Google launches Gemini Omni 1.1 Flash for video creation

Google has introduced Gemini Omni 1.1 Flash, a video generation model that gives you more control over scenes, transitions, and final quality. It is now available to developers through the Gemini API and Google AI Studio.

The update is designed for editing tools, audiovisual workflows, and applications that can produce video without relying on manual adjustments for every change. The difference is that you can now better define how a shot starts, continues, and ends.

Longer, more consistent scenes

The scene extension feature lets you continue an existing video from the point where it ended. Gemini Omni 1.1 Flash analyzes up to 10 seconds of previous context, compared with the last second used by earlier models.

That helps preserve the characters’ appearance, camera movement, and the logic of the scene. You can extend a video in 10-second segments until reaching a maximum cumulative duration of 40 seconds.

In practice, this lets you turn a short clip into a longer sequence without the character changing appearance or the environment jumping noticeably.

Control over the first and last frame

Developers can also specify the opening and closing frame of a shot. The model generates the video connecting both points, which is useful for creating:

• Camera movements around a character.
• Zoom or lateral-sweep transitions.
• Loops that end similarly to how they begin.
• Continuous shots without visible cuts.

This lets you describe a complex scene and lock its endpoints, instead of leaving the AI to decide entirely how the shot evolves.

Faster, cheaper previews

To test ideas, Omni 1.1 Flash can generate videos at 360p up to 60% faster and at one-third the cost of a standard 720p generation, according to Google.

The company presents this mode as a drafting tool for creating storyboards, testing prompts, and repeating scenes before producing the final version. Once the result is defined, the model can generate outputs in 1080p or 4K for higher-quality work.

The 360p resolution does not replace the final video, but it does reduce the time and cost of iteration. That difference matters when a scene needs dozens of attempts.

Video references for preserving characters and movements

The API also lets you include up to three seconds of reference video as part of the multimodal input. An application can use existing videos to preserve a character’s movements, visual style, or behavior.

For example, a developer could assign the dance steps from three different videos to different characters and ask them all to perform together in the same space, in a single shot and without cuts.

The feature is designed for creative and editing tools that need more than a written description. References provide information about rhythm, posture, and movement that text alone cannot always express.

Where you can use it

Google is adding Gemini Omni 1.1 Flash to its developer ecosystem:

• Google AI Studio, for testing the model and building prototypes.
• Gemini Enterprise Agent Platform, for enterprise deployments.
• The API, along with documentation, examples, and prompting guides.

The model is also coming to global subscribers of Google AI Plus, Pro, and Ultra within Google Flow. Scene extension will be available to those same plans in the Gemini app.

For you, the most important change is not just that AI can generate video. It is that you now have more tools to repeat an idea, correct it, and turn it into something usable: you can work with fast drafts, lock the endpoints of a shot, reuse references, and then produce a high-resolution output.

The next thing to watch will be consistency in longer projects. Google currently limits cumulative extension to 40 seconds and video references to three seconds, so complex productions still require splitting the work into multiple shots and reviewing the result.