ECGQuest: leaderboard
ECGQuest (Emory University, 2026) tests contextual knowledge of electrocardiography with true/false statements that GPT-4o generated from 23 ECG references and Computing in Cardiology proceedings, each paired with a negated form so both labels are balanced. Zero-shot accuracy in percent on the 1,050-question held-out test set; 50 is chance and higher is better.
Metric: Accuracy (%; 1,050 held-out true/false questions on ECG knowledge from 23 references and Computing in Cardiology proceedings, true and negated forms balanced; zero-shot, one True or False answer per question). Source: arxiv.org. Saturation forecast: Around September 2027. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 74.4 | #167 |
| 2 | GPT-4o | 71.7 | #741 |
| 3 | Gemma 4 31B | 71 | #394 |
| 4 | DeepSeek R1 Distill Qwen 14B | 69.8 | #2582 |
| 5 | Gemma 3 27B | 67.1 | #1631 |
| 6 | Gemma 4 26B | 66.5 | #612 |
| 7 | Gemma 3 12B | 64.6 | #2000 |
| 8 | GPT-3.5 | 63.1 | #2417 |
| 9 | Gemma 4 12B | 62.9 | #928 |
| 10 | Qwen 2.5 14B Instruct | 60.2 | #1732 |
| 11 | MedGemma-4B | 59.5 | #2444 |
| 12 | DeepSeek-R1-Distill-Qwen-7B | 59 | #7996 |
| 13 | Llama 3.1 8B Instruct | 58.4 | #4246 |
| 14 | Mistral 7B Instruct (v0.3) | 57.6 | #6653 |
| 15 | Gemma 3 4B | 56.4 | #5064 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=ecgquest · How It Works · Data refreshed daily, snapshot 2026-11-13.