Google is now letting Gemini Advanced subscribers test its powerful Veo 2 text-to-video AI model, which can create short, high-resolution videos with stunning “cinematic realism.”
Starting today, users can choose Veo 2 from the Gemini model menu on both web and mobile and enter a prompt to generate an eight-second video in 720p resolution.
Each user will have a monthly limit on how many videos they can generate. Google says it will alert users as they near that limit. Videos are downloadable in MP4 format, and mobile users can instantly share their creations on platforms like TikTok and YouTube via a new share button.
Veo 2 is designed to understand real-world physics, lighting, and human movement better than before. This allows it to create lifelike scenes, smooth character animations, and more detailed visuals. All videos come with a SynthID watermark to show they were AI-generated.
In addition to Veo 2, Google also introduced Whisk Animate, a new tool that turns static images into animated eight-second videos. This feature is now available to Google One AI Premium subscribers globally through Google Labs.
Google first introduced Veo 2 in December via its Vertex AI platform, where it cost 50 cents per second to use. The model has since been integrated into YouTube’s experimental Dream Screen feature, allowing creators to produce AI-powered Shorts videos.
