CaReCoS - Inferred Questions: leaderboard
The inferred tier of CaReCoS (Centific, 2026): questions about heart, lung and cough spectrograms that need clinical reasoning grounded in the spectrogram rather than retrieval of an annotated attribute, credited by an LLM judge when the core clinical concept or conclusion matches the reference. The score is accuracy in percent; higher is better.
Metric: Accuracy (%) of one-to-two-sentence answers to clinically grounded questions about a mel-spectrogram image of the recording, judged by an LLM against the reference (strict factual matching for explicit questions, relaxed semantic matching for inferred ones); the 2,520 inferred questions that require clinical reasoning grounded in the spectrogram; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 | 84.6 | #229 |
| 2 | Gemini 2.5 Pro | 71.2 | #266 |
| 3 | Gemini 2.5 Flash | 42.5 | #453 |
| 4 | Qwen2.5-Omni-7B | 25.2 | #1789 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=carecos-inferred-questions · How It Works · Data refreshed daily, snapshot 2026-11-13.