Google has had a dedicated video generation model for a while now. Veo has been powering text-to-video features across the Gemini app and Flow for the past year. But at Google I/O 2026, the company introduced something it says is a step beyond that. Gemini Omni is a new multimodal model that doesn’t just take a text prompt and return a clip. It reasons across text, images, audio, and video together to produce a single, consistent output.
Instead of describing a scene from scratch, you can feed Gemini Omni an existing video clip, a voice memo, a sketch, or any mix of those, and build from there. You can refine the result through conversation, swapping characters, backgrounds, or objects just by asking. Google’s claymation protein folding demo at I/O showed what that looks like in practice. One prompt, a stop-motion style explainer with voice narration.
The first version rolling out now, Gemini Omni Flash, generates clips up to 10 seconds long. Google says that’s a product choice rather than a model limit, and longer videos are coming. Every clip gets an invisible SynthID watermark baked in, verifiable through the Gemini app or Google Search.
The Feature Google Is Holding Back
Here’s the one thing Gemini Omni won’t do yet. Audio and speech editing, changing what someone says in an existing video, is absent from this launch. Google said explicitly that it’s still working out how to bring that to users responsibly. It’s a notable gap given how close the demos came to that territory. You can create a personal digital avatar to appear in videos, but it requires recording yourself speaking a series of numbers first as an anti-deepfake step.
Gemini Omni Flash is live now in the Gemini app and Google Flow for Plus, Pro, and Ultra subscribers. YouTube Shorts users get it free later this week. A developer API arrives in the coming weeks, and a higher-tier Omni Pro model was teased with details to follow. The speech editing capability will come eventually. Google just isn’t ready to ship it.
