Gemini Flash 1.1 Extends AI Videos to 40 Seconds

Google pushed out version 1.1 of Gemini Omni Flash on Thursday, its video generation model, and the update changes how far a single clip can stretch. Scenes now extend in 10-second increments up to a 40-second cap, a jump from earlier limits that kept most AI-generated clips short and choppy at the seams.

The mechanics behind that extension got an overhaul too. Previous versions of the model looked at only the last second of existing footage when deciding how to continue a scene, which tended to produce jarring jumps in motion and lighting. Omni 1.1 Flash instead scans up to ten seconds of prior video, giving it more context to keep characters, backgrounds, and camera movement consistent as a clip grows.

Google also opened up new ways to steer output. Developers can now feed in three seconds of outside footage as a style reference, letting a model carry over a specific character design or motion pattern, or lock in start and end frames to generate camera movement between two fixed points. Pricing and resolution options changed as well, with a new low-cost draft mode aimed at faster iteration before committing to a final render.