Google ships Gemini Omni 1.1 Flash with 4K upscaling and 40-second scene extension

Google has shipped Gemini Omni 1.1 Flash, a major update to its Omni generative video model that Google says makes it ready for professional work. According to the company's developer blog, the release is available immediately through the Gemini API in Google AI Studio and targets developers building generative video workflows, creative tools and media editing software.

Scenes can now run to 40 seconds

The headline change is scene extension. Rather than producing isolated clips, Omni 1.1 can carry on generating footage from the point where an existing video ends. When extending a scene, the model takes into account up to 10 seconds of prior context, and Google notes that earlier versions looked only at the final second of a clip.

That added memory matters for coherence. Google claims it yields more consistent visuals and better story continuity, which makes it practical to assemble longer narratives or branch off in new creative directions. Extensions run in 10-second increments, up to a cumulative maximum of 40 seconds per video.

Keyframe control and video references

Omni 1.1 also introduces first and last frame interpolation: a developer supplies a starting frame and an ending frame, and the model fills in continuous footage between them. Google points to orbiting camera moves, zooms and loops that need clean endpoints as natural use cases.

For consistency across shots, the model now accepts up to three seconds of video as a reference within multimodal prompts. That lets a generation stay aligned with existing footage, for instance mapping reference dance performances onto supplied characters inside one continuous take.

Cheaper drafts, 4K finishes

The update also sets up a two-tier resolution workflow. Lightweight 360p previews generate up to 60% faster, a figure Google bases on system throughput of 360p versus the standard 720p, and cost roughly a third as much. Google frames these as tools for fast prototyping and storyboard iteration.

For final renders, Omni 1.1 can output 1080p video or upscale to 4K, resolutions Google considers suitable for professional pipelines.

Where to get it

Omni 1.1 is rolling out across Google's developer ecosystem: directly in Google AI Studio, through the Agent Platform API for enterprises, and with supporting documentation, a cookbook and prompting guides covering scene extension, video references and upscaling.

On the consumer side, Google says Omni 1.1 is available globally to Google AI Plus, Pro and Ultra subscribers in Google Flow, with scene extension also live in the Gemini app for the same tiers. The announcement, authored by Google DeepMind product manager Alisa Fortin, also surfaced on the Hacker News front page.

Why it matters

Generative video has so far been held back by short clip lengths and weak consistency between shots. This update attacks both limits directly: a 10-second context window, 40-second cumulative scenes, keyframe-bound interpolation and reference video together push the technology from novelty clips toward material that could sit inside a real edit.

The pricing split is telling too. Inexpensive 360p drafts paired with costly 4K finishes mirror how production pipelines already operate: iterate cheaply, and render at high resolution only when the shot is settled. And by shipping the same model to API developers and consumer subscribers on the same day, Google signals that it expects generative video to become a standard capability across its products rather than a niche developer offering.