Google has launched Genie 4, the latest iteration of its browser-based world-generation project. While the model was initially unveiled in October, access was previously restricted to a select group of testers. Now, subscribers to the Ultra tier can utilize the tool to create interactive environments in real time.
Unlike traditional game engines, Genie generates scenes dynamically as users explore them. The fourth model allows for mid-session adjustments, enabling users to modify weather conditions, add characters, or alter events via text prompts without restarting. Additionally, Genie 4 generates reactive audio, including sound effects and dialogue.
Google continues to offer the previous version, Genie 3, which now supports three-minute sessions, 360-degree panoramas, and the linking of multiple environments. Users can choose between Genie 4 for enhanced visual and audio quality or Genie 3 for longer exploration capabilities.
The company notes that the technology uses predictive modeling rather than static rule-based engines. Google acknowledges potential limitations, such as object drift and complex physics challenges. All generated content includes SynthID watermarks and undergoes automated safety screening.




