MedMosaic - Speech and Sound MCQ: leaderboard
Metric: Accuracy (%; multiple choice, extracted answer letter against the answer key; 10,884 questions on synthetic MTS-Dialog consultations with inserted clinical sounds). Source: arxiv.org. Saturation forecast: Around December 2026. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 68.3 |
| 2 | Gemini 2.5 Flash | 60.2 |
| 3 | GPT-4o Audio | 35.2 |
| 4 | Gemma 3n E4B (IT) | 33.3 |
| 5 | Qwen2.5-Omni-7B | 32.5 |
| 6 | Phi-4 Multimodal Instruct | 26.6 |
Interactive version: theaggregate.ai/benchmark?slug=medmosaic-speech-and-sound-mcq · How It Works · Data refreshed daily, snapshot 2026-09-26.