DeepMind Unveils Genie 3: Advancing Real-Time Interactive 3D World Generation

DeepMind has announced Genie 3, its latest AI model designed to generate interactive 3D environments in real-time. Building upon its predecessor, Genie 2, this new model can create dynamic worlds from simple text prompts or images, allowing users to navigate these environments at 720p resolution and 24 frames per second.

A notable feature of Genie 3 is its ability to maintain consistency within the generated worlds over extended periods. Unlike earlier versions, it retains objects in place even when users move the camera away and return later. This capability is achieved through the model's memory of previously generated content, enabling it to simulate real-world physics and interactions more accurately.

DeepMind envisions Genie 3 as a tool for training AI agents in diverse scenarios, contributing to the development of artificial general intelligence (AGI). By providing a platform for AI systems to learn from interactive experiences, Genie 3 aims to bridge the gap between simulated environments and real-world applications.

Currently, access to Genie 3 is limited to a select group of academic researchers and creative professionals. DeepMind has not yet announced plans for broader public availability.