Google’s latest Gemini Omni model can turn real videos into surreal dreams

Video creation has been one of the most compelling creative uses of AI. Among the platforms that have helped fuel this phenomenon is Google’s Veo, especially Gen 3, which has proven incredibly powerful at creating entire scenes with consistent elements and near-perfect lip-sync. While Veo 3 (and the new 3.1) is limited to creating purely AI-generated videos with text and audio, Google is introducing a new model at Google I/O 2026 that goes a step further by letting you modify real-life footage into stunning clips.

Gemini Omni is Google’s new class of multimodal models that can improve real footage into something that would likely only exist in your mind if it weren’t for AI. It’s coming first in the form of Omni Flash, which, Google says, can combine multiple forms of input — text, audio, statics, and video — to produce something radically different in any of these formats. However, it’s starting with video, where users will be able to create videos as wild as their wildest dreams, while ensuring character stability across multiple frames.

Gemini Omni not only produces videos, but it also essentially considers the story by analyzing multiple aspects simultaneously. What’s even better is that you can refine the videos it creates using natural language signals until you find one that matches your vision (or even better).

With Gemini Omni, you can get started by adding your context to images, audio or video, or simply use text to describe your vision. You can also upload a digital version of yourself and use avatars to create videos with characters that look and sound like you.

Google also emphasizes that the videos will follow real-life physics, thanks to the model’s understanding of gravity, kinetic energy, and even fluid dynamics. We’ll have to wait to see if this actually applies to real-life outputs.

Google has shared two examples to demonstrate the Gemini Omni’s capabilities, one featuring comedian Adam Waheed and the other featuring YouTuber Happy Kelly. Along with Adam Waheed’s video at the top, here’s the video featuring Happy Kelly:

Gemini Omni is launching with its new Flash model in the Gemini app and will be available to all paying customers of Google AI Plus, Pro and Ultra tiers. Omni will also be available through Google’s AI film-making tool Flow.

If you want to try it out for free, Omni will also be available through YouTube, where you can create remixes of existing shorts. Apart from the regular YouTube app, it will also be available in YouTube Create. However, Google has not yet confirmed whether creators will be able to limit or restrict the remixing of their content using AI.