Medmarks - MedHallu Medium: leaderboard
Metric: Score (%). Source: medmarks.ai. 71 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 85.53 |
| 2 | GPT-5.1 (Medium) | 74.06 |
| 3 | Claude Sonnet 4.5 | 73.86 |
| 4 | Ling-flash-2.0 | 72.2 |
| 5 | Qwen 3 235B A22B (Thinking) | 66.67 |
| 6 | GLM-4.7 FP8 | 66.49 |
| 7 | GPT-OSS-120B (High) | 66.45 |
| 8 | GPT-OSS-20B (High) | 65.83 |
| 9 | GPT-OSS-20B (Medium) | 63.16 |
| 10 | Grok 4 | 62.63 |
| 11 | Hermes 4 70B | 62.61 |
| 12 | INTELLECT-3 | 62.47 |
| 13 | Qwen 3 VL 30B A3B (Thinking) | 61.86 |
| 14 | GPT-OSS-20B (Low) | 61.14 |
| 15 | Llama 3.1 8B Instruct | 60.39 |
Interactive version: theaggregate.ai/benchmark?slug=medmarks-medhallu-medium · How It Works · Data refreshed daily, snapshot 2026-09-05.