Google launches Gemini Omni Flash for realistic multimodal video generation
Google just dropped Gemini Omni Flash, a new family of generative models that lets you make realistic videos from text, images, clips, and even speech.
Announced at I/O 2026 and available starting today, it's a big step up from their earlier Genie model, which was all about virtual experiences.
AI Plus subscribers get immediate access
The first version, Gemini Omni Flash, is out for AI Plus subscribers right now. If you don't have a subscription, no worries, it'll be free on YouTube Shorts and YouTube Create later this week.
Early demos showed off claymation-style science explainers and marble runs with lifelike physics and sound.
Gemini Omni videos carry AI watermark
Every video made with Gemini Omni gets an AI watermark so viewers know it's generated content.
Audio editing is still limited for safety reasons.
Google teased a higher-level "Omni Pro," with details coming soon.