PDF OCR to Markdown - Scanned PDF & Image to Text for RAG avatar

PDF OCR to Markdown - Scanned PDF & Image to Text for RAG

Pricing

from $3.97 / 1,000 ocr pages

Go to Apify Store
PDF OCR to Markdown - Scanned PDF & Image to Text for RAG

PDF OCR to Markdown - Scanned PDF & Image to Text for RAG

Convert PDFs, scans, images, DOCX and PPTX to clean Markdown for LLMs and RAG. Text layer used when present, Tesseract OCR for scans (EN, NL, DE, FR, ES, IT, PT, German Fraktur), per-page confidence. $0.00097 per text page, $0.00397 per OCR page.

Pricing

from $3.97 / 1,000 ocr pages

Rating

0.0

(0)

Developer

DigiVibeTech

DigiVibeTech

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

6 days ago

Last modified

Features

Limited permissions

Share

What does PDF OCR to Markdown do?

It turns PDFs, scanned documents, images, Word (DOCX, with headings, lists and tables) and PowerPoint (PPTX) files into clean Markdown that you can feed to an LLM, a RAG pipeline, a search index or a spreadsheet.

  • Pages with a real text layer (born-digital PDFs) are extracted directly: fast, exact and cheap.
  • Scanned pages and images are read with Tesseract 5 OCR, the open-source engine, which runs entirely inside the Actor. Your files are not sent to any third-party AI service.
  • Every page gets an OCR confidence score. Pages that can't be read (blank, handwriting, very poor scans) are flagged and not charged.
  • RAG-ready: optional chunks (paragraph-aware, with page numbers) at no extra cost.
  • German Fraktur (blackletter) is supported, alongside English, Dutch, German, French, Spanish, Italian and Portuguese.

Why this one

  • Text-layer pages are extracted directly ($0.00097) and only scanned pages use OCR ($0.00397), so mixed PDFs cost less.
  • Tesseract 5 OCR runs inside the Actor: your files are not sent to a third-party AI service. Every page gets an OCR confidence score.
  • Blank and unreadable pages are flagged and not charged, and RAG chunks with page numbers are included at no extra cost.
  • PDF, images, DOCX and PPTX in, Markdown out, including German Fraktur.
  • A 10-page scan costs $0.04 here. Pdf OCR API by cspnair, a popular PDF OCR actor on Apify, charges $0.01 per PDF (checked 29 Oct 2026), so it is cheaper for long scans and we are cheaper for short or mostly digital PDFs.

Screenshots

From a real run on 3 Nov 2026 (Europe/Amsterdam) on pages 1-2 of a public-domain 1946 Civil Aeronautics Board accident report scan (Wikimedia Commons): 1 document, 2 scanned pages read with OCR (mean confidence 0.78), billed as 2 OCR-page events.

Input settings Results table One result item Run cost breakdown

Quick start: example input, output and cost

OCR for scanned PDFs and images: text and Markdown for RAG and LLMs, per-page confidence, German Fraktur. $0.00097 per text page, $0.00397 per OCR page, blank pages free.

Example input

{
"fileUrls": [
"https://upload.wikimedia.org/wikipedia/commons/3/33/CAB_Accident_Report%2C_National_Airlines_Flight_23.pdf"
],
"pageRange": "1-3"
}

Cost example: A 50-page scanned PDF x $0.00397 = about $0.20; a 50-page digital PDF x $0.00097 = about $0.05.

Example output (real run, 2026-11-04 UTC)

From a real run of this actor on a public-domain 1946 scanned accident report PDF from Wikimedia Commons, pages 1–3, English OCR (plus a missing file and a blocked internal address, which are not charged). One file, so the excerpt has 1 record. The scan is old, so the OCR text has typos; per-page confidence is in warnings. Markdown shortened:

