Limitations:
- Omni currently offers 10-second video generations, with longer ones coming soon.
- Uploading audio references and scene extensions is not yet supported in the Gemini API for this model.
- Video references up to 3 seconds long are accepted by the API schema, but are currently not processed correctly by the model.
- Character consistency during scene changes or panning movements has some limitations, but we are working to improve this.
Gemini Omni is available in public preview starting today on Google AI Studio and Gemini API. For a complete list of model capabilities and region-specific restrictions, please check the developer documents.
Build both models today
The real magic happens when you put these models together. Apply Nano Banana 2 Lite as a model to quickly generate an image, then pass that image as a reference to Gemini Omni Flash to animate it into high-quality video. Additionally using Interaction API For these multi-turn experiences, you can maintain session history and context so that users can chain together up to three consecutive edits.
To support you get started, we’ve created some demo apps that you can remix and let you see how to combine Nano Banana 2 Lite and Gemini Omni Flash into one process.