MedMosaic - Lung Sound MCQ: leaderboard
Metric: Accuracy (%; multiple choice, extracted answer letter against the answer key; 585 questions on lung sound recordings from HLS-CMDS). Source: arxiv.org. Saturation forecast: Around March 2028. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 52.8 |
| 2 | Gemini 2.5 Flash | 51.6 |
| 3 | Gemma 3n E4B (IT) | 47.9 |
| 4 | Phi-4 Multimodal Instruct | 40.3 |
| 5 | Qwen2.5-Omni-7B | 36.6 |
| 6 | GPT-4o Audio | 4.1 |
Interactive version: theaggregate.ai/benchmark?slug=medmosaic-lung-sound-mcq · How It Works · Data refreshed daily, snapshot 2026-09-26.