[
{
"sourceUrl": "https://upload.wikimedia.org/wikipedia/commons/3/33/CAB_Accident_Report%2C_National_Airlines_Flight_23.pdf",
"status": "ok",
"fileType": "pdf",
"pageCount": 12,
"pagesProcessed": 3,
"pagesTextLayer": 0,
"pagesOcr": 3,
"pagesLowConfidence": 0,
"meanOcrConfidence": 0.797,
"billedPages": {
"page-text": 0,
"page-ocr": 3
},
"languages": [
"en"
],
"markdown": "<!-- page 1 -->\n\nDocket No, SA~109 ee Pile No. 3425-h5 CIVIL AERONAUTICS BOARD\n\nACCIDENT INVESTIGATION REFORT\n\n4\n\nJuly 8, 15H6 Released: July 12, 1946\n\nAdopted:\n\nRATIONAL AIRLINES - BAUANA RIVER, FLORIDA, OCTOBER 11, 19h5\n\nThe Accident\n\nNational Airlines! Plight 23 of October 10, 195, en reute from …",
"characters": 5410,
"pages": [
{
"page": 1,
"method": "ocr",
"markdown": "Docket No, SA~109 ee Pile No. 3425-h5 CIVIL AERONAUTICS BOARD\n\nACCIDENT INVESTIGATION REFORT\n\n4\n\nJuly 8, 15H6 Released: July 12, 1946\n\nAdopted:\n\nRATIONAL AIRLINES - BAUANA RIVER, FLORIDA, OCTOBER 11, 19h5\n\nThe Accident\n\nNational Airlines! Plight 23 of October 10, 195, en reute from Mesa York, N, Y., …",
"confidence": 0.768,
"rotatedDeg": 0,
"billedAs": "page-ocr",
"lowConfidence": false,
"warning": "medium OCR confidence 0.77 - check this page"
},
{
"page": 2,
"method": "ocr",
"markdown": "- Tower that the rurmay lights Wore inoperative and that considerable delay\n\na Slight turbulence at that flight altitude, Captain Stoia requested change -\n\nof clearance to 000 feet. Ap\n\nproval for change of altitude was obtained | 7 - after a delay due to traffic, and shortly after passing Daytona B …",
"confidence": 0.79,
"rotatedDeg": 0,
"billedAs": "page-ocr",
"lowConfidence": false,
"warning": "medium OCR confidence 0.79 - check this page"
},
{
"page": 3,
"method": "ocr",
"markdown": "was the cirection of landing intended, However, shortly before the 2ir- 5 4 . ‘ . ~ - Qs craft arrived over the field, the tover personnel shanged the rumiay iights\n\nto Noe 15 since that runway was the longest one available at Banana River.\n\nbh\n\nThe aircraft passed over the field at an altitude of 1 …",
"confidence": 0.834,
"rotatedDeg": 0,
"billedAs": "page-ocr",
"lowConfidence": false
}
],
"warnings": [
"page 1: medium OCR confidence 0.77 - check this page",
"page 2: medium OCR confidence 0.79 - check this page"
],
"chunks": null,
"engine": "Tesseract 5 OCR (Apache-2.0) + PDFium text layer (pypdfium2, Apache-2.0/BSD)",
"processingSec": 3.7,
"processedAt": "2026-11-04T06:31:41.354788+00:00"
}
]

Full sample from the same run (≤50 rows): JSON · CSV

Try it: run with {"fileUrls":["https://upload.wikimedia.org/wikipedia/commons/3/33/CAB_Accident_Report%2C_National_Airlines_Flight_23.pdf"],"languages":["en"],"pageRange":"1-3"}. That is 3 scanned pages × $0.00397 (page-ocr) = $0.0119 (about one cent). Pages with a text layer are $0.00097 each. Apify's free plan includes $5 of prepaid usage every month, so this run fits in it many times over.

Why use it?

This Actor
Price per scanned (OCR) page$0.00397
Price per text-layer page$0.00097
Start fee per runnone
Blank, unreadable or failed pagesfree
Accuracypublished (see the benchmark below)
Per-page confidenceyes
Historic German Frakturyes

How much does it cost?

You pay per successfully extracted page, and only after the result is saved:

EventPriceWhen
page-text$0.00097the page has a usable PDF text layer; also each PPTX slide or ~3,000 characters of DOCX text
page-ocr$0.00397the page was read with OCR

Some examples:

  • 100 born-digital PDF pages cost about $0.10 ($0.097).
  • 100 scanned pages cost about $0.40 ($0.397).
  • A 300-page scanned book costs about $1.19.

