Active Evidence-Seeking and Diagnostic Reasoni: leaderboard
Large language models perform well on static medical examinations, yet clinical diagnosis often requires iterative evidence gathering under uncertainty. Building on prior interactive evaluation efforts, we introduce an OSCE-inspired standardized patient simulator and a controlled, reproducible benchmark for active diagnostic inquiry. Across 468 cases and 15 models in our protocol, we observe that
Metric: Task 2: Active Seeking Exact Accuracy (self-reported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 49.4 | #152 |
| 2 | Qwen 3 32B | 48.3 | #443 |
| 3 | Qwen 3 14B | 45.1 | #537 |
| 4 | Qwen 3 8B | 39.5 | #674 |
| 5 | Qwen 2.5 32B Instruct | 39.1 | #498 |
| 6 | GLM-4 32B | 35.7 | #459 |
| 7 | Qwen 2.5 14B Instruct | 34 | #647 |
| 8 | DeepSeek R1 Distill Qwen 32B | 33.9 | #655 |
| 9 | Llama 3 8B Instruct | 24.8 | #1129 |
| 10 | Qwen 2.5 7B Instruct | 23.3 | #862 |
| 11 | DeepSeek R1 Distill Qwen 14B | 23.3 | #844 |
| 12 | DeepSeek-R1-Distill-Qwen-7B | 6 | #1415 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=active-evidence-seeking-and-diagnostic-reasoni · How It Works · Data refreshed daily, snapshot 2026-10-20.