Google DeepMind Releases Genie 4 World Model With Real-Time Audio and Mid-Simulation Editing
Genie 4 doubles scene consistency to two minutes while enabling real-time audio and freeform edits to weather, camera angles, objects, and people. Google also tested Gemini 4 Argon directing simulations for potential robot-scenario testing before real-world deployment.
Google DeepMind released Genie 4, its latest AI world model, in research preview on October 27, 2026. The model generates more photorealistic and interactive simulated environments than its predecessor, supports real-time audio, and allows users to type text prompts mid-simulation to alter scenes.
Genie 4 introduces several capabilities not present in Genie 3, which launched as a research preview roughly a year earlier. Users can make freeform edits during a simulation — changing weather, camera angle, or adding objects and people — while the model maintains a consistent scene for up to two minutes, double the previous limit. Audio generation includes music, ambient sounds, and interactive dialogue that matches the simulated environment. The model also features improved physics and more human-like characters.
Google’s Gemini AI model is more tightly integrated into Genie 4, functioning as a director that plans steps and checks frames while Genie handles visuals. In a research experiment tied to Gemini Robotics, a Gemini 4 Argon model broke down tasks into steps that Genie 4 simulated in real time — a potential method for testing robot scenarios before real-world deployment.
“What we’ll increasingly see in the future are these hybrid world models, where you have a combination of Gemini writing code and text, and Genie ingesting that in real time and simulating reality at different levels of abstraction,” said Tejas Kulkarni, research director at Google.
Early testers have flagged film prototyping as a possible use case, alongside the previously pitched applications of gaming, interactive learning, and robotics. Google DeepMind engineer Diego Rivas noted that the real-time capability is particularly useful “in the ideation phase, in the storyboarding phase.”
The team acknowledged current limitations: details and objects can drift during longer simulations, and audio accuracy and in-world text legibility remain unsolved. Genie 4 will roll out to Google AI Ultra subscribers via Project Genie and early access partners of the Gemini API and Google AI Studio in the coming weeks.