Gemini Omni 1.1 Flash Extends AI Video to 40 Seconds With 4K Upscaling

Google shipped Gemini Omni 1.1 Flash on August 28, an updated video-generation model that stretches AI-made clips to 40 seconds, adds frame-to-frame transitions, and can output up to 4K. The update reaches developers immediately and is rolling out to consumer subscribers through the Gemini app and Google Flow the same day.

The headline change is how the model extends a scene. Earlier versions of Gemini's video generator only looked at the final second of a clip when continuing it, which made longer generations drift in lighting, character detail, and camera framing. Omni 1.1 Flash analyzes up to 10 seconds of prior context instead, and Google says that carries a clip through extensions in 10-second increments up to a 40-second total.

Keyframes, a faster draft mode, and 4K output

The model also adds keyframe-based transitions: pick a starting frame and an ending frame, and Gemini fills in the motion between them. Google points to orbiting camera moves, zoom transitions, and seamless looping clips as the intended use, the kind of shots that were previously hard to control with a single text prompt.

For iteration, there's a new 360p draft mode that Google says renders roughly 60% faster and at a third of the cost of a standard 720p generation, meant for testing an idea before committing to a full render. On the other end, finished clips can now be output at 1080p or upscaled to 4K for production use. Google's post doesn't specify whether the 4K output is natively generated or upscaled from a lower resolution.

Who gets access, and where

Developers can use Omni 1.1 Flash immediately through Google AI Studio and the Gemini Enterprise Agent Platform API, with a pricing table published alongside the technical documentation. On the consumer side, the full model is available in Google Flow for all Google AI Plus, Pro, and Ultra subscribers worldwide. Inside the regular Gemini app, only the scene-extension feature is rolling out to those same three paid tiers globally; Google hasn't said whether free Gemini users will get any part of the update.

The model also accepts short video clips, up to three seconds, as a reference input alongside text and images, which Google frames as useful for matching a specific style or subject across a new generation.

Why the context window matters

AI video's biggest tell has always been the seams: a character's shirt changes color mid-clip, or a background object flickers between frames. Widening the context window from one second to ten is a direct fix for that, and it's the kind of incremental model change that determines whether AI-generated video is usable for anything beyond a novelty clip. A 40-second ceiling is still short of a real scene, but it's a meaningful jump from the handful of seconds most consumer video models have shipped with this year.

How to try it

• Open Google Flow on an AI Plus, Pro, or Ultra account to generate with the full Omni 1.1 Flash feature set, including keyframe transitions and 4K output.
• In the Gemini app, look for scene extension when continuing an existing AI-generated clip.
• Developers can pull the model directly from Google AI Studio or the Gemini Enterprise Agent Platform API.

Google previously ran a limited-time free trial of the original Gemini Omni video tool, which we covered when that offer was extended into August. Omni 1.1 Flash is a paid-tier feature from launch, with no equivalent free window announced so far.