Medmarks - Verifiable: leaderboard
Metric: Weighted Mean Win Rate (%). Source: medmarks.ai. Saturation forecast: Around November 2027. 89 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 64.52 |
| 2 | GPT-5.1 (Medium) | 62.08 |
| 3 | Grok 4 | 61.59 |
| 4 | Claude Sonnet 4.5 | 60.71 |
| 5 | GPT-5.2 (Medium) | 60.5 |
| 6 | GLM-4.7 FP8 | 60.1 |
| 7 | Qwen 3.5 27B (Thinking) | 59.33 |
| 8 | Qwen 3.6 35B A3B (Thinking) | 59.32 |
| 9 | Gemma 4 31B (IT) | 59.05 |
| 10 | Qwen 3.5 35B A3B (Thinking) | 58.56 |
| 11 | Qwen 3.6 27B (Thinking) | 58.5 |
| 12 | Qwen 3 235B A22B (Thinking) | 58.42 |
| 13 | Muse Glimmer 30B (xHigh) | 57.59 |
| 14 | Muse Glimmer 30B (High) | 57.43 |
| 15 | Qwen 3 Next 80B A3B (Thinking) | 56.95 |
Interactive version: theaggregate.ai/benchmark?slug=medmarks-verifiable · How It Works · Data refreshed daily, snapshot 2026-10-09.