Skip to content

Quickstart — your first extraction

Updated

1. Get an API key#

Sign in to the client portal at https://app.docxtract.io/ and generate a key from Settings. Keys look like sk_ followed by 32 hex characters.

Note
The separator matters. sk_ with an underscore is a DocXtract key. sk- with a hyphen belongs to a different API provider and will not authenticate here.

2. Find out which document types your key can use#

curl https://api.docxtract.io/v3.1/models \
  -H "Authorization: Bearer YOUR_API_KEY"

This call costs no credits, so use it freely — including to populate dropdowns in your own UI. It returns only the document types your key is permitted to use.

3. Extract a document#

curl -X POST https://api.docxtract.io/v3.1/documents \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "file=@invoice.pdf" \
  -F 'options={"model":"invoice"}'

Accepted files: PDF, JPG, PNG, up to 10 MB and 150 pages.

Warning
options is a JSON string in a form field, not nested form data. And if you omit document_type/model, it defaults to invoice — a passport sent without it is extracted as an invoice. Always set it explicitly.
Anatomy of the request and the response
Anatomy of the request and the response

4. Read the response#

{
  "success": true,
  "data": { "vendor": "ABC Suppliers Pvt Ltd", "total": 5900.00 },
  "processing_time_ms": 5577,
  "model_used": "invoice",
  "pages": 1,
  "extraction_id": "abc123"
}

Metadata sits at the root level beside data, not nested under meta. The field is extraction_id, not id, and it appears only when store_db is true.

data varies by document type — an invoice returns vendor/totals/line_items, a passport returns name/number/dob. Only the envelope is stable, so treat data as an open map.

5. Handle the multi-page case#

A PDF of more than 3 pages does not return 200. It returns 202 with a chunk manifest and requires two more calls. This is not optional and cannot be turned off, so any integration that accepts arbitrary PDFs must handle it — see Multi-page documents.

Base URL#

https://api.docxtract.io/ — with no /api prefix. The endpoint is /v3.1/documents.

docxtract.rpatech.ai is a legacy alias that redirects here; do not use it in new code.

Next steps#

  • API versions — build against v3.1
  • Error codes — branch on error.code, never on the message
  • Interactive reference and Postman collection: https://app.docxtract.io/api-reference.php

Related