Google's Genie 4 Signals Shift Toward Prompt-Based Game Worlds
AI-generated image · US National Wire
The new world model allows for real-time scene updates via natural language, challenging traditional development pipelines.
Google DeepMind has released Genie 4, a world model designed to generate photorealistic, interactive experiences. According to reporting from TechCrunch and The Verge, the model allows users to steer simulations in real time using natural language prompts to change camera angles, weather, or add new objects and people.
Powered by Gemini 3.8 Flash, Genie 4 doubles the simulation length of its predecessor, Genie 3, now supporting experiences up to two minutes. TechCrunch reports that the model integrates real-time audio, including ambient noise, music, and interactive dialogue. Google is positioning the tool for prototyping game ideas, robotic training scenarios, and filmmaking pre-visualization.
While Google DeepMind engineer Diego Rivas told TechCrunch that the tool is currently most valuable for rapid ideation and storyboarding, the model's capabilities suggest a move toward hybrid systems. Tejas Kulkarni, a research director at Google, described a future where Gemini writes code and text that Genie ingests to simulate reality in real time.
Google warns that the technology is still in its early stages; both The Verge and TechCrunch note that object permanence and fine details can shift as simulations run longer, and in-world text is not yet consistently legible.
Genie 4 is currently in research preview for trusted testers. Google stated it will soon be available to Google AI Ultra subscribers via Project Genie, as well as early access partners of Google AI Studio and the Gemini API.