Google Rolls Out Gemini Omni 1.1 Flash, Extends AI Video Clips to 40 Seconds

Google released a new version of its generative video model, Gemini Omni 1.1 Flash, on August 27, 2026. The update gives creators and developers significantly more control over how AI-generated videos are made and edited, and it is launching immediately in popular creative tools, including Adobe Firefly and Figma Weave.

The biggest change in Gemini Omni 1.1 Flash is how it extends scenes. The previous version of the model, when generating a continuation of a shot, only took into account the last second of the preceding segment, which often led to jumps in camera movement, lighting shifts, or inconsistent characters. The new model analyzes up to 10 seconds of earlier footage, which Google says improves visual consistency and narrative continuity between consecutive segments.

Frame and Reference Control

Developers also gained the ability to precisely set the first and last frame of a generated shot using the FIRST_FRAME and LAST_FRAME markers. The model then fills the space between them with a smooth transition, making it easier to build complex camera movements or create video loops where the final frame flows seamlessly back into the first.

Another new feature is support for reference material: up to three video clips, each up to three seconds long, can now be attached to a generation request, which helps preserve the consistent look of a character or object across shots. One remaining limitation is that footage cannot be inserted into the middle of an existing clip or added at the beginning, the model can only continue a scene from its end.

Cheaper Drafts, Pricier Final Output

Google also introduced a 360p draft mode designed for quickly testing ideas before generating the final version. Such a preview is produced up to 60 percent faster than a standard 720p video and costs roughly a third as much. At the other end of the scale is 4K output, achieved by upscaling the source material, intended for publication-ready production.

In the API, input costs $1.50 per million tokens, regardless of whether it is text, image, video, or audio. Output costs $9 per million text tokens and $17.50 per million video tokens, which in practice works out to roughly $0.10 per second of 720p footage.

Integrations with Creative Tools

The new model launched immediately in external creative platforms. Adobe added it to Firefly, and Figma integrated it into its Weave tool. Runway incorporated Gemini Omni 1.1 Flash into its existing prompt-, image-, and video-based workflows, while GMI Cloud provides centralized access to the model for creators using its infrastructure.

With scene extension, richer reference material, and 4K resolution, Gemini Omni Flash lets teams move from generating videos to actually directing them - Itay Schiff, Creative Director at Figma Weave

That reliability matters more than any single feature. Omni has made AI-generated video a viable option for a segment of creators who couldn't previously rely on it - Louisa Guo, VP of Marketing at GMI Cloud

Limitations and Availability

The model does not yet support system instructions, generation temperature control, or negative prompts, and audio track editing remains unavailable. Full language support is currently limited to English, and source material cannot be pulled directly from YouTube URLs. For individual creators and businesses, this means Gemini Omni 1.1 Flash is best suited to short promotional and social media formats rather than full-length productions with sound design.

For Polish advertising agencies and video content creators, the update means cheaper and faster campaign prototyping, thanks to the 360p draft mode, before committing to generating an expensive full-resolution version. Integration with Adobe Firefly and Figma Weave, tools already widely used in Polish design studios, could accelerate adoption of this technology among local creative teams.