Availability and evidence
Genie 4 entered a research preview for trusted testers on October 27. Google described simulations that users could steer with natural language, with native audio including ambient sound, music and dialogue. Gemini models assist scene coherence and reasoning. [1][2]
The preview did not establish general availability. Google said AI Ultra subscribers in Project Genie and early-access Gemini API and AI Studio partners would receive access later. It also acknowledged drift in visual detail and object permanence, along with limitations in physics, audio accuracy and long-term consistency. [2]
Sources
- We’re announcing Genie 4, our new world model for real-time interactive simulation.
Google · Oct 27, 2026
- Genie 4: A new generation of world simulation
Google DeepMind · Oct 27, 2026
- Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering
TechCrunch · Oct 27, 2026