MedStreamBench: leaderboard
MedStreamBench (2026) evaluates medical video understanding under temporally bounded evidence across retrospective, present, future and proactive settings built from 22 surgical and medical datasets. The overall score weights content correctness 0.7 and response timing 0.3 on streaming items; percent, higher is better.
Metric: Overall score (%; item-level score averaged over 5,419 QA items: content correctness for single-turn items and 0.7 x content + 0.3 x responsiveness for streaming future and proactive items; scored 0-1, x100). Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 37.28 | #266 |
| 2 | Qwen 2.5 VL 7B | 26.62 | #1787 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=medstreambench · How It Works · Data refreshed daily, snapshot 2026-11-13.