Protenix is ByteDance Research's structure prediction model for proteins, small molecule ligands, and multi-component complexes. It supports modeling of all components as a unified system, with full transparency into training data, model scale, and inference approach —  for teams who need open access to next-generation prediction capabilities.

Key capabilities:

• Protein monomer, multimer, and protein-ligand complex prediction

• Models proteins and small molecule ligands as a single complex

• Improved quality with fast and private MSA support (GPU-accelerated)

• License: Apache 2.0

Learn more about Protenix.