Protein Structure · MIT × Recursion
Structure and binding affinity in one pass, approaching FEP accuracy 1000× faster.
30s–5min
Runs on Managed GPU endpoint
Runs on managed compute. Availability and startup time depend on current capacity.Run this model
About this model
Boltz-2 co-folds a protein with its ligand and predicts binding affinity in the same forward pass. It approaches free-energy-perturbation accuracy at roughly a thousandth of the cost, which makes it the default engine for hit-to-lead triage inside Corollary. MIT licensed end to end, so there is no commercial-use asterisk.
Reported performance
Standardized I/O contract
Every model in the Hub speaks the same contract, which is what lets the Router and the agent call any of them without special-casing.
Inputs
Target protein sequence
Ligand SMILES
Outputs
All-atom coordinates in PDB format, viewable and downloadable.
Binding affinity estimate for the protein–ligand pair.
pLDDT score per residue (0–100). Above 70 is generally reliable.
