The OpenFold Consortium, a prominent nonprofit organization focused on artificial intelligence research, has announced the preview release of OpenFold3. This open-source deep learning model is designed to predict the three-dimensional structures of complex proteins and their associated molecules with exceptional accuracy. The announcement was made on October 28, 2025, from Davis, California.
OpenFold3 was trained on a dataset comprising over 300,000 publicly accessible, experimentally determined structures, supplemented by a curated synthetic database containing more than 13 million structures. This extensive training enables OpenFold3 to significantly enhance the efficiency of in silico screening for biomolecules.
The model serves as a foundational tool for the development of next-generation protein AI applications, with potential applications spanning drug discovery, enzyme design, biosensor creation, and biomaterials innovation. By providing a robust framework for protein structure prediction, OpenFold3 is poised to accelerate research and development in various scientific fields.
This initiative reflects the growing importance of open-source solutions in scientific research, particularly in the realm of biochemistry and molecular biology, where accurate predictions of biomolecular structures are critical for advancing therapeutic and technological innovations.
