Google may showcase a new video-generation model called Omni at I/O 2026

Google launches new Gemini Omni model – text-to-video generation

Recently a new version of Google's Gemini Omni appeared in closed demos. It has already shown impressive results: users can create short videos, edit them directly in the chatbot window, and use ready-made templates.

Key features

* Video remixes – the model can combine existing videos and add new elements to them.
* In‑dialog editing – all changes happen right in chat, without needing to switch to a separate editor.
* Templates – prewritten scenarios allow quick content generation.

Within the Gemini app, metadata indicates that Omni is an extension of the Veo platform. It’s still unclear what role it will play in Google’s overall product ecosystem.

Test clips

1. Professor on the board – AI generated a scene where a teacher writes a mathematical proof, correctly displaying formulas and the sequence of explanations.
2. Two men over spaghetti by the sea – the second scene looks fairly realistic but contains typical artifacts of generative models.

Both tests used 86 % of the daily limit for AI Pro tier accounts.

Product status

The model is not yet available to the general public. However, Google has confirmed its commitment to video generation, especially after a competitor discontinued support for Sora. More detailed information about Gemini Omni and other generative media innovations is expected to be presented at Google I/O 2026, where key platform updates are traditionally announced.