Genie 4: A new generation of world simulation
2026-10-29 · Google DeepMind
Genie 4: A New Generation of World Simulation
On October 27, 2026, Google announced Genie 4, a new world model that unlocks unprecedented realism and improves real-time interactive simulation. This release marks a significant step forward in world simulation technology.
Core Features and Multimodal Capabilities
Genie 4 is natively multimodal, meaning it can simultaneously generate audio alongside the visual world. This includes music, ambient noises, and interactive dialogue, creating a highly immersive experience.
The model is built upon the capabilities of the Gemini family:
- Gemini Omni: Anchors the simulations in real-world appearance and behavior, ensuring visual and physical authenticity.
- Other Gemini Models: Provide the complex reasoning necessary to maintain a simulation’s coherence and continuity over time.
With this foundation, users can simply set an event in motion, and Genie 4 will simulate what happens next. This includes how the world responds, how it sounds, and how the people within it move, react, and speak.
Shaping World Simulation in Real-Time
Genie 4 allows users to modify simulations in real-time using natural language. Key interactive capabilities include:
- Dynamic Modification: Users can change the weather, move the camera, or introduce new people and objects on the fly.
- Contextual Adaptation: Gemini contextually adapts user prompts to match the surrounding scene seamlessly.
- Autonomous Monitoring: Between user inputs, Gemini monitors and steers the scene to ensure the world progresses cohesively.
Demonstrations highlight Genie 4's enhanced photorealism, fluid movement, and native audio generation. For instance, one simulation uses a reference image to create a scene featuring a floating platform, wild grasses, a twisted tree, and a glowing amber sphere, with a middle-aged man playing a brass saxophone. Genie 4 accurately renders the 16mm film aesthetic and generates realistic lake ambience, wind, and saxophone sounds. Other examples include a sweeping pink satin fabric billowing through a sun-drenched meadow and a first-person POV hot air balloon ride over Antarctica.
Pushing the Frontier in Realism
Genie 4 pushes the frontier in realism for physics and human action. The model ensures that the world's responses are physically believable and that human movements and reactions are natural. All accompanying sounds in the demonstrations are produced directly by Genie 4.
A Crucial Aspect of Building AGI
Understanding and simulating the world is a crucial aspect of building Artificial General Intelligence (AGI). Genie 4 advances the ability to create safe, infinitely variable simulations. These simulations have broad applications, including:
- Developing training environments for robots and agents.
- Unlocking the creation of new interactive experiences.
Rollout and Access
Google has started rolling Genie 4 out as a research preview for trusted testers to gather feedback. Future release plans include:
- Availability for Google AI Ultra subscribers in Project Genie.
- Early access for partners of the Gemini API and Google AI Studio.
Interested users can join the waitlist for Genie 4 in Project Genie.