Gemini Omni 1.1 Flash Launches with 40-Second Video Extension and

Google has launched Gemini Omni 1.1 Flash, a new iteration of its generative AI model designed specifically for video creation and editing. The update introduces several advanced capabilities: extending existing video clips, generating intermediate frames between two specified keyframes, using short reference videos during generation, and producing output at resolutions up to 4K.

Extended Video Understanding and Generation

Gemini Omni 1.1 Flash can now analyze up to 10 seconds of preceding footage when extending an existing video — a significant upgrade from earlier versions that relied solely on the final second of input. Developers may extend clips in increments of 10 seconds, with a maximum total extension duration of 40 seconds. This expanded temporal context allows content creators to complete scenes while preserving visual continuity and narrative coherence.

Keyframe Interpolation for Seamless Motion

A core new feature is the ability to define a start frame and an end frame, then instruct the model to generate all necessary intermediate frames to bridge them. The system handles complex camera motions — including object rotation and zoom — and supports transitions that produce continuous or loopable sequences. This shifts creative control from full text-to-video generation toward targeted outcome specification.

Reference-Based Video Synthesis

The model supports using reference videos up to three seconds long to guide new video generation. This functionality helps retain specific visual attributes from the reference — such as character appearance and environmental details. Google demonstrated the use of multiple reference clips within a single continuous output to reconstruct distinct movements across one unified video sequence.

Low-Resolution Drafts for Rapid Prototyping

Google has added a 360p output option to accelerate early-stage experimentation. According to the company, generating drafts at this resolution is up to 60% faster than at 720p and costs approximately one-third as much. This tier is intended primarily for rapid prototyping and storyboarding workflows, with final outputs producible at 1080p or 4K after scene approval.

Access and Integration Pathways

Developers can access Gemini Omni 1.1 Flash via the Gemini API in Google AI Studio. Enterprise customers may integrate it through the Agent Platform API. Google has published documentation, usage guides, ready-to-run examples, and prompt-writing instructions tailored to video generation tasks. The model is globally available to subscribers of Google’s AI Plus, AI Pro, and AI Ultra plans through Google Flow and the Gemini mobile application. Scene extension functionality is also accessible within the Gemini app for those same subscribers.

Google confirmed that Gemini Omni 1.1 Flash has already begun integration into video production pipelines at several companies and platforms, including Adobe, Figma Weave, GMI Cloud, and Runway. The release reflects Google’s strategic shift toward AI tools that go beyond ground-up video synthesis — instead emphasizing comprehension of existing footage, intelligent scene completion, precise motion and transition control, and consistent visual-spatial fidelity.