Google introduced a new AI video-generation model, Gemini Omni Flash, at Google I/O 2026, positioning it as a major upgrade over its existing generative video tools, including Veo and Nano Banana.
The new model is part of Google’s newly announced Omni family and combines Gemini intelligence with advanced world understanding to create more accurate and editable AI-generated videos. According to Google, Omni Flash has a stronger understanding of movement, physics, objects, and scene continuity compared to its earlier models.
During the keynote presentation, Google DeepMind CEO Demis Hassabis demonstrated the new model live on stage, showcasing how it can generate detailed videos from simple prompts and transform existing footage into stylized AI-created worlds.
“Google Omni combines Gemini Intelligence with the best of our models for a new level of world understanding,” Hassabis said during the presentation.
He added that the model performs better than Google’s previous video-generation systems because it understands “complex ideas, for example, how kinetic and potential energy work.”
Google says this improved understanding allows the AI to create more realistic movement, object interactions, and scientifically accurate scenes than earlier models, such as Veo and Nano Banana.
Google Omni Flash: What is it and how does it work
Gemini Omni Flash is designed to generate and edit videos using natural conversational prompts. Unlike traditional AI video tools that rely on rigid commands, Omni Flash allows users to describe scenes in everyday language and refine outputs interactively.
During the demonstration, Hassabis showed a prompt asking the AI to create a “claymation explainer of protein folding, everything is made out of clay, no hands, stop motion, accurate.” The model then generated an astylized stop-motion-style educational video entirely from the text instructions.
Google says Omni Flash has a deeper understanding of physics, motion, and real-world behavior, helping it produce videos that appear more natural and coherent.
The company also demonstrated how users can upload their own videos and conversationally modify them. According to Hassabis, Omni Flash can “change your reality into something cool” based on prompts while preserving the original motion and structure of the footage.
Google showcased examples where the model adjusted artistic styles, inserted new visual elements, and transformed environments into entirely different worlds.
Hassabis also said the Omni model offers “a more natural way to edit videos with conversational language,” reducing the need for complex editing workflows.
Google confirmed that Gemini Omni Flash is currently available in the Gemini app, Google Flow, and YouTube Shorts, while API support for developers is expected to arrive soon.
Google strengthens SynthID to help identify AI-generated videos
Alongside Omni Flash, Google also announced major updates to SynthID, the company’s AI watermarking and content identification system designed to help users identify AI-generated media.
During the presentation, Google revealed that SynthID is expanding through partnerships with several major AI companies, including NVIDIA, OpenAI, Kakao, and ElevenLabs.
Google says the upgraded system will make it easier to identify whether videos, images, audio clips, or text were generated or modified using AI systems such as Gemini Omni Flash.
The company explained that SynthID embeds invisible digital watermarks directly into AI-generated content without affecting media quality. These markers can later be detected to verify whether content was created using participating AI tools.
Google positioned the expansion as part of a wider effort to improve transparency around generative AI as AI-created videos become increasingly realistic and difficult to distinguish from authentic footage online.
Google Omni Flash: Launch timeline
Google confirmed that Gemini Omni Flash is available starting today inside the Gemini app, Google Flow, and YouTube Shorts.
The company also announced that API access for developers will arrive “soon”, though an exact release timeline has not yet been shared. Gemini Omni is rolling out today to all Google AI Plus, Pro, and Ultra subscribers.
Hassabis added that more details about the upcoming Omni Pro model will be revealed later.
