Google Launches Gemini Omni AI Model For Video Generation And Editing: What You Need To Know

Google has officially launched Gemini Omni, a new generative AI model that can create and edit high-quality videos using a combination of text, images, audio, and video as input, marking a significant expansion of the company's multimodal AI capabilities.

The announcement was made on May 19, 2026, by Koray Kavukcuoglu, Chief Technology Officer of Google DeepMind and Chief AI Architect at Google, via the company's official blog.

What Is Gemini Omni?

Gemini Omni is described by Google as a model where Gemini's reasoning ability meets the ability to create. It is designed to accept any combination of input types, text, images, audio, and video, and produce video output that is grounded in Gemini's real-world knowledge across history, science, and cultural context.

The first model in the Omni family, Gemini Omni Flash, is the version launching today. Google says further models in the family will follow, with support for additional output modalities such as image and audio generation planned for later releases.

Conversational Video Editing

One of the standout features of Gemini Omni is its ability to edit video through natural language conversation. Users can issue sequential instructions that build on one another, adjusting camera angles, transforming environments, modifying objects, or altering on-screen action, without losing consistency between edits. The model maintains character continuity and scene coherence across multiple editing turns.

Physics Understanding and World Knowledge

Google states that Gemini Omni features an improved intuitive understanding of physical forces including gravity, kinetic energy, and fluid dynamics, enabling more realistic video scenes. Beyond visual realism, the model can draw on Gemini's broader knowledge base to produce meaningful, contextually accurate content, including educational explainers generated from short descriptive prompts.

Multi-Input Creation

Gemini Omni can accept multiple reference inputs simultaneously. Users may combine a character image, a motion reference video, an audio track, and a text prompt in a single request to produce a unified video output. Google noted that voice references for audio generation are supported at launch, with additional audio input types to follow.

Safety, Watermarking, and Responsible AI

Google confirmed that all videos generated with Gemini Omni will carry an imperceptible SynthID digital watermark. Users can verify AI-generated content through the Gemini app, Gemini in Chrome, and Google Search. The company also confirmed support for Avatars, a feature that lets users generate videos with their own digital likeness and voice. Google noted that broader speech and audio editing capabilities are still being reviewed for responsible deployment.

Availability

Gemini Omni Flash is rolling out globally today to Google AI Plus, Pro, and Ultra subscribers via the Gemini app and Google Flow. It is also available at no charge to users of YouTube Shorts and the YouTube Create App. Developers and enterprise customers can expect API access in the coming weeks.