Pro (Preview) | 66.8 |
| 3 | MiMo-V2.5-Pro | 66 |
| 4 | GPT-5.4 | 64.1 |
| 5 | DeepSeek V4 Flash | 63.9 |
| 6 | Qwen 3.7 Max | 63.1 |
| 7 | Claude Sonnet 4.6 | 61.7 |
| 8 | MiniMax-M2.7 | 60 |
| 9 | DeepSeek V4 Pro | 59.5 |
| 10 | Kimi K2.6 | 56.7 |
| 11 | GLM-5.1 | 54.2 |
| 12 | Lingshu-32B | 50.9 |
| 13 | MedGemma-27B | 49.5 |
| 14 | Lingshu-7B | 47.1 |
Interactive version: theaggregate.ai/benchmark?slug=gpagentbench-2k-diagnosis-accuracy · How It Works · Data refreshed daily, snapshot 2026-09-26.