What Happened
On 2026-05-19, Google began rolling out the first Gemini Omni model, Omni Flash, to Gemini, Flow, and YouTube Shorts.
Why It Matters
The launch connected Gemini's reasoning and world knowledge with generative media, making iterative video editing an interactive multimodal workflow. Maintaining a scene while changing its details across successive requests is a different workflow from generating unrelated clips. The model made iteration, input references, and continuity central to the creative process described at launch.
Technical Details
Users could combine text, images, audio, and video inputs to generate video and refine scenes through conversation. Videos included SynthID watermarks. Image and audio output were described as future additions, so the launch should be documented as video-first.