API Reference
Libraries
API Overview
Predictions
Run prediction models on molecular inputs. Each application is available as its own endpoint with application-specific inputs and outputs.
PredictionsStructure And Binding
Predict 3D structure coordinates, per-residue confidence scores, and binding metrics for a molecular complex. Supports explicit glycan graphs and covalent attachments, optional template-guided folding, and per-protein MSA control.
Start a structure and binding prediction
List structure and binding predictions
Retrieve a structure and binding prediction
Delete prediction data
Estimate cost for a structure and binding prediction
PredictionsAdme
Predict Tier 1 ADME summary values for a batch of small molecules specified by SMILES.
Start an ADME prediction
List ADME predictions
Retrieve an ADME prediction
Delete ADME prediction data
Estimate cost for an ADME prediction
Small Molecule
Small Molecule Pipeline: design novel small molecules and screen compound libraries against protein targets. Includes de novo generation and virtual screening.
Small MoleculeDesign
Generate novel small molecules optimized for binding to a protein target. Results are scored by binding confidence (likelihood of binding, for hit discovery), optimization score (binding strength ranking, for lead optimization), and structure confidence.
Start a small molecule de novo design run
List small molecule design runs
Get small molecule design run status
Get generated small molecule designs
Stop a running small molecule design run
Resume a stopped small molecule design run
Delete small molecule design run data
Estimate cost for a small molecule design run
Small MoleculeLibrary Screen
Screen an existing library of small molecules against a protein target. Results are scored by binding confidence (likelihood of binding, for hit discovery), optimization score (binding strength ranking, for lead optimization), and structure confidence.
Start a small molecule library screen
List small molecule library screens
Get small molecule library screen status
Get screened small molecule candidates
Stop a running small molecule library screen
Resume a stopped small molecule library screen
Delete small molecule library screen data
Estimate cost for a small molecule library screen
Small MoleculeExplore
Explore a large library of small molecules against a protein target without screening all of it. Submit the whole library and a budget; molecules are chosen to score as results arrive, so each choice is informed by everything scored so far. Results use the same scores as a library screen, and progress reports the library size alongside the budget.
Start a small molecule library exploration
Get small molecule exploration status
Get scored molecules from an exploration
Stop a running small molecule exploration
Resume a stopped small molecule exploration
Protein
Design novel protein binders, redesign selected residues in fixed structures, and screen protein libraries against targets.
ProteinDesign
Generate binder or generic protein designs. New requests use the top-level type discriminator (binder or generic), while the legacy target plus binder_specification body remains accepted for migration. Binder requests can share one CIF across target and binder, sample uniformly across multiple specifications, or use Boltz-managed curated antibody and nanobody defaults. Results are discriminated by type: binder runs include binding metrics, while generic runs return structure and secondary-structure metrics only. A generic request can use a fusion_protein entity to concatenate two or more ordered fixed, designed, or template-backed protein segments into one output chain.
Start a protein design run
List protein design runs
List curated protein design specifications
Get protein design run status
Get generated protein designs
Stop a running protein design run
Resume a stopped protein design run
Delete protein design run data
Estimate cost for a protein design run
ProteinSequence Redesign
Redesign selected protein residues in one fixed CIF structure. Use the top-level type discriminator to choose binder redesign, with target and binder chain roles, or generic redesign. Every chain in the input structure must be assigned exactly once. Binder results include binding and structure metrics; generic results include structure and secondary-structure metrics.
Start a protein sequence redesign run
List protein sequence redesign runs
Get protein sequence redesign run status
Get redesigned protein sequences
Stop a running protein sequence redesign run
Resume a stopped protein sequence redesign run
Delete protein sequence redesign run data
Estimate cost for a protein sequence redesign run
ProteinLibrary Screen
Screen an existing library of proteins against a target structure. Results are scored by binding confidence (likelihood of protein-protein interaction) and structure confidence.
Start a protein library screen
List protein library screens
Get protein library screen status
Get screened protein candidates
Stop a running protein library screen
Resume a stopped protein library screen
Delete protein library screen data
Estimate cost for a protein library screen
Share Links
Share read-only access to predictions and pipeline runs by issuing time-limited links that visitors can open without an API key or, for email-restricted links, after signing in with an allowed email. A share link is scoped to a single workspace and bundles one or more predictions and pipeline runs. The link ID is itself the bearer credential; treat it as a secret. Create, retrieve, and archive require an API key or supported OAuth bearer token with read permission on every referenced resource. Retrieving metadata remains available after expiry or archive. Viewing content and listing shared pipeline results are gated by the link ID and the link’s access mode. Archiving a link revokes public access immediately; subsequent content reads return 404. The underlying predictions and pipelines are unaffected and remain accessible through their own authenticated endpoints.
Create a share link
Retrieve a share link
Archive a share link
Read a share link
List pipeline results from a share link
Admin
Manage workspaces and API keys. Requires an admin API key. Admin keys have full access to all management and compute operations across all workspaces in the organization.
AdminWorkspaces
Workspaces provide isolated environments for organizing predictions and pipeline runs across teams, projects, or customers. Each workspace has independent data retention settings, can be associated with workspace API keys, and can have a lifetime spending limit for tenant-level budget enforcement. Spending limits use milli-USD and begin tracking usage when first configured. Admin keys can create or change a limit; a workspace key can read the limit for its own workspace.
Create a workspace
List workspaces
Get a workspace
Update a workspace
Archive a workspace
Get a workspace spending limit
Set a workspace spending limit
AdminAPI Keys
API keys authenticate requests to the Boltz API. There are two key types: admin keys have full access to all management and compute operations across the organization, while workspace keys are scoped to a single workspace and can only perform compute operations (predictions, protein design, small molecule design) within that workspace. Keys can be created in live or test mode. Test keys (prefixed sk_bc_*_test_) create test-mode resources with synthetic data and no GPU cost. Every resource includes a livemode field indicating its mode.
Create a workspace API key
List API keys
Revoke an API key
AdminUsage
Retrieve aggregated usage data for the organization. Usage can be grouped by workspace and/or application, and filtered by time range, workspace, and application.
Get usage data
Cli
Check the installed boltz-api CLI version against the currently published CLI release and the minimum version supported by the Boltz API.
CLI version metadata
Auth
Inspect the authentication context for the current credential, including the organization or workspace scope for API keys and the available organization memberships for OAuth bearer tokens. OAuth callers can use this information to choose which organization to send with future requests.