OpenFold Consortium Releases OpenFold3 for AI Research

The OpenFold Consortium released OpenFold3, an open-source deep learning model designed to predict 3D structures of biomolecular complexes ^[1][3]. This update marks the first time the group has shared its full training stack, model weights, and datasets with the public ^[1]. These resources aim to foster independent validation and transparency within the field of AI-driven biology research ^[1][2]. By making the model weights available, researchers can now verify and build upon the protein structure predictions more effectively ^[3]. The consortium’s decision to provide full data access addresses long-standing challenges regarding scientific reproducibility in the industry ^[2]. This move positions OpenFold3 as a primary tool for scientists working on complex biomolecular interactions ^[2][3].