Gemini Omni Flash

What it does

Gemini Omni Flash is a video-to-video model served through the gemini provider: you give it an existing video and a written instruction, and it returns an edited video with native audio. It is a preview model, so treat its behaviour as subject to change. On pipe2 it is wired into a single pipeline, Video Editor, where the edit is described in plain natural language. Input is video plus text; output is video.

When to use it

• You want to run the Video Editor pipeline on footage you already have, rather than generating a new clip from scratch.
• You want to change the style or overall look of a video and can describe the target in a sentence.
• You want to add effects to existing footage without opening an editing timeline.
• You want to modify specific scenes inside a clip while leaving the rest of it alone.
• You want the edit to carry audio through natively instead of handling sound as a separate step.

What to know before you run it

This model is video-to-video only. It needs an existing video as input; it will not generate a clip from a prompt alone. On the Video Editor pipeline no resolution, duration, or aspect-ratio options are listed for this model, so there is no tier to pick here. The live price table on this page is the authority on what a run costs. Because it is a preview model, its behaviour and availability can change. The pipeline covers style changes, added effects, scene modifications, and whole-look transformations, all driven by the instruction you write. One worked example is published on this page; start there to see the instruction style that works.

Also known as

You may see this model referred to as Gemini Omni, Gemini Omni Flash Preview, or by its API-style identifier gemini-omni-flash-preview. All of these refer to the same model listed here as Gemini Omni Flash.