Invoice OCR API: best way to get structured fields back
The API takes an invoice file and returns the fields as JSON: vendor, number, dates, line items, tax and totals. Structured e-invoices are parsed; scans are recognised.
5 documents free. PDF, photo or XML — data out as CSV, Excel or JSON.
Request and response
Send a file (PDF, image, XML, DOCX, XLSX) and get back a document object with header fields, line items and totals.
Every field carries the value; confidence flags mark values that deserve a look.
Why not a generic OCR API
Generic OCR returns text with coordinates. You then write the rules that turn text into vendor, number and totals — per supplier, forever.
Invoice-specific extraction returns the fields, not the pixels.
Structured files first
If the file is ZUGFeRD, XRechnung, Factur-X, UBL or Peppol, the API reads the XML: exact values, no recognition, no cost.
Only images run through recognition.
Limits worth knowing
Multi-invoice PDFs are split page by page and by invoice boundaries.
Rotated or dark scans are handled, but a hand-written total on a photo is where recognition still struggles. Check the total.
Frequently asked
What does an API call cost?
Structured e-invoices: nothing. Scans: €0.07 per page.
Is there a free plan for developers?
Five documents to test, then pay per page — no monthly minimum.
Which languages are supported?
Documents in Latin and Cyrillic scripts are read; the fields are language-independent.