Google DeepMind has announced Genie 4, a world model that produces interactive simulated environments in real time. The post is dated October 27, 2026.
What it does
Genie 4 is natively multimodal. Music, ambient sound and spoken dialogue are generated together with the world itself, and characters can react to user input or to events in the scene. The user steers the simulation with keyboard controls for movement and camera, plus text prompts that change things such as the weather or add a person or object.
DeepMind says the model builds on the Gemini family. Gemini Omni grounds the simulation in real-world appearance and behavior, while other Gemini models handle the reasoning that keeps the scene coherent. The company highlights improved photorealism, more natural human movement and physics such as aerodynamics and collisions.
Gemini as a planner
The most relevant part for anyone thinking about agents is the pairing with Gemini. In the described setup, Gemini 3.8 Flash acts as a guiding agent: it watches the generated frames, splits a goal into steps, checks whether each step succeeded and revises the plan if the world changes or the user intervenes. In a robotics example, Gemini 4 Argon breaks a task into steps that are fed to Genie 4 as text captions. The demonstrations include a Universal Robots UR5 handling a pipette and Apptronik's Apollo humanoid playing checkers. DeepMind frames the goal as testing how AI systems handle different environments before real-world use.
Availability
Genie 4 is a research preview for trusted testers, who come from creative industries, journalism, education, gaming and engineering. It is described as coming soon to Google AI Ultra subscribers in Project Genie and to early access partners of the Gemini API and Google AI Studio. A waitlist for Project Genie is open. The source gives no pricing and no date for broader release.
DeepMind says enterprise partners are evaluating the model for fast scenario generation and are identifying gaps. Creative teams are using it for pre-visualization, and game developers for quick prototypes.
Limits and safeguards
DeepMind calls Genie 4 early and experimental. Visual details and object permanence can drift in longer sessions, after many prompt edits or while moving through a scene. Complex physical interactions, real-time audio accuracy, legible text, spatial accuracy and long-term consistency remain open research areas. All video and audio it produces carries SynthID watermarking, and the company says it ran internal safety evaluations before the preview.
What is still unknown
The source does not say how Gemini API access will be priced, what usage limits will apply or when partners will get it. Those details decide whether this becomes something an operations team can use.
Top comments (0)