Gemini Omni

The model supports natural language video editing that maintains scene consistency across edits, motion and style transfer from reference materials, and character or object swapping via text commands. It understands physics-based rendering concepts like gravity and fluid dynamics, and can synchronize on-screen text with action. Gemini Omni is available through the Gemini app, Google Flow, and YouTube Shorts, and leads head-to-head evaluations for video editing and text-to-video. A Google AI subscription is required for access.