Objaverse is a large-scale dataset containing 800K+ labeled 3D objects, each with a name, description, tags, and other metadata. It encompasses various object types, including static objects, animated objects, characters with articulated annotations, decomposable models, indoor and outdoor environments, and exhibits diverse visual styles. Objaverse can be utilized for generating 3D models, enhancing 2D instance segmentation, demonstrating open-vocabulary AI, and researching the robustness of CLIP.