PDF OCR
Page updated
Price: $4.00 per 1,000 OCR pages, flat on every Apify plan ($0.80 per 1,000 text-layer pages; $8.00 per 1,000 large 10–20 MB images), no start fee. Free-plan cap: up to 25 pages per run on the Apify Free plan. Prices checked on 2026-10-12.
PDF OCR reads a public PDF URL and returns the text, using OCR when the pages are scans rather than a text layer. The record includes the page count and a plain-text body.
Treat the text as a working copy. Handwriting and damaged scans are outside what the tool can promise, and nothing here is a certified transcript.
Who it is for
Analysts who have public PDF links, including scans, and need the text in a file.
- Read a scanned public report that is not a text PDF.
- Pull text from a set of public filings stored as scans.
- Archive the words from a public slide or brochure PDF.
- Check whether a PDF already contains a text layer before relying on OCR.
Inputs
| Field | Type | Required | Description |
|---|---|---|---|
| urls | array of URLs | No | Public links to PDFs or images (PNG, JPEG, TIFF, GIF, WebP, BMP). |
| mode | string | No | auto (text layer first, OCR only where needed), text (never OCR) or ocr (OCR every page). Defaults to auto. |
| extractTables | boolean | No | Add clean tables from text-layer pages as CSV strings. Defaults to false. |
| keyValueStore | string | No | Optional. One of your Apify key-value stores with uploaded files to process. |
| dataset | string | No | Optional. One of your Apify datasets; the url field of each item is processed. |
Example input
{
"urls": [
"https://example.com/scans/annual-report.pdf"
],
"mode": "auto"
}Example output
[
{
"url": "https://example.com/scans/annual-report.pdf",
"pageCount": 3,
"ocr": true,
"text": "Sample text recovered from a scanned page. Not a real document."
}
]Price
$4.00 per 1,000 OCR pages ($0.80 per 1,000 pages with a text layer; $8 per 1,000 large 10-20 MB images)
Free-plan cap: Up to 25 pages per run on the Apify Free plan
Billing happens on Apify, not on this website.
Limits
- Printed text is the realistic case. Handwriting, heavy stamps, and very small type are unreliable.
- A PDF that is not public, or that requires a login, is not fetched.
- The output is plain text. It is not a certified transcription and it does not rebuild tables perfectly.
What it does not do
- It does not promise a word-perfect legal copy.
- It does not describe images, charts, or photos beyond text that is actually printed.
- It does not accept a private upload on this website. Runs happen on Apify.
FAQ
What if the PDF already has real text?
The tool still returns the text. OCR matters when the pages are scans. A text PDF does not need to be guessed at.
Which languages work?
OCR is English only. OCR text is a searchable draft, so check numbers from receipts or invoices against the original.
Can I upload a file on lepidata.com?
No. This site does not take files. The tool runs on Apify and reads public PDF URLs.
Why is the price missing?
It has not been added to the site config.
Are phone photos of paper supported?
Only if they are inside a PDF at a public URL. Crooked photos and handwriting should be treated as low confidence.