Pass native video to Gemini Flash in Agent Chat

Feature request for product/service

– Other –

Describe the request

Let users attach video files (.mp4, .mov, .webm) in Agent chat the same way as screenshots — either as native multimodal input for vision-capable models, or with optional server-side frame extraction + timestamps.

Use cases:

Review screen recordings / product demos without manual ffmpeg frame extraction
Debug UI motion, pacing, and transitions in marketing or app footage
Analyze bug repro clips with timecoded observations

Hey @Marten_Kauksi, thanks for the suggestion.

Video attachments in Agent chat (with the clip passed to a vision model) are something the team has been building and testing internally. We still have to work out some issues with larger video attachments. Hope we can get this to you someday!