SurgiQ: leaderboard
Metric: Micro accuracy over all 13,055 questions: accuracy (%) on SurgiQ's four-option surgical multiple-choice questions, zero-shot; the answer is the option label with the highest next-token log-probability (plain and space-prefixed labels), options shuffled once. Source: arxiv.org. Saturation forecast: Around December 2026. 35 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 72B | 68.08 |
| 2 | Qwen 2.5 72B Instruct | 68 |
| 3 | Qwen 2.5 32B | 65.43 |
| 4 | Qwen 2.5 32B Instruct | 64.92 |
| 5 | GPT-OSS-120B | 64.63 |
| 6 | Ministral-3-14B-Reasoning-2512 | 63.48 |
| 7 | DeepSeek R1 Distill Qwen 32B | 60.37 |
| 8 | GPT-OSS-20B | 57.56 |
| 9 | K2 Think V2 | 56.15 |
| 10 | DeepSeek R1 Distill Qwen 14B | 55.92 |
| 11 | Ministral-3-8B-Instruct-2512 | 54.4 |
| 12 | BioMistral-7B | 47.14 |
| 13 | MedGemma-4B-IT | 44.08 |
| 14 | Gemma 2 9B (IT) | 43.87 |
| 15 | MedGemma-27B-IT | 38.58 |
Interactive version: theaggregate.ai/benchmark?slug=surgiq · How It Works · Data refreshed daily, snapshot 2026-09-29.