Genie 4, DeepMind’s interactive world model, is now in the Gemini API and AI Studio for paid developers. It streams 720p worlds with audio, for up to 120 seconds a session.
A day after launching Genie 4 for AI Ultra subscribers, Google opened it to developers in the Gemini API and AI Studio. It generates an explorable video world in real time, with sound, dialogue and live typed changes.
It streams 720p at 24 frames per second for about $0.10 per second, around $6 a minute, on the paid tier only. Sessions are capped at 120 seconds, and billing continues until you end the session.
Key Capabilities:
- Inputs: A text prompt, one 16:9 reference image, keyboard controls and live text changes.
- Where: The Interactions API in the Gemini API, plus Google AI Studio, served from the US, Western Europe and Japan.
- Consumer side: Project Genie opened to AI Ultra subscribers on Nov 18, and some AI Pro users are coming off the waitlist.