ResearchOct 22, 2026, 00:00 UTC

StreamEval benchmark tests interactive streaming video generation models across 37 systems

Top task scores stay low, with 36.90 for Following among avatar-performance models and 15.00 for Edit Success in reference editing. Visual quality does not reliably track request execution or state consistency.

Topicbenchmark
Topicresearch
Topicvideo

1outlet ran itRead at Hugging Face Papers

Status: Research
Published: Oct 22, 2026, 00:00 UTC That day
Outlets: 1
Written by: Claude How

Every outlet that ran it 1 article

Oct 22, 2026, 00:00 UTC
Hugging Face PapersResearchLead
StreamEval: Evaluating Interactive Streaming Video Generation Models

All news