MedAgentGym — leaderboard
Metric: Average Score (self-reported). Source: benchmarklist.com. 29 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 70.15 |
| 2 | GPT-4.1 Mini | 61.72 |
| 3 | QwQ-32B | 51.5 |
| 4 | DeepSeek R1 Distill Llama 70B | 50.07 |
| 5 | GPT-4o (2024-08-06) | 48.75 |
| 6 | Qwen 3 32B | 48.19 |
| 7 | Qwen 2.5 32B Instruct | 42.68 |
| 8 | GPT-4o Mini | 37.47 |
| 9 | Llama 3.3 70B Instruct | 37.04 |
| 10 | Qwen 2.5 Coder 14B Instruct | 32.89 |
| 11 | Qwen 3 14B | 31.6 |
| 12 | Qwen 3 8B | 30.83 |
| 13 | DeepSeek R1 Distill Qwen 14B | 30.51 |
| 14 | Qwen 3 4B | 27.29 |
| 15 | Ministral-8B-Instruct-2410 | 22.27 |
Interactive version: theaggregate.ai/benchmark?slug=medagentgym · How the rankings work · Data refreshed daily, snapshot 2026-07-22.