Extract a document (schemaless)

POST

/api/v1/extract

curl --request POST \
  --url https://api.langparse.dev/api/v1/extract \
  --header 'Content-Type: application/json' \
  --header 'X-Api-Key: <X-Api-Key>' \
  --data '{ "fileBase64": "example", "fileName": "invoice.pdf", "hint": "Pull line items with description, qty, unit price.", "model": "example", "pages": [ 3, 5 ], "workflowId": "example" }'

Parse WITHOUT a model — the parser infers the best structure and returns it as JSON. Returns a job id; poll GET /api/v1/documents/{id} for the result.

Authorizations

• apiKey
• bearerAuth

Request Body^required

Provide fileBase64 (inline). Optional hint steers the shape; model overrides the extractor.

Media typeapplication/json

object

fileBase64

required

Base64-encoded file bytes (≤ ~12 MB).

string format: byte

fileName

string

Example

invoice.pdf

hint

Free-text guidance for the structure to infer.

string

Example

Pull line items with description, qty, unit price.

model

Extractor model id (else the default vision model).

string

pages

Only extract these 1-indexed pages (e.g. [3,5]). Omit for all pages.

Array<integer>

Example

[
  3,
  5
]

workflowId

Run this workflow on the file after the document is created.

string

Responses

202

Extract job accepted

Media typeapplication/json

object

id

string

status

string

Allowed values: queued

debugId

string

pollUrl

string

Example

{
  "status": "queued"
}

401

Missing or invalid API key

402

Insufficient credits