Google DeepMind Unveils Genie 4 World Model With Native Audio and Real-Time Steering

Google DeepMind's Genie 4 world model generates interactive worlds with native audio, free-form edits, and real-time natural-language steering.

Google DeepMind announced Genie 4 on Tuesday, its newest world model for real-time interactive simulation, bringing what the company calls unprecedented realism and native audio to AI-generated environments that users can steer on the fly with natural language.

Genie 4 generates the most photorealistic and interactive worlds DeepMind has produced to date, according to coverage from TechCrunch following a press briefing. Users can type prompts mid-simulation to change a scene — moving the camera, changing the weather, or introducing new people and objects — while the model generates synchronized audio on the fly, including ambient sound, music, and interactive dialogue.

World models — AI systems that simulate interactive environments rather than just predict text or render static images — have become one of the most contested frontiers in AI. Follow the latest AI model releases on AI Buzz Wire as the race accelerates.

Built on Gemini, Directed by Gemini

Genie 4 is deeply integrated with Google's Gemini model family. According to DeepMind's announcement, Gemini Omni anchors simulations in real-world appearance and behavior, while other Gemini models provide the complex reasoning needed to keep a simulation coherent over time.

TechCrunch described the architecture as Gemini acting as a director that plans steps and checks frames while Genie creates the visuals. In a research experiment tied to Gemini Robotics, a Gemini 4 Argon model broke tasks down into steps that Genie 4 then simulated in real time — a demonstration DeepMind positions as a possible way to rehearse robot scenarios in simulation before they happen in the physical world.

"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," Tejas Kulkarni, research director at Google, said during a briefing, according to TechCrunch.

Double the Persistence, Better Physics

The generational leap over Genie 3 — released as a research preview a little over a year ago — is substantial. Genie 4 can maintain a consistent scene for up to two minutes, double its predecessor's limit, and offers better physics and more humanlike characters, per TechCrunch. It also supports free-form edits that were not possible before, such as changing weather or camera angles in real time or adding new people or objects to a running simulation.

The model is not without gaps. DeepMind's team acknowledged during the briefing that details and objects can drift the longer a simulation runs, and that the company has not yet solved how to ensure generated audio remains accurate or that in-world text stays legible.

A Consolidating World Model Market

Genie 4 arrives as the world model industry begins to consolidate. TechCrunch notes the release comes a couple of months after Runway launched GWM Worlds 2, a playable world model with audio generation. Meanwhile, Fei-Fei Li's World Labs is being acquired by AMD, Amazon is in talks to buy Decart, and Rivian has acquired Third Dimension to bring simulation capabilities in-house for self-driving testing.

Google is clearly trying to stay ahead of that crowded field — and investors have noticed. Press reports noted Alphabet shares rose more than 2% after the Genie 4 unveiling, which landed just ahead of the company's quarterly earnings report on Wednesday. The predecessor Genie 3 rattled video game stocks when it debuted last year, as investors grappled with the implication that real-time generated worlds could eventually compete with traditional game development.

From Gaming to Film Storyboards to Robotics

DeepMind frames world simulation as a crucial step toward AGI, enabling safe training environments for robots and agents. The pitched commercial use cases span gaming, interactive learning, and manufacturing. One speculative example from the company: entering a painting of a historical scene and chatting with its subjects about the event portrayed.

But early users are pointing somewhere unexpected: film pre-production. "The real time aspect becomes most valuable in the areas where keeping a state of flow in the creativity space becomes incredibly useful," said Diego Rivas, a Google DeepMind engineer, during the press briefing, per TechCrunch. "For example, in the ideation phase, in the storyboarding phase — anytime you're kind of in need of rapid ideation."

Trusted testers have already been using Genie 4 to prototype films, and Rivas suggested that users could eventually rely on Gemini Omni to turn those prototypes into finished creative assets, though he stressed those capabilities remain a ways off.

Availability

Genie 4 is rolling out now in research preview to a limited group of trusted testers. In the coming weeks, it will reach Google AI Ultra subscribers through Project Genie — Google's tool for generating interactive worlds from text or image prompts — as well as early access partners in the Gemini API and AI Studio.

Whether world models become the next platform shift or remain an impressive research demo, Genie 4 makes one thing clear: the line between watching AI-generated video and playing inside it is disappearing.