Open-source structure prediction with binding affinity scoring
Boltz-2, from Boltz PBC, is the first fully open-source structure prediction model to combine 3D complex prediction with binding affinity estimation — moving beyond static structures toward a more complete picture of molecular interactions. It supports user-defined distance constraints, multi-chain template integration, and conditioning on experimental methods, giving computational scientists more control over how predictions are generated.
Key capabilities:
Protein, ligand, and multi-component complex structure prediction
Binding affinity prediction — not just structure, but strength of interaction
User-defined distance constraints and multi-chain template support
Fully open-source and commercially available

Explore AI biotech resources
Browse practical AI use cases, prompts, and stories from scientists using Benchling AI to move faster and discover more.