OCR: Document Text Recognition with 98%+ Accuracy
A free month of testing on your own documents.
What Is OCR and How Does Text Recognition Work?
OCR turns images into text. Our API uses AI models on GPUs: it reads more accurately than traditional tools and does it hundreds of times faster.
- Printed and handwritten text
- JPG, PNG, PDF, TIFF, BMP formats
- Automatic page layout detection
- Table and form recognition
- JSON output
Document Digitization: What Problems Does OCR Solve?
Retyping one invoice takes a few minutes. At a hundred documents a day, that's a full-time job. OCR does the same in seconds.
Piles of Documents
Archives, invoices, contracts: turn paper into searchable text in minutes.
No More Retyping
Document data lands in your system without anyone typing it in.
Data Chaos
Extract data from tables and forms straight into JSON, ready for import.
Multilingual Documents
Text recognition in multiple languages including Polish, English and German.
How Does the OCR Text Recognition API Work?
- 1
Send a Document
Upload via API: JPG, PNG, PDF (multi-page), TIFF or BMP.
- 2
AI Analyzes
The model recognizes layout, tables, text and document structure on GPU.
- 3
Get Your Data
Text or JSON, tables, form fields, paragraphs. Ready to import.
OCR API on GPU: 98% Accuracy and Hundreds of Pages per Minute
98%+ Accuracy
AI models trained on millions of documents. On printed text, nearly flawless.
Hundreds of Pages per Minute
Processing on NVIDIA GPUs. Stack of invoices in seconds, not hours.
Tables? No Problem
Automatic tabular data extraction with structure preserved.
Batch Processing
Send hundreds of documents, the API handles them in parallel.
GDPR by Default
Documents processed on servers in Poland. Full regulatory compliance.
Straightforward integration
REST API with OpenAPI docs. Ready-made code examples come with your access.
OCR Use Cases: Invoices, Contracts, Archives
Looking for invoice OCR? It has its own page
Automatic invoice reading: tax IDs, amounts, dates, bank account and line items straight to JSON. Sample output, pricing and FAQ.
Document OCR: Frequently Asked Questions
What is OCR and what is it for?
OCR (Optical Character Recognition) extracts text from images: scans, photos and PDF files. It turns a document into text you can search, copy and process automatically. Our OCR API also reads tables, form fields and handwriting.
How accurate is document recognition?
On printed text we reach over 98% character-level accuracy. Handwriting and low-quality scans score lower, which is why we offer a free month of testing on your own documents so you can measure the result before deciding.
Which file formats does the OCR API support?
JPG, PNG, TIFF, BMP and multi-page PDF files. You send the document through a REST API and receive text or JSON with the structure preserved: paragraphs, tables and fields.
How much does document OCR cost?
You pay per processed page: from PLN 0.06 per page in the base tier, less at higher volumes. New clients test for free during the first month and then pay half the rate. The calculator on this page gives you the exact price.
Is the OCR GDPR compliant? Where are documents processed?
Documents are processed only on our own servers in Poland. We never send them to third-party providers or outside the EU, and we do not keep files longer than it takes to return the result. We sign a data processing agreement.
Can the OCR read invoices and extract the data?
Yes. From invoices we extract tax IDs, document numbers, dates, net, VAT and gross amounts, bank account numbers and line items. Details, a sample JSON output and pricing are on the invoice OCR page.
How fast is OCR on GPU?
A single page takes a fraction of a second, and hundreds of documents sent at once are processed in parallel. A stack of a thousand invoices takes minutes, not hours.
See What OCR Costs
Set how many document pages you process per month. You pay only for processed pages.
Your starting price
PLN 300/ mo.PLN 600
At this volume you pay PLN 0.03 per page, with tables, forms and handwriting included.
On AWS Textract (forms and tables) the same volume costs about PLN 2,400 per month.
Volume tiers
- first 10k pagesPLN 0.03 / page
- 10k to 50k pagesPLN 0.025 / page
- 50k to 200k pagesPLN 0.021 / page
- over 200k pagesPLN 0.0175 / page
Tiers are billed separately, like tax brackets: volume that falls into a lower tier keeps that tier's rate. Rates in the table already include the -50% intro discount.
Your first month of testing is free. The -50% intro rate for new clients kicks in after it.
Order this package
Leave your contact details. The calculated volume reaches us with your inquiry, and we get back within 24h with access and details.
Test OCR for a Month, Free
Send your own invoices, contracts or scans and see how much data the API extracts without manual corrections.
We respond within 24h. We help with the rollout.