Try Pro FREE for 2 weeksGet Started

gemini-omni-flash-preview — Video Generation AI Model

Gemini Omni takes text, images or a clip and returns video, and edits conversationally: the node emits an Interaction ID, and feeding it into another Omni node refines the same clip rather than starting over.

Try in Studio

Details

  • CategoryVideo Generation
  • Year2026
  • LicenseGoogle Cloud Terms

Compliance & Provenance

  • ProviderGoogle · GPAI
  • EU AI Act RiskLimited Risk
  • Art. 50 TransparencyRequired — AI-generated outputs are marked

Inputs & Outputs

TextInputstring

[all tasks] Prompt describing the video to generate, or the edit to apply.

Image 1Inputimageoptional

[Image to Video / Reference to Video / Edit] Source image, or reference image 1. Ignored by Text to Video.

Image 2Inputimageoptional

[Reference to Video / Edit] Reference image 2. Ignored by Text to Video and Image to Video.

Image 3Inputimageoptional

[Reference to Video / Edit] Reference image 3. Ignored by Text to Video and Image to Video.

VideoInputvideooptional

[Edit only] The existing clip to modify. Ignored — and not uploaded — by every other task.

Previous Interaction IDInputstringoptional

[any task] Interaction ID output by a previous Omni node. Continues editing that same video instead of generating a new one.

VideoOutputvideo

Generated or edited video.

Interaction IDOutputstring

Wire into another Omni node's Previous Interaction ID port to keep editing this video.

Parameters

modelstringdefault Omni Flashoptions: Omni Flash

Model variant

Which Omni tier to call. 'Omni Flash' (default) is the only tier currently published; it is tuned for high-speed video generation and editing.

taskstringdefault Text to Videooptions: Text to Video, Image to Video, Reference to Video, Edit

Task (decides which input ports apply)

What Omni should do. 'Text to Video' (default) generates from the prompt alone and ignores any wired images. 'Image to Video' animates a supplied still. 'Reference to Video' uses up to three images as subject/style references. 'Edit' modifies an existing clip — supply it on the Video port, or chain a Previous Interaction ID to edit the clip a prior Omni node produced.

aspect_ratiostringdefault 16:9options: 16:9, 9:16

Aspect ratio (not used by Edit)

Output video aspect ratio. '16:9' for landscape / cinema, '9:16' for vertical / mobile. Omni exposes no resolution control — it is selected automatically. IGNORED by the Edit task: an edit inherits the source clip's geometry, and the API rejects the field outright if it is sent, so it is dropped from the request for that task.

Tags

  • video-generation
  • video-editing
  • text-to-video
  • image-to-video
  • multimodal
  • conversational-editing

Alternatives in Image/Video Models

  • GPT Image

    OpenAI GPT-Image text-to-image / edit. Needs API key.

  • Nano Banana

    Google Gemini-based image gen/edit with multimodal reasoning. Needs API key.

  • Sora 2

    OpenAI Sora 2 text-to-video / img2video. Needs API key.

  • VEO

    Google VEO 3.1 video gen (up to 4K, with native audio). Needs API key.

  • Video Analysis

    Gemini video analysis. Feed a video and get a transcript, timecodes, or highlight analysis as text. Needs API key.

Resources

All AI models