Gemini Omni: Google's Video-First AI Creator Expands From Any Input
Google announces Gemini Omni, a video-first AI model that can generate and edit high-quality videos from a wide range of inputs (images, audio, text, and video) with conversational, build-on-each-step editing. The Omni Flash model rolls out today to the Gemini app, Google Flow, and YouTube Shorts, and can transform existing footage, add new characters or objects, adjust environment and style, and even create a digital avatar voiced like the user. It leverages physics-aware rendering, knowledge from Gemini, and will use SynthID watermarks to verify generation. Audio output starts with voice references only, and some editing features are still being tested to ensure responsible use and privacy protections. Availability begins for AI Plus, Pro, and Ultra subscribers globally, with expansion to YouTube Create/App users this week.