PDF to Markdown and Text: Tables, Pages, Metadata for LLMs

Give it PDF links or upload PDF files, and get back clean Markdown and plain text, ready for an LLM prompt, a RAG pipeline or a search index. Ruled tables come out as Markdown tables, big titles as headings, and every page also comes on its own with its own text.

It runs on your files only. Nothing is scraped, there are no logins and no proxies, