The paradigm of video generation and editing is shifting from tedious timeline operations to a few casual conversations with AI.
At Google I/O on May 19, 2026, Google launched Gemini Omni (first release: Omni Flash), its first universal world model that fuses Gemini's reasoning intelligence with media generation. It allows high-fidelity video generation and conversational editing via mixed text, image, video, and audio inputs.
Traditional video generation models like Sora only support one-shot prompt generation — if something's off, you start over. Gemini Omni's breakthrough is conversational video editing. You can tell the AI: "Change the background to the surface of Mars" or "Zoom in on the character's face." As you modify parts of the scene, the AI maintains character appearance and scene tone across multiple rounds, ensuring physics like water flow and gravity don't break.
To achieve this consistency, the model deeply integrates Gemini's reasoning architecture, focusing on simulating physical world common sense like gravity, fluid dynamics, and kinetic energy.
Gemini Omni Flash is available now to Google AI Plus, Pro, and Ultra subscribers in the Gemini App and Google Flow. It's also rolling out to YouTube Shorts and YouTube Create. All generated videos are automatically watermarked with SynthID (Google's invisible digital watermarking for AI content traceability).