The Beijing Academy of Artificial Intelligence (BAAI) has recently open-sourced Uni3D, a 3D vision general model with 1 billion parameters. This model is capable of processing point cloud data, achieving comprehensive technical breakthroughs in mainstream 3D vision tasks, and demonstrating exceptional general vision capabilities. The development team of Uni3D drew on design experiences from 2D vision models and introduced a multi-modal alignment training method, allowing it to directly inherit knowledge learned from 2D vision models, thereby acquiring powerful 3D vision abilities. The model has achieved state-of-the-art results in various 3D vision tasks, showcasing its strong versatility and transfer learning capabilities. The BAAI states that the open-source release of Uni3D lays a solid foundation for future research and applications in 3D computer vision.