What you need to know
- Google teased Gemini 2.5 Flash at its Cloud Next 2025 conference, saying the model is “coming soon.”
- The company’s Veo 2 video-generation model is becoming widely available as part of the Google AI Studio.
- The Live API for Gemini is also becoming available in preview for real-time interactions.
Google Cloud Next 2025 is officially here, and the first set of developer-focused announcements haven’t disappointed. This week, Google is announcing that its Veo 2 video-generation model is production ready and widely available for use via the Gemini API in Google AI Studio. Additionally, the company confirms that Gemini 2.5 Flash — a streamlined thinking model for basic tasks — is “coming soon.”
As a video-generation model, Google says that Veo 2 is “able to follow both simple and complex instructions, as well as simulate real-world physics in a wide range of visual styles.” Veo 2 will work with both text or image prompts, or both. This means you could describe to Veo 2 what you want your AI-generated video to turn into, provide the model with an image to work with, or a blend of the two options.
Users can try out Veo 2 in the Google AI Studio, and it can produce videos at 720p resolution at 24 frames per second. Currently, the big limitation is length. Videos created with Veo 2 can only be a maximum of eight seconds long. As an enterprise-grade tool, Veo 2 costs $0.35 per second of video generated.
As an example of what the model can do, Google shows off what Wolf Games is doing with Veo 2 in the video below.
