Medmarks - Med-HALT Reasoning FCT: leaderboard
The "Med-HALT Reasoning FCT" subtask of Medmarks. Living medical-model benchmark spanning verifiable and open-ended clinical evaluation families with model/system comparisons.
Metric: Score (%). Source: medmarks.ai. Saturation forecast: Estimated already saturated. 89 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 94.56 | #66 |
| 2 | Grok 4 | 93.47 | #181 |
| 3 | GPT-5.1 (Medium) | 93.36 | #135 (GPT-5.1) |
| 4 | Muse Glimmer 30B (xHigh) | 91.57 | #178 (Muse Glimmer 30B) |
| 5 | Muse Glimmer 30B (High) | 91.55 | #178 (Muse Glimmer 30B) |
| 6 | Qwen 3.5 27B (Thinking) | 90.87 | #187 (Qwen 3.5 27B) |
| 7 | Qwen 3 235B A22B (Thinking) | 90.07 | #313 (Qwen 3 235B A22B) |
| 8 | GPT-5.2 (Medium) | 89.98 | #110 (GPT-5.2) |
| 9 | Claude Sonnet 4.5 | 89.73 | #142 |
| 10 | GPT-OSS-120B (Medium) | 87.96 | #341 (GPT-OSS-120B) |
| 11 | GPT-OSS-120B (High) | 87.67 | #341 (GPT-OSS-120B) |
| 12 | Qwen 3.6 27B (Thinking) | 87.59 | #170 (Qwen 3.6 27B) |
| 13 | Qwen 3 Next 80B A3B (Thinking) | 86.86 | #320 (Qwen 3 Next 80B A3B) |
| 14 | GPT-OSS-120B (Low) | 86.7 | #341 (GPT-OSS-120B) |
| 15 | Nemotron 3.5 Lightning | 86.5 | #467 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=medmarks-med-halt-reasoning-fct · How It Works · Data refreshed daily, snapshot 2026-10-20.