MedBench v5 - CCR-LLM: leaderboard
The text track of MedBench v5 (2026) Clinical Cognitive Responsiveness: the macro-average of 36 clinical tasks spanning medical exam questions, consultation, documentation, diagnosis and treatment reasoning, quality control and safety, each normalized to 0-100. Higher is better.
Metric: Macro-average score (0-100) over the 36 text tasks of the MedBench v5 Clinical Cognitive Responsiveness LLM track (medical knowledge QA, language understanding and generation, clinical reasoning and decision-making, safety and ethics; macro-recall and LLM-judge scoring), each task normalized to 0-100; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 69.16 | #79 |
| 2 | Gemini 3.1 Pro (Preview) | 68.61 | #94 |
| 3 | Kimi K2.6 | 68.19 | #178 |
| 4 | GLM-5.1 | 67.7 | #179 |
| 5 | GPT-5.5 | 67.37 | #49 |
| 6 | Grok 4.20 | 66.87 | #297 |
| 7 | Seed 2.0 Pro | 66.56 | #211 |
| 8 | DeepSeek V4 Pro | 65.86 | #188 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=medbench-v5-ccr-llm · How It Works · Data refreshed daily, snapshot 2026-11-13.