Set a maximum cost per run and the Actor will never go over it. Pages that don't fit are skipped and explained, not charged. Apify's free plan credit covers about 1,000+ scanned pages a month.

Input

{
"fileUrls": ["https://example.com/scan.pdf", "https://example.com/photo-of-page.jpg"],
"languages": ["nl", "en"],
"ocrMode": "auto",
"pageRange": "",
"maxPagesPerFile": 200
}
  • fileUrls: direct links to PDF, DOCX, PPTX, PNG, JPG, TIFF (including multi-page), WEBP, GIF or BMP files, up to 200 MB each. Share links to public Google Drive and Dropbox files are converted automatically.
  • languages: up to 3 languages. Choose de-fraktur for old German print.
  • ocrMode: auto uses the text layer when present and OCR otherwise; always OCRs every page; never uses the text layer only.
  • pageRange: for example 1-5, 2,4,9 or 10-.
  • chunkSize / chunkOverlap: set e.g. 1500 / 150 to get a chunks array for embeddings. Each chunk lists its source pages.

Output

You get one dataset item per document:

{
"sourceUrl": "https://example.com/scan.pdf",
"status": "ok",
"fileType": "pdf",
"pageCount": 12,
"pagesOcr": 12,
"pagesLowConfidence": 1,
"meanOcrConfidence": 0.91,
"billedPages": {"page-text": 0, "page-ocr": 11},
"markdown": "<!-- page 1 -->\n\nFirst paragraph ...",
"pages": [{"page": 1, "method": "ocr", "confidence": 0.93, "markdown": "...", "billedAs": "page-ocr", "lowConfidence": false}],
"warnings": ["page 7: low OCR confidence 0.41 (...) - not charged"]
}
  • Each page starts with an invisible <!-- page N --> marker, so you can cite or split by page.
  • Very large documents (over 2 million characters) store the full Markdown in the run's key-value store, and the dataset item links to it.
  • Files that fail are listed in the ERRORS key-value record with the reason. They are never charged.

Accuracy benchmark (real scans, not generated pages)

We measured the Actor on 35 real book and archive pages from Wikisource. The scans are public domain, and the ground truth is human-proofread ("validated") text under CC BY-SA. The metric is character error rate (CER); lower is better.

MaterialPagesMedian CERMedian word errorMedian confidence
English, 19th–20th c. print144.8%8.1%0.92
Dutch, modern print51.8%9.0%0.94
German, modern print46.3%13%0.90
German Fraktur, de-fraktur65.8%28%0.86
German Fraktur, plain de (for comparison)613.6%50%0.66
Dutch, 17th–18th c. print213.8%40%0.81
Handwriting (Kurrent)4~71%, not supported–0.33 (not charged)

Some of the remaining error comes from page headers, footnotes and multi-column layouts that the proofread text arranges differently.

Limitations (honest list)

  • Handwriting is not supported. It is detected by low confidence and not charged.
  • Layout: the output is paragraphs in reading order. Tables come out as text lines rather than Markdown tables, and complex multi-column pages may mix columns.
  • No images or formulas: pictures are skipped, and math is OCR'd as plain text.
  • Speed: about 1.5–4 seconds per scanned page per vCPU with one language. Every extra language adds time (with 3 languages it is about 6 s per page). Give the Actor more memory (1 vCPU per 4 GB) to process pages in parallel.
  • Downloads: files must be reachable by a direct link. Servers that block automated downloads (HTTP 403) are reported, not charged.
  • Documents are processed inside your Apify run and are not sent to any external API.
  • Only process documents you have the right to process.
  • It uses Tesseract (Apache-2.0), tessdata_fast models (Apache-2.0) and PDFium via pypdfium2 (Apache-2.0/BSD).

FAQ

Does it work with photos of documents? Yes, as long as they're reasonably sharp and flat. Rotation is detected automatically.

Can I use it through the API, Make, Zapier or n8n? Yes. It works like any Apify Actor: call it with the input JSON and read the dataset.

Found a problem? Open an issue on the Issues tab. We read every one.

Maintained by DigiVibeTech. Questions or bugs: use the Issues tab or digivibetech.com/support; we reply within 2–5 business days.