Genie 4

Genie 4 is Google DeepMind's natively multimodal world model, announced on 27 October 2026. Instead of rendering a fixed video clip, it builds a playable 3D environment you can steer in real time with plain language, complete with physics, ambient audio, and characters that answer you. Genie 4 is built on Gemini: Gemini Omni grounds the visuals, and Gemini 3.8 Flash runs a real-time agent loop that observes, plans, and replans.

genie4.art is an independent, vendor-neutral guide to the model. It covers what Genie 4 does, how it differs from Genie 3, frames from the official demos, the staged access rollout through trusted testers, AI Ultra via Project Genie, and the Gemini API, plus practical prompting notes for building usable worlds. It also states plainly what Google has not published, such as frame rate, resolution and pricing.