Skip to main content
POST
(ESM3, ESMFold2) Folds proteins. Defaults to esmfold2-fast-2026-05 if no model is given
On this operation the ESMFold2 models accept inputs up to 768 tokens. For asynchronous ESMFold2 predictions, and for inputs up to 4096 tokens, use Batch folding with endpoint: "fold".

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Headers

accept
string | null

Body

application/json
model
enum<string> | null

You may have access to additional, private models as well. These will be visible on the Rate Limits page under the Developer Console

Available options:
esm3-open-2024-03,
esmfold2-fast-2026-05,
esmfold2-2026-05,
esmfold2-2026-05-cutoff-2025,
null
potential_sequence_of_concern
boolean
default:false

Disclose potential sequences of concern. For approved users, such sequences will not go through additional safety filtering. Reach out if you are interested in using this.

sequence
string | null

Input sequence to be folded.

msa
object | null

(ESMFold2) MSA data for the query sequence.

num_loops
integer
default:20

(ESMFold2) Number of trunk loops for iterative refinement.

Required range: 0 <= x <= 20
num_sampling_steps
integer
default:100

(ESMFold2) Diffusion ODE solver steps. Lower for speed, higher for quality.

Required range: 1 <= x <= 100
lm_dropout
number
default:0.3

(ESMFold2) Dropout probability on LM pair embeddings. When > 0, dropout is applied.

Required range: 0 <= x <= 1
lm_mask_pct
number
default:0

(ESMFold2) Fraction of sequence residues randomly masked before the PLM backbone. If not provided, defaults to 0.1 for ESMFOLD2_FAST and 0.0 for ESMFOLD2

Required range: 0 <= x <= 1
msa_max_depth
integer | null
default:1024

(ESMFold2) Number of MSA rows randomly subsampled each loop. Set to null to disable (sets msa_subsample_at_inference to False).

Required range: 1 <= x <= 16384
msa_column_mask_rate
number
default:0.1

(ESMFold2) Fraction of MSA columns randomly masked in non-query rows for inference-time diversity.

Required range: 0 <= x <= 1
include_distogram
boolean
default:false

(ESMFold2) Whether to include distogram predictions in the response.

include_pae
boolean
default:false

(ESMFold2) Whether to include Predicted Aligned Error (PAE) matrix in the response.

include_pair_chains_iptm
boolean
default:false

(ESMFold2) Whether to include pair-chain IPTM predictions in the response.

include_embeddings
boolean
default:false

(ESMFold2) Whether to include sequence and pair embeddings in the response.

Response

Successful Response

model
enum<string>
required

You may have access to additional, private models as well. These will be visible on the Rate Limits page under the Developer Console

Available options:
esm3-open-2024-03,
esmfold2-fast-2026-05,
esmfold2-2026-05,
esmfold2-2026-05-cutoff-2025
created
string
required

ISO formatted date time

potential_sequence_of_concern
boolean
default:false

Indicate that the returned sequence may be of potential concerns.

warning_messages
string[] | null

Warning messages returned by FastAPI.

coordinates
(((number | null)[])[])[] | null

Coordinates of the N, CA, C atom sequence as a Nx37x3 array.

complex
object | null

Molecular complex with support for proteins, nucleic acids, and ligands.

plddt
number[] | null

Per position plddt predictions for a protein or molecular complex. Usually produced by structure decoder.

ptm
number | null

Predicted TM score of a protein or molecular complex. Usually produced by structure decoder.

interface_ptm
number | null

Interface TM score of a protein or molecular complex.

pae
number[][] | null

Predicted Aligned Error matrix for the protein or molecular complex. Shape is L x L. Values are rounded to 2 decimals (well within 0.5 A bin width)

pair_chains_iptm
number[][] | null

Pair-chain IPTM predictions for a protein. Shape is C x C, where C is the number of chains in the protein.

output_embedding_sequence
number[][] | null

Output embedding sequence from the model. Only present if requested.

output_embedding_pair_pooled
number[][] | null

Output embedding pair mean-pooled from the model. Only present if requested.

residue_index
integer[] | null

Residue indexes from the model. Only present if embeddings are requested.

entity_id
integer[] | null

Entity ids (e.g. chain ids) from the model. Only present if embeddings are requested.