BRIDGE Medical Leaderboard - CoT: leaderboard
Published Average Performance across 87 tasks for CoT protocol; five-shot is the Few-Shot setting. No across-protocol aggregate.
Metric: Average Performance (%). Source: huggingface.co. Saturation forecast: Around September 2028. 112 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B (IT) | 45.88 | #176 |
| 2 | Gemma 4 26B A4B (IT) | 45.17 | #297 |
| 3 | Gemini 2.5 Flash | 43.29 | #245 |
| 4 | Qwen 3.8 27B (Non-reasoning) | 42.94 | #105 (Qwen 3.8 27B) |
| 5 | Qwen 3 Next 80B A3B (Thinking) | 42.9 | #320 (Qwen 3 Next 80B A3B) |
| 6 | DeepSeek R1 | 42.1 | #251 |
| 7 | Gemini 2.0 Flash (001) | 41.98 | #360 |
| 8 | Qwen 3 30B A3B 2507 (Thinking) | 41.43 | #376 (Qwen 3 30B A3B 2507) |
| 9 | GPT-4o (2024-08-06) | 40.66 | #336 |
| 10 | Gemini 1.5 Pro (002) | 40.53 | #349 |
| 11 | Qwen 3 Next 80B A3B Instruct | 40.5 | #368 |
| 12 | Qwen 3 235B A22B (Thinking) | 40.14 | #313 (Qwen 3 235B A22B) |
| 13 | Qwen 3 30B A3B (Thinking) | 39.35 | #493 (Qwen 3 30B A3B) |
| 14 | DeepSeek R1 Distill Llama 70B | 38.95 | #594 |
| 15 | Mistral Large 2 (Nov) Instruct (2411) | 38.9 | #465 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=bridge-medical-leaderboard-cot · How It Works · Data refreshed daily, snapshot 2026-10-20.