Gemini has not yet released the Omni video model, but has pushed it to some users for testing. Testers have reported that the sound quality is far superior to Veo.

According to WatchTower Beating monitoring, with a week still to go before Google I/O 2026, Gemini's new video model "Omni" has already been unexpectedly discovered by users. Multiple Reddit users have reported that over the past week, when opening the Gemini App, a new video creation entry has repeatedly popped up, labeled on the interface as "Powered by Omni," appearing alongside the existing Veo 3.1 (internally codenamed Toucan).

One user who has actually tried it out gave high praise, stating that Omni is one of the best video models he has ever seen, with impressive keyword compliance and seamless multi-angle transitions. He specifically noted that the audio and environmental quality generated by Omni is more than a notch above the Veo series, and it even automatically adds background music that matches the scene. However, he also mentioned two obvious issues: the rate limit is extremely strict, with Pro subscription users exhausting 80% of their quota after generating only two videos; and celebrity portraits are still blocked, as the classic test of Will Smith eating spaghetti does not pass through.

Currently, Gemini's multimedia generation is divided: videos rely on Veo 3.1, and images rely on the Nano Banana series. If Omni is a unified model, it means that Google is consolidating text, image, and video generation capabilities into a single architecture. DeepMind CEO Hassabis publicly stated last year that Gemini and Veo would be merged, and Omni is likely the realization of this plan. Google has not yet officially confirmed this model, with an announcement expected at the I/O event on May 19.