Google DeepMind has released the generative video model "Gemini Omni 1.1 Flash." This model is an update designed for practical operation in production environments.
As a new feature, a mechanism to extend scenes by inheriting existing video has been added. The model can analyze up to 10 seconds of footage as past context. This is a significant improvement over previous models, which could only reference the final one second.
This makes it easier to maintain visual consistency and narrative coherence. Scenes can be extended in 10-second increments, with a maximum total length of 40 seconds.
Additionally, a feature has been implemented to generate smooth transitions by specifying start and end frames. Furthermore, the model now supports high-quality output in 4K resolution.
Developers can also establish a workflow to quickly create low-resolution drafts to validate ideas. This is expected to increase prototyping speed and reduce costs.
Access is available via Google AI Studio or the Gemini Enterprise Agent Platform. Integration into generative video workflows and media editing software is supported through the Gemini API.
Source: Gemini Omni 1.1 Flash (HN 296pt, 236 comments) (HN Search (backfill), 2026-08-28)