World Labs debuts Atlas AI model for spatial intelligence

World Labs, the artificial intelligence startup co-founded by computer vision pioneer Fei-Fei Li, has introduced Atlas, a multimodal world model designed to advance spatial intelligence.

Atlas is described as an “omni model” capable of generating expansive, highly detailed 3D environments from limited inputs, such as a single image or a few smartphone photos. The model can produce up to one minute of 1440p video with what the company calls “pixel-perfect” camera control, allowing users to simulate space and time by viewing scenes from any angle.

The technology aims to bridge the gap between simulated environments and physical reasoning. By enabling AI to understand 3D objects, people, and their spatial interactions, World Labs seeks to develop physics-aware models that can reason within the physical world. This approach is central to Li’s theory that true artificial general intelligence requires native spatial intelligence.