Google DeepMind Rolls Out Genie 4 World Model in Research Preview

Google DeepMind announced Genie 4, its new world model for real-time interactive simulation, on October 27, 2026, releasing it as a research preview to trusted testers.

The announcement, authored by Raia Hadsell, Diego Rivas and Hang Qi, describes Genie 4 as natively multimodal, with music, ambient noises and interactive dialogue generated simultaneously alongside the simulated world. DeepMind says the model “unlocks unprecedented realism and improves real-time interactive simulation,” and delivers enhanced photorealism, fluid movement and native audio, with the accompanying sound in its demonstrations produced directly by the model.

The model incorporates and builds on capabilities from the Gemini family. Gemini Omni anchors the simulations in real-world appearance and behavior, while other Gemini models provide the complex reasoning needed to help maintain a simulation’s coherence and continuity. With that foundation, a user can set an event in motion and Genie 4 simulates what happens next: how the world responds, how it sounds, and how the people in it move, react and speak.

DeepMind describes understanding and simulating the world as a crucial aspect of building AGI. The company says Genie 4 advances its ability to create safe, infinitely variable simulations that can support training environments for robots and agents as well as the creation of new interactive experiences.

Four Capability Areas

DeepMind’s demonstrations are recordings of real-time interactions in which users navigate worlds using standard keyboard controls and a camera, steering the scene with text prompts as it evolves. The first documented capability is real-time control: a user can modify a running simulation with natural language, changing the weather, moving the camera, or introducing a new person or object, and Gemini contextually adapts the prompt to match the surrounding scene. Between inputs, Gemini also monitors and steers the scene to keep the world progressing cohesively.

On physics and human action, DeepMind says Genie 4 introduces a new standard of realism to real-time interactive simulation, modeling real-world physics scenarios such as aerodynamics and collisions. The company describes the largest advance as its simulation of people: characters that move naturally, respond organically to their environments, and can be interactively steered in real time.

DeepMind describes Genie 4 as its first real-time world model to generate audio natively. Each world carries the ambient sound of its own setting, from wind in the trees to waves crashing at the beach. The model also introduces interactive dialogue that adapts on the fly, with characters conversing and reacting to user inputs or events in the world, all in real time.

For complex, multi-step events, Gemini 3.8 Flash serves as a guiding agent in the loop. As Genie 4 generates a world in real time, Gemini continuously observes the scene, breaks goals into a sequence of steps, and reviews incoming frames to verify that each step succeeded, adapting its plan as the world shifts or the user intervenes. DeepMind says this unlocks worlds with reliable causality, procedural behaviors and multi-step reasoning.

The company has also been experimenting with Gemini 4 Argon and Genie 4 together with Gemini Robotics. In one documented example, Argon breaks a user-instructed task into steps that are interleaved as text captions while the simulation renders, and the two systems simulate Apptronik’s Apollo humanoid playing a game of checkers, with the simulation adapting as the task progresses toward completion.

Trusted Testers and Future Work

DeepMind characterizes Genie 4 as a major research milestone that remains an early, experimental model, and it has been working with trusted-tester cohorts across creative industries, journalism, education, gaming and engineering. According to the company, filmmakers are testing Genie 4 as a rapid “draft mode” for pre-visualization and ideation, stepping into a live simulation of a scene and running multiple takes in real time while adjusting camera angles, lighting, dialogue and action. Photojournalists are exploring how a single 2D archival photograph can become a grounded simulation of that environment in seconds, and teams are testing how populated, dynamic worlds can help audiences experience historical spaces or visualize how short-term environmental events such as flash floods or wildfires unfold.

Game developers and creative technologists are testing whether they can go from an initial idea to an explorable prototype in seconds, checking visual styles, material aesthetics and interactive mechanics through natural language prompts before writing engine code. Enterprise partners are evaluating Genie 4 as a tool for rapid zero-to-one scenario generation and identifying gaps DeepMind could address to better serve simulation needs.

Looking ahead, DeepMind identifies hybrid world models as a promising path for future progress: systems that combine the logical, structured reasoning of code and language models with the rich, sensory representation of pixel-level simulation. In this approach, Gemini maps structure and plans the narrative timeline behind the scenes while Genie generates the immediate physical reality, rendering how the world looks, sounds and responds pixel by pixel.

Safety, Limitations and Availability

DeepMind says Genie 4 includes strict safety guardrails consistent with its safety policies across other models, and that it evaluated the model through internal safety evaluations before the preview to prevent the generation of harmful or malicious content. The company says it intentionally deployed Genie 4 to trusted tester partners first to gather early feedback. All video and audio generated by Genie 4 will include SynthID, DeepMind’s imperceptible digital watermarking technology for identifying AI-generated content across visual and audio modalities.

On limitations, DeepMind states that while Genie 4 can maintain coherent environments over short sessions, visual details and object permanence can still shift or drift over time, especially as a simulation runs longer, after multiple complex prompt edits, or when navigating throughout a scene. Complex physical interactions, real-time audio accuracy, legible text rendering, spatial accuracy and long-term consistency remain active areas of ongoing research.

Genie 4 succeeds Genie 3, which DeepMind announced on August 5, 2025 and described as its first world model to allow interaction in real time. Genie 3 generated dynamic worlds that users could navigate at 24 frames per second, retaining consistency for a few minutes at a resolution of 720p, and it was released as a limited research preview to a small cohort of academics and creators.

DeepMind says Genie 4 is coming soon to Google AI Ultra subscribers in Project Genie and to early access partners in the Gemini API and Google AI Studio, and the waitlist for Genie 4 in Project Genie is open.