Genie 3, a foundation world model by Google DeepMind, signifies progress toward artificial general intelligence. It generates various 3D environments, capable of sharing interactions in real-time and maintaining physical consistency over time. The model supports extensive duration generation and promptable world events. Genie 3 enhances training for agents in general tasks. It builds on previous models like Genie 2 and Veo 3, showcasing improved capabilities, important for applications in education, gaming, and prototyping.
Genie 3 is the first real-time interactive general purpose world model that goes beyond narrow world models. It can generate both photo-realistic and imaginary worlds.
With a simple text prompt, Genie 3 can generate multiple minutes of diverse, interactive, 3D environments at 24 frames per second with a resolution of 720p.
The model's simulations stay physically consistent over time because it can remember what it previously generated, an emergent capability not explicitly programmed.
World models are key on the path to artificial general intelligence, especially for embodied agents, where simulating real world scenarios is particularly challenging.
Collection
[
|
...
]