Google has opened Genie 4, its interactive world model, to developers through the Gemini API and Google AI Studio, a day after the model reached Google AI Ultra subscribers inside Project Genie. Developers can now build with a model that generates explorable worlds in real time, rather than watching demonstrations of one.
Starting from a text prompt, a reference image or both, Genie 4 streams continuous 720p video at 24 frames a second with synchronised native audio, from ambient soundscapes to character dialogue. Users move through the scene with keyboard controls and can type new prompts mid-session to change the weather, lighting, objects or events, while Gemini reasoning keeps the world state consistent.
The price is about $0.10 for every second of streamed output, with stereo audio included. Sessions are capped at 120 seconds, so a maximum-length session costs roughly $12, and access requires a paid-tier Gemini API key. Every output carries a SynthID watermark, and prompts showing real people, public figures or copyrighted brand imagery can be blocked or cut off mid-stream.
Google names three uses for the API: prototyping interactive experiences, visualising designs for spatial concepts, and training robotics and AI agents. The format has limits: output is fixed at 16:9 landscape, portrait video is not supported, and Google warns that visual stability may degrade after 60 to 90 seconds.
World models are fast becoming a new creative category, sitting alongside video generators such as Runway and Pika and image tools like Midjourney. For developers curious about Google’s wider AI line-up, our Gemini page covers the model family behind it, and the Gemini vs Copilot comparison shows how it stacks up against Microsoft’s assistant.
Source: PPC Land