Generative AI models have traditionally specialized in specific tasks — excelling at text generation, image creation or video production, but rarely all three at once.
This week at Google’s I/O conference, the search giant unveiled an AI model it claims can do all three well: Gemini Omni, “where Gemini’s ability to reason meets the ability to create.”
Omni can create explainers, stylized clips and simulations grounded in “real-world knowledge” such as physics and cultural context.
“Omni has an improved intuitive understanding of forces like gravity, kinetic energy and fluid dynamics, allowing you to create more realistic scenes,” wrote Koray Kavukcuoglu, Google’s chief AI architect and CTO of Google DeepMind, in a blog post.
Google said the model also allows users to edit videos through natural language prompts while maintaining scene consistency, physics and character continuity across multiple edits.
The first release in the Omni model family is Gemini Omni Flash, which is a lightweight, faster version optimized for speed and responsiveness rather than maximum quality or reasoning depth.
Omni Flash is rolling out this week through the Gemini app and Google Flow for Google AI Plus, Pro and Ultra subscribers globally. It is also launching on YouTube Shorts and the YouTube Create App now. Developers and enterprise users will get access in coming weeks through APIs.
Google said all videos generated with Omni will include its “imperceptible” SynthID watermarking technology to identify it as AI in Google Search, Gemini app and Gemini in Chrome.