INFERENCE INFRASTRUCTURE FOR SCIENTIFIC AI
Run biology, chemistry, and physics AI models through one unified API. No GPU management. No infrastructure. Pay per call.
import curieai
client = curieai.Client(api_key="sk-...")
# Fold a protein sequence
result = client.fold("MKTIIALSYIFCLVFADYKDDDDK")
print(f"Confidence: {result.confidence:.1f}%")
result.save_pdb("my_protein.pdb")9 models liveBiology · Chemistry · PhysicsPay per call500 free calls
SCIENTIFIC AI MODEL CATALOG
9 models live. Biology, chemistry, and physics. Pay per call.
BIOLOGY● LIVE
ESMFold v1
Fast single-sequence protein structure prediction from amino acid sequence.
INPUT
Sequence / FASTA
OUTPUT
PDB + pLDDT score
Meta AI~200ms$0.005/call
CLICK TO EXPLORE →BIOLOGY● LIVE
ESM-2 650M
Protein language model embeddings (1280-dim) for downstream ML tasks.
INPUT
Amino acid sequence
OUTPUT
1280-dim embedding
Meta AI~90ms$0.002/call
CLICK TO EXPLORE →BIOLOGY● LIVE
ProteinMPNN
De novo protein sequence design from backbone structure.
INPUT
PDB structure
OUTPUT
Designed sequences
Baker Lab~300ms$0.005/call
CLICK TO EXPLORE →CHEMISTRY● LIVE
ChemBERTa-2
Molecular embeddings and property prediction from SMILES strings.
INPUT
SMILES string
OUTPUT
Embedding + properties
HuggingFace~80ms$0.002/call
CLICK TO EXPLORE →CHEMISTRY● LIVE
RDKit
Molecular properties, fingerprints, and Tanimoto similarity from SMILES.
INPUT
SMILES string
OUTPUT
Properties + fingerprint
RDKit~20ms$0.001/call
CLICK TO EXPLORE →CHEMISTRY● LIVE
MolT5
Natural language descriptions of molecules from SMILES strings.
INPUT
SMILES string
OUTPUT
Text description
Google Research~400ms$0.003/call
CLICK TO EXPLORE →PHYSICS● LIVE
MACE-MP-0
Universal ML force field — energy and forces for any material system.
INPUT
Atomic coordinates
OUTPUT
Energy + forces
Cambridge / DeepMind~150ms$0.005/call
CLICK TO EXPLORE →PHYSICS● LIVE
NequIP
Equivariant neural network interatomic potential for molecular dynamics.
INPUT
Atomic coordinates
OUTPUT
Energy + forces
MIT / Harvard~200ms$0.005/call
CLICK TO EXPLORE →PHYSICS● LIVE
DeePMD
Deep potential molecular dynamics force field for atomistic simulations.
INPUT
Atomic coordinates
OUTPUT
Energy + forces + virial
DeepModeling~200ms$0.005/call
CLICK TO EXPLORE →COMING SOON · Boltz-2 · DiffDock · RFdiffusion · AlphaFold 3 · Nucleotide Transformer · scGPT · ChemFormer
BUILD WITH THE CURIE API
Install the SDK and start running scientific AI models in minutes.
import curieai
client = curieai.Client(api_key="sk-...")
# Fold a protein sequence
result = client.fold("MKTIIALSYIFCLVFADYKDDDDK")
print(f"Confidence: {result.confidence:.1f}%")
result.save_pdb("my_protein.pdb")Ready to run your first model?
500 free API calls per month. No credit card required.
CREATE FREE ACCOUNT →