Google introduces Genie 4 with phased access

Google announced Genie 4 on October 27, 2026, as a “world model,” a model that generates interactive simulations and lets users modify them as they unfold. The company presents this research as a way to expand creative tools in fields such as film and journalism. Google announcement.

Directing scenes as they unfold

With Genie 4, users can start an event and direct the scene in real time using natural language. Google’s examples include moving the camera, changing the weather, and introducing characters while the simulation is still running. Interaction described by Google.

Genie 4 also generates audio on the fly: ambient sounds, music, and interactive dialogue accompany the events in each simulation. Audio described in the announcement.

How Gemini models contribute

According to Google, Genie 4 combines capabilities from the Gemini family. Gemini Omni provides appearance and behavior grounded in the real world, while other Gemini models contribute reasoning to maintain the simulation’s coherence and continuity. Capabilities attributed to Gemini models.

Phased access

The Genie 4 rollout began with a limited group of trusted testers. Google announced that it will soon reach Google AI Ultra subscribers through Project Genie. The other planned access routes are for early-access partners through Gemini API and AI Studio. Availability announced by Google.