Which Flow model can extend a video? Veo 3.1 Lite today, Omni soon

In Google Flow's current models table, only Veo 3.1 Lite lists Extend videos, limited to 8 seconds; Veo 3.1 Fast and Quality do not, and Gemini Omni Flash 1.1 shows Extend as coming soon (read 2026-10-02). The Gemini API already documents extension for Omni. Sume has no extend task, so continuity there means chaining clips.

Flow's table

Google's Flow help page lists supported features per model (Flow Help, read 2026-10-02). It does not give per-model credit costs.

The API is ahead of Flow

The Gemini API page for Omni Flash lists extend as a task: it appends to the end of a video, takes 3 to 10 second continuations, and the total can reach 40 seconds (Gemini API docs, read 2026-10-02). It also says uploaded videos for editing or extension must be 10 seconds or less, and that extension only appends to the end.

So the same model can extend through the API while the Flow table still marks it as coming soon. Check the Flow page again before you plan a Flow-only workflow.

On Sume

Sume's gemini-omni-flash-1.1 modes are text to video, image to video with an optional end image, reference to video, and a video_url edit. There is no extend mode (Video Router).

The practical stand-in is a chain: pull the last frame of clip one with video frames, then submit it as the first frame of clip two. That gives visual continuity, not the model's own look at the prior seconds of motion, so cuts can show.

Choosing

Need a real continuation today in a UI: Flow with Veo 3.1 Lite, capped at 8 seconds. Need Omni extension: the Gemini API. Need one job pipeline with webhooks and a single bill: chain Omni clips on Sume and accept that it is a continuation by first frame.