Holtercare-Bench - Presence (Video): leaderboard

Metric: Accuracy (%; multiple-choice closed QA on whether a given rhythm or abnormal waveform occurs in the recording; the ECG rendered as a sliding-window video, accelerated or truncated to fit input limits; zero-shot on the held-out test cases of Holtercare-23K, 20 percent of 788 clinical Holter recordings of 13-24 hours). Source: arxiv.org. Saturation forecast: Around 2029. 7 models tracked.

Top models

#ModelScore
1InternVL3.5-8B67.97
2Gemini 3 Flash58.17
3Lingshu-7B57.84
4MiniCPM-V-4.5-8B52.29

Interactive version: theaggregate.ai/benchmark?slug=holtercare-bench-presence-video · How It Works · Data refreshed daily, snapshot 2026-09-29.