Source time Edition date
Close readGoogle DeepMind
Genie 4: A new generation of world simulation
Genie 4：新一代世界模拟
Genie 4 promises unprecedented real-time realism, but readers should weigh how much compute such world models consume against the practical limits of deploying them.
Judgment
Teams evaluating interactive-world tooling must decide whether to join the Genie 4 waitlist and build against a trusted-tester preview, accepting limited access and undisclosed real-time compute costs, or wait for broader consumer and developer release.
Basis
Raia Hadsell, Diego Rivas and Hang Qi frame Genie 4 as a natively multimodal world model that generates music, ambient noise and dialogue alongside the simulation, with Gemini Omni anchoring appearance and other Gemini models supporting coherence. They say it is rolling out first as a research preview for trusted testers, then to Google AI Ultra subscribers in Project Genie and early-access partners of the Gemini API and Google AI Studio. The mechanism — real-time world plus audio generation — is what makes deployment cost and access the gating issues.
Unclear The source gives no compute requirements, latency figures, pricing, or hardware specifications for running Genie 4 in real time.
Next Check whether the Genie 4 waitlist page specifies Project Genie access terms for Google AI Ultra subscribers or any published usage limits.