Google extends access to VEO 3, its viral new video model, via the Gemini app

About 100 hours after the first start, Google opens the AI ​​video model VEO 3 for users in 71 additional countries.

The news came from Josh Woodward, Vice President of Gemini on Google, who stated the announcement via X. The EU countries are not included in the rollout for the time being.

Gemini Pro subscribers receive a test package of ten VEO 3 generations via the web interface. These ten generations are a unique range of tests.

Users with an ultra subscription of $ 250 per month receive the maximum number of generations that Google allows, including daily updates. In Flow mode, which is particularly aimed at AI filmmakers, Ultra users receive 125 generations per month, and Pro -subscribers receive ten generations per month.

There are still some restrictions. At the moment, VEO 3 only works in the Gemini Pro web version and currently only supports the English audio edition – although other languages ​​can occasionally appear. The flow mode does not support the voice output when users upload their own pictures.

Precise input requests, viral videos

According to the noise audio overviews of the last year, VEO 3 is also developing with a limited access to Google's second AI virus sensation. Users flood social media with demo videos that show how the combination of audio and video sets a new quality standard for AI-generated content. The Google model also follows the input requests with impressive precision.

Entry request: “The camera follows a dachshund that runs through a living room and an open front door and on a veranda. It stands on the upper staircase with a view of the neighborhood while an ice cream truck drives past.” | Video: Nick Matarese about X

VEO 3 also makes it easier to create false content that looks real and sounds. In an example, users generated a fake interview in a fictional car show – something that could just as easily be applied to invented protest material or other misleading materials.

Video: László Gaál about X

VEO 3 confirms concerns about the role of generative AI in the spread of disinformation and shows how far the technology has come. A few years ago, creating a “Deeppake” – even the face of someone else in one video – required hours and serious technical skills. Today a single line of text generates realistic scenes with pictures and tones. This makes it more important for people than ever to check visual content from sources that have not been observed.