Gemini Omni scene extension: app, Flow, API or Sume?

Where can you use Gemini Omni scene extension? According to Google's launch post, Omni 1.1 Flash is in Google AI Studio, the Gemini Enterprise Agent Platform API, Google Flow and the Gemini app. Sume's gemini-omni-flash-1.1 generates clips but does not expose extend, so extension to 40 seconds needs one of Google's own surfaces.

Surfaces Google lists

Flow access is for AI Plus, Pro and Ultra subscribers, and the Gemini app offers scene extension to Plus, Pro and Ultra. The post says scene extension analyzes up to 10 seconds of prior context, up from 1 second, and extends in 10 second increments up to 40 seconds.

The API detail

The Gemini API Omni docs describe multi-turn editing through previous_interaction_id, extension that appends to the end only, and uploaded input videos of 10 seconds or less. Those are properties of Google's API, and Sume's docs state it does not expose them.

What Sume gives you

Sume's row covers 3 to 10 seconds at 360p, 720p, 1080p and 4K, 16:9 and 9:16, up to 10 images and up to 3 reference videos of at most 3 seconds each. A longer piece is a sequence of clips: end one clip on a frame that starts the next, then join them on a timeline render.

This is assembly, not model-side extension, so continuity depends on your prompts and reference frames.

Choosing a surface

If you are exploring and want scene extension without writing code, the Gemini app or Flow are the places Google lists, with a paid Google plan. If you want to script it, the Gemini API docs describe the method, and you will call Google directly. If your pipeline already runs on Sume for other models and a 10 second clip is enough, gemini-omni-flash-1.1 is one id among the others, with the same job, status and content endpoints.

Whichever you choose, keep first and last frames of each clip so you can join segments yourself if extension is not available.