At yesterday’s Google I/O 2026 event, the company not just unveiled Gemini 3.5 Flash, it also introduced probably one of the wildest Gemini upgrades we have seen in years. Here, I am talking about the Gemini Omni model.
Gemini Omni is not a normal AI model, here’s why
Google says that the new model can create videos from almost any kind of input, starting with videos. The first release under this lineup, Gemini Omni Flash, is rolling out globally to everyone right now. In case you missed yesterday’s keynote and wondering what exactly makes Gemini Omni one of the most capable, or as I mentioned, wildest Gemini upgrades yet?
All this while, we have come across AI video generators that mostly focus on text prompts, but Gemini Omni goes well beyond that. In simple words, you can now throw different types of media into Gemini Omni and ask it to generate an entirely new video around it. Crazy, right?
Google also showcases some prompts where users transformed sculptures into flowers, turned mirrors into liquid effects, or completely changed environments while maintaining the original subject structure. Here, it is worth noting that Google is also focusing on “world knowledge” with Omni. The company claims that the model understands concepts like gravity, kinetic energy, fluid dynamics, and even historical or cultural context.
Also read: Google Aims to Reinvent Online Shopping with Universal Cart This Year; Here’s How
If you are someone who likes diving deep into scientific topics, Gemini Omni is perfect for you. One of the highlights Google showcased was a claymation-style protein folding example created entirely through prompts. For those wondering what else Gemini Omni can do, well, the model can also work with user-uploaded references. For example, creators can combine images, audio clips, and videos together into a generated sequence. Google says even voice-based references are supported initially, while broader audio input support is coming later down the line.
Google integrates Gemini Omni to YouTube Shorts
As part of the latest integration, Google is pushing Gemini Omni Flash to YouTube Shorts and the YouTube Create app, which, by the way, is free for now. Meanwhile, Gemini app subscribers on AI Plus and other plans will get access to Gemini Omni globally.
Not to forget, the company has also teased a new feature called AI Avatars, powered by Gemini Omni. You can create digital versions of yourself using your own voice and likeness for generated videos. The company says safeguard policies and SynthID watermarking are built into every Omni-generated clip.
