Exhibitor login
AI Insider 28 October 2026

Google introduces Genie 4: Innovative world model with sound

Google introduces Genie 4: Innovative world model with sound

Google DeepMind has recently announced *Genie 4*, an advanced world model that can simulate realistic interactive environments with sound in real time. Users can ignite the environment using natural language, with sound, characters, and physics immediately adapting to the given instructions. For now, the model is only accessible to a select group of testers, and details about availability in the Netherlands remain unknown.

Genie 4 is the successor to Genie 3 and introduces several key innovations. Unlike previous versions, Genie 4 enables sound integration for the first time, generating music, ambient sounds, and character dialogues simultaneously with the visuals. Additionally, the model offers more photorealism and more natural movement of people and objects, thanks to improved physics such as collisions and air resistance. All of this is made possible by the use of the Gemini Omni, which enhances visual reality, while the Gemini 3.8 Flash model acts as the central driving brain.

With Genie 4, users can easily make adjustments during simulations, such as changing weather conditions or moving cameras and characters, without having to start over. This real-time aspect is particularly valuable in creative processes, such as film production, game design, education, and robotics, where it provides a safe training environment for robots. All generated output receives a *SynthID* watermark, confirming the origin of the AI-generated material.

At this time, Genie 4 is only available in a testing phase for trusted users, with broader accessibility expected via Google AI Ultra and the Gemini API in the future. However, Google has not announced any specific pricing or release date. Availability in certain countries or languages, including Dutch, remains unclear for now. In the same sector, competitor Runway is also active with its GWM Worlds 2 model, which strengthens the competition in real-time, interactive environments.

Read the full article from AI Insider.