A new video generation model is apparently coming to Gemini, with “Omni” producing some pretty impressive inital results.
Video generation is perhaps one the most impressive, but also most polarizing aspects of generative AI. Google has built out Veo as its video generation model for a while now, but seems to have something new in the pipeline.
At least one Gemini user was prompted to “Create with Gemini Omni,” which Google describes as follows:
Meet our new video generation model. Remix your videos, edit directly in chat, try a template, and more.
How “Omni” fits into the broader context of Gemini and Veo isn’t entirely clear at the moment, but metadata suggests “Omni” is an extension of Veo. But, regardless of that, the output here looks pretty impressive. One demo used input of “a professor writes out a mathematical proof for trigonometric identities on a traditional chalkboard, explaining the step he is currently on in the equation,” and while there are still some obvious tells in the final output, the video does a great job of handling the text while putting out a fairly realistic video.
Meanwhile, a second prompt asked for a scene of two men eating spaghetti – in reference to the Will Smith test – again with fairly realistic results. It’s nothing entirely groundbreaking, but the output is quite good.
The prompt here was: “Can you create a scene with two men at a table seaside at an upscale restaurant on outdoor deck seating. They are at a circular table with a nice white table cloth, and all of the fancy accessories, all the spoons forks and knives, fancy napkins, centerpiece. One man is Distinguished: A mature African-American man in his 50s with a short beard and confident posture, wearing a tailored, sophisticated suit, the other is is friend, both approaching the table to eat a plate of spaghetti. In the beginning the men approach the table, exchange brief niceties, and begin to eat the spaghetti calmly In between bites sharing conversation.”
A “usage” tab also showed up for this user, with these two prompts taking up 86% of daily usage on an AI Pro plan (though the user did say some usage on Gemini Flash during the same day).
We recently spotted Google’s intention to add more explicit usage limits.
Google hasn’t announced Gemini “Omni” yet, but previously said that “video’s here to stay” in commitment to the technology following the announcement that OpenAI would kill off video generation through its Sora model earlier this year. With I/O 2026 right around the corner, that’s probably where we’ll hear more about Google’s plans for Gemini and video generation.
More on Gemini:
Follow Ben: Twitter/X, Threads, Bluesky, and Instagram
FTC: We use income earning auto affiliate links. More.