Models · 2026-10-27

Google unveils Genie 4: AI builds interactive worlds with more realism

Google has announced Genie 4, a world model that generates interactive virtual worlds with more realistic visuals, audio, and changes made through text commands.

✨ What's new in Genie 4

Google has announced Genie 4, a so-called world model that creates a virtual world where you can move around and interact with objects. The company names four main changes. The worlds are more realistic. Users can update an experience in real time by typing a command in natural language. The model can generate audio for the world. Under the hood, it uses Gemini 3.8 Flash to help with simulations.

The length of generated experiences has grown from one minute, as with the previous Genie 3 model, to two minutes.

📌 What it could be used for

Google suggests Genie 4 could help with pre-visualization of scenes for filmmakers, rapid prototyping of game ideas, and creating training scenarios for robots.

At the same time, the company warns that as simulations run longer, details can shift over time and objects may not behave as expected. This is worth keeping in mind if you expect fully stable results.

🔹 How to try it

For now, Genie 4 is launching as a "research preview" for "trusted testers." Google says it will come soon to Google AI Ultra subscribers. You can already join a waitlist to apply for access to Project Genie. The first source also links to the model's page on the DeepMind blog.

Sources: 🔗 The Verge — AI