MedQ-Bench - No-Reference Reasoning: leaderboard
Open-ended quality assessment of a single medical image, judged by an LLM on completeness, precision, consistency and the final quality verdict — four 0-2 criteria summed to 7.
Metric: Overall Score (0-7). Source: huggingface.co. Saturation forecast: Around December 2026. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 5.68 | #98 |
| 2 | GPT-4o | 5.32 | #343 |
| 3 | Gemini 2.5 Pro | 5.02 | #152 |
| 4 | Grok 4 | 5.02 | #181 |
| 5 | InternVL3-8B | 4.98 | #620 |
| 6 | InternVL3-38B | 4.96 | #405 |
| 7 | Mistral Medium 3 | 4.56 | #500 |
| 8 | Claude Sonnet 4 | 4.53 | #199 |
| 9 | Qwen 2.5 VL 7B | 4.37 | #643 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=medq-bench-no-reference-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.