World Labs Shares Atlas Multimodal World Model

Justin Johnson, co-founder of World Labs and a computer vision researcher, posted that the team is sharing Atlas. He described the new multimodal world model as special because it unites generation and reconstruction. These two tasks form the core of his visual intelligence work for more than a decade. World Labs builds large world models focused on spatial intelligence. The announcement came directly from Johnson on social media and supplies the initial details released by the company about this project.