MedMosaic - Heart Sound MCQ: leaderboard
Metric: Accuracy (%; multiple choice, extracted answer letter against the answer key; 9,729 questions on heart sound recordings from HLS-CMDS and CirCor). Source: arxiv.org. Saturation forecast: Around March 2027. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 61.7 |
| 2 | Gemini 2.5 Flash | 47.2 |
| 3 | Gemma 3n E4B (IT) | 42 |
| 4 | Phi-4 Multimodal Instruct | 41.3 |
| 5 | Qwen2.5-Omni-7B | 33.8 |
| 6 | GPT-4o Audio | 3.2 |
Interactive version: theaggregate.ai/benchmark?slug=medmosaic-heart-sound-mcq · How It Works · Data refreshed daily, snapshot 2026-09-26.