On Tuesday, Google initiated the rollout of its Veo 2 artificial intelligence model to the Gemini chatbot platform. This advanced feature will be available exclusively to paid subscribers of Gemini, allowing eligible users to create eight-second videos based on natural language text prompts. Veo 2 was first unveiled in December 2024, serving as an upgrade to the previous Veo AI model, and it is also integrated with Google’s Vertex AI platform, contributing to features such as YouTube’s Dream Screen.
Veo 2 in Gemini Can Generate 720p Videos
In a recent blog post, Google outlined the global rollout of this video generation capability in Gemini, available in all supported languages. This feature will not be available to users on the free tier.
Users can select Veo 2 from the model picker menu within the Gemini web client and mobile applications. The model is capable of producing an eight-second video using text prompts, outputting the content in 720p resolution with a 16:9 aspect ratio, and allowing for downloads in MP4 format. Google has implemented a monthly limit on the number of videos a user can create, with notifications for users approaching this threshold.
Moreover, Google has introduced sharing options for the generated videos. On the Gemini app, users can easily upload their creations to social media platforms, including TikTok and YouTube, by utilizing the share button. Although the feature is currently rolling out, it may take several weeks before all eligible users worldwide can access it.
Upon the model’s introduction, Google highlighted that Veo 2 showcases significant improvements in realism and detail. It has the capability to interpret cinematographic terminology, allowing users to include specific camera details, such as lens type, movements, and cinematic effects, within their text prompts, which the AI can then follow.
Additionally, Google stated that the video generation model has been designed to minimize “hallucinations,” which means users can expect a decrease in anomalies like extra fingers, distorted limbs, or unanticipated objects appearing in their videos.
