Google's Omni 1.1 Flash brings 4K video and longer scenes

If you have spent any time trying to generate video with AI, you know the frustration: keeping a character consistent for more than a few seconds usually feels like rolling dice.

Google rolled out Gemini Omni 1.1 Flash, pushing its video generation capabilities straight into production-ready territory. Available immediately through the Gemini API in Google AI Studio, as well as in Google Flow and the Gemini app for paid subscribers, this update turns unpredictable video generation into a practical pipeline.

Giving AI video a real memory

The biggest leap here is visual consistency over time.

Previous video models typically only referenced the final frame or the last second when continuing a shot, which almost always led to shifting backgrounds and morphed faces. Omni 1.1 Flash fixes this by analyzing up to 10 seconds of prior footage. You can now stretch videos in 10-second increments up to a total length of 40 seconds, keeping characters and environment details intact throughout the sequence.

Another huge addition for creators is start-and-end frame interpolation. Instead of guessing how the camera will move, you can lock in your opening and closing frames. The model then fills in the continuous motion between them, making complex camera pans, zooms, and seamless loops genuinely feasible.

Google also added multimodal video referencing, letting you feed the model up to 3 seconds of reference video. If you want a specific character to mimic dance moves or gestures from a real clip, the model can translate those exact movements into the final generation.

Fast drafts in 360p, final exports in 4K

Google clearly realized that rendering high-resolution experiments is both too slow and too expensive for serious workflows.

To fix that, Omni 1.1 Flash introduces a lightweight 360p draft mode. Generating in 360p runs up to 60% faster and costs just one-third of the standard 720p render. You can rapidly storyboard on the go in the Google Flow app or test API logic in minutes, pick the best variation, and only then commit computational resources.

When you are ready for final delivery, the model supports upscaling to crisp 1080p or 4K resolution suitable for professional post-production. Early adopters like Adobe Firefly, Figma Weave, and Runway are already plugging the model into their production pipelines.

Why it matters

We think Omni 1.1 Flash marks an important shift in generative video from party tricks to dependable tooling.

For creators and developers, this changes the day-to-day math. The ability to churn through rough cuts at a fraction of the cost, anchor key camera positions with first and last frames, and stretch clips to 40 seconds without continuity collapsing removes the biggest headaches in AI video. It is cheap enough to experiment with right now and precise enough to build into automated production pipelines.

For the broader industry, Google is shifting the competitive ground. Rather than just bragging about raw visual fidelity, it is packaging speed, practical controls, and tiered pricing into developer-friendly APIs. That will force rivals to rethink their own developer ergonomics and render costs.

A 40-second ceiling is still short of full cinematic production, but for social media, ad creatives, and rapid prototyping, Google just handed builders a much sharper toolkit.