MedQ-Deg - Mild Degradation (L1): leaderboard
Metric: Accuracy (%) on the MedQ-Deg items with mild degradations (L1, the diagnostic features still intact), multiple-choice medical VQA items merged from OmniMedVQA, GMAI-MMBench and MedXpertQA, each image corrupted by modality-specific degradations whose severity three radiologists calibrated (24,894 QA pairs over 7 modalities and 18 degradation types), single-letter answer, temperature 1.0, mean of 3 runs; higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 40 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B Instruct | 71.87 | #264 |
| 2 | Qwen 3 VL 30B A3B Instruct | 70.79 | #365 |
| 3 | Gemini 2.5 Pro | 69.93 | #145 |
| 4 | GPT-5 | 69.09 | #91 |
| 5 | GPT-5.1 | 67.34 | #131 |
| 6 | Gemini 2.5 Flash | 67.18 | #237 |
| 7 | GPT-4.1 | 66.5 | #240 |
| 8 | GPT-4.1 Mini | 65.87 | #346 |
| 9 | GPT-4o | 65.22 | #333 |
| 10 | GLM-4.5V | 64.57 | #339 |
| 11 | Gemma 3 27B | 58.91 | #596 |
| 12 | Claude Sonnet 4.5 | 58.28 | #138 |
| 13 | Qwen 2.5 VL 72B Instruct | 58.28 | #364 |
| 14 | MedGemma-4B | 58.12 | #731 |
| 15 | Qwen 2.5 VL 32B Instruct | 55.67 | #443 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=medq-deg-mild-degradation-l1 · How It Works · Data refreshed daily, snapshot 2026-10-11.