News, Service

Google Unveils Video Generation Model 'Gemini Omni'

May 20, 2026

Google unveiled a new family of generative AI models called 'Gemini Omni' at Google I/O 2026. The first model is 'Gemini Omni Flash.' Google describes this model as one that can create whatever it wants from any input, and it initially offers video creation and editing capabilities.

Gemini Omni can use text, image, audio, and video inputs together. Users can upload existing videos and edit them using natural language, such as changing scenes, modifying backgrounds, or adjusting camera angles. Google introduced maintaining character consistency, physical naturalness, and scene flow while preserving the context of previous requests as a core feature.

While the existing video generation model Veo was primarily focused on text-based video generation, OmniFlash differs in that it can take video itself as input to create new videos. According to The Verge, OmniFlash can currently generate videos up to 10 seconds in length that include audio, and Google is preparing to support the generation of longer videos.

Google positions Gemini Omni as akin to a "video version of Nano Banana." While Nano Banana demonstrated Gemini's multimodal capabilities in image generation and editing, Omni extends that direction to video. Google DeepMind explained that Omni constructs scenes by leveraging Gemini's knowledge of physics, history, science, and cultural contexts.

The rollout is also extensive. Gemini OmniFlash is available on the Gemini app, Google Flow, and YouTube Shorts. Google AI Plus, Pro, and Ultra subscribers can use it on the Gemini app and Google Flow, while it will be available for free on YouTube Shorts and the YouTube Create app starting this week. APIs for developers and enterprise customers are also scheduled to be provided.

Content credibility mechanisms are also included. Google DeepMind stated that content generated or edited with Gemini Omni includes SynthID digital watermarks and C2PA content credentials. Users can verify creation within the Gemini app, and verification features will be supported in Chrome and Google Search in the future.