Google has launched Gemini Omni, a new multimodal AI model built for video creation and editing. The model accepts a combination of text, images, video clips, audio, and drawings as inputs – making it one of the more versatile video AI tools announced to date.
The first release in the Omni family, Gemini Omni Flash, is being rolled out across the Gemini app, Google Flow, and YouTube Shorts.
What sets Gemini Omni apart is its conversational editing approach. Users can issue instructions in natural language, with each prompt building on previous edits. The model maintains continuity across scenes, characters, and visual elements – addressing one of the more persistent pain points in AI video generation.
The tool also introduces an avatar feature, allowing users to generate videos using a personalised digital version of themselves, complete with their own voice.
On the trust and verification front, Google confirmed that all Gemini Omni-generated content will carry SynthID digital watermarks. Users can verify AI-generated videos through the Gemini app, Gemini in Chrome, and Google Search.
Under the hood, Gemini Omni combines the reasoning capabilities of the Gemini model family with video generation functions informed by physics, historical context, and visual consistency – going beyond surface-level editing.
The launch extends Google’s growing generative media ecosystem, which already includes image creation and editing tools. Support for additional audio input formats is expected in subsequent updates.
