Google has updated its Gemini Omni Flash video model to version 1.1, with changes aimed at making AI video generation cheaper, faster to test, and easier to control across a longer sequence.
The update focuses on a practical problem in generated video: keeping motion, characters, and visual continuity stable as a clip grows. Gemini Omni 1.1 Flash now gives developers more ways to guide a result before committing to higher-resolution output.
Scene extension gets a longer memory
One of the main changes is how scene extension works. Instead of analyzing just the last second of an existing video, Gemini Omni 1.1 Flash can now analyze up to ten seconds of footage.
That matters because a model extending a clip needs context. A single second may not capture the full motion, character position, visual rhythm, or camera direction of the scene. With up to ten seconds available for analysis, the model has more information to preserve a consistent look and flow.
Google says scenes can be extended in 10-second increments up to 40 seconds. That gives developers a clearer structure for building longer outputs: generate, review, extend, and repeat within a defined limit.
The result is not just longer video. It is a workflow that can better support continuity across multiple generated segments, especially when the visual state at the end of one segment needs to carry into the next.
Style references and keyframes add more control
Gemini Omni 1.1 Flash also allows developers to upload up to three seconds of external footage as a style reference. That reference can be used to carry over characters or motion patterns.
This gives developers another way to direct the model without relying only on written instructions. A short clip can communicate timing, movement, and visual behavior in a way that text may describe less precisely.
The update also supports setting start and end frames to create camera movements between keyframes. In practical terms, that gives the generation process a clearer beginning and destination for the shot.
These controls point to a more structured approach to AI video generation. Rather than treating each output as a single isolated prompt, developers can use visual inputs, reference footage, and keyframes to shape how a scene moves over time.
Draft mode lowers the cost of iteration
The other major change is a 360p draft mode. According to the source, it runs up to 60 percent faster at a third of the cost compared to 720p.
That is important because video generation often requires iteration. Developers may need to test motion, framing, character consistency, or scene extension before producing a final version. A lower-cost draft mode makes that testing stage less expensive.
Once the draft is acceptable, videos can be upscaled to 1080p or 4K. That separates creative exploration from final delivery, letting developers evaluate an idea quickly before spending more on higher-resolution output.
The published per-second pricing is:
- $0.03 for 360p
- $0.10 for 720p
- $0.15 for 1080p
- $0.30 for 4K
This pricing structure makes resolution a direct cost decision. Teams can use 360p when speed and iteration matter most, then move to 1080p or 4K when the visual result is ready for higher-quality output.
What the update means for AI video work
Gemini Omni 1.1 Flash is not described as a complete change in what AI video generation is. The update is more about control, cost, and repeatability.
The ability to analyze up to ten seconds of existing footage can help preserve visual continuity. The option to extend scenes in 10-second increments up to 40 seconds gives developers a clearer production rhythm. The three-second style reference adds a visual guide for characters or motion patterns. Start and end frames create another path for controlling camera movement.
Together, those features make the model more flexible for developers who need to test, revise, and refine generated video. The 360p draft mode is especially relevant because it changes the economics of experimentation: low-resolution tests can happen faster and at lower cost before upscaling to 1080p or 4K.
Gemini Omni 1.1 Flash is available through Google AI Studio and the developer docs.