FullOCR
Pricing

Pay for pages, not for seats

Our own model cost is $0.082 a page. Everything below is priced from that, not from a competitor's ad. There is no cheaper tier that skips the verification, because a document you cannot trust is not worth converting.

Pricing

Pay for pages, not for seats

Our own model cost is $0.082 a page. Everything below is priced from that, not from a competitor's ad. There is no cheaper tier that skips the verification, because a document you cannot trust is not worth converting.

Every account starts with 5 free pages a month — no card, no watermark, all three formats.

Pages

Pay only for what you convert. No subscription, no seats.

$0.24 per page

$6 minimum order

  • Word, HTML and plain text
  • Two independent readings, then adjudication
  • Equations compiled and made editable
  • Figures redrawn as vector shapes
  • Reprint-test fidelity report
  • Standard queue, delivered within 12 hours

Certified

When the document has to stand up to somebody else's scrutiny.

$0.95 per page
  • Everything in Studio
  • Human adjudication of every flagged region
  • Character-to-pixel provenance for the whole document
  • Signed certificate of conformance
  • Insured accuracy guarantee
  • For courts, archives and publishers

Two more ways to buy

Rush

Interactive queue — minutes instead of hours. Added to any per-page price.

$0.16 add-on, per page

FullOCR Desktop

Runs on your own Mac or PC with no internet. Produces an editable Word draft with a fidelity score and a map of the regions worth checking, and escalates only those pages to the full cloud protocol. One payment per major version, 100 escalation pages included.

$149 one time

Air-gapped node

The complete cloud protocol installed inside your own network, for institutions whose documents may never leave the building. Annual licence per node, with onboarding and protocol tuning.

$4,000 per node, per year

Prices in US dollars. A free allowance of 5 pages a month needs no card. Volume above 5,000 pages a month is quoted individually.

Questions

Frequently asked

How is this different from ordinary OCR?
Ordinary OCR answers the question 'what does this say?'. FullOCR answers 'what does this page look like, exactly?' — and then rebuilds it as an editable Word document. The layout, the page breaks, the boxes, the underlines and the equations are all part of the answer.
Which languages can it read?
Persian, Arabic and English are fully supported today, including mixed documents with formulas. The interface is available in English, Persian, Arabic, Spanish, Hindi and Chinese.
What file types can I upload?
PDF, JPG, PNG, TIFF, WEBP and HEIC, including the multi-page outputs of desk and overhead scanners. Files up to 400 MB.
Can I send a link instead of uploading?
Yes. Paste a Google Drive, OneDrive or Dropbox share link and the file is fetched, converted and added to your history.
Are the equations really editable?
Yes. They are native Word equations, so you can click into them and change them. Every one is compiled during conversion to prove it is well formed.
What does the fidelity score mean?
We render our own Word output back into an image and compare it with your original scan. Page fidelity measures how closely the text blocks, margins and columns match; the divergence map shows where they differ. No human reference answer is involved.
Do you keep my files?
Only if you want us to. You can have files deleted immediately after conversion, and you can delete any job from your history at any time.
Can I run it on my own machine?
Yes, for organisations that need it. There is a licensed offline installation for archives, courts, publishers and libraries that cannot send documents to a cloud service.

Send us your hardest page

The one with the table that never survives, or the formula that always breaks, or the Persian text that always reverses. That is the page we built this for.

Convert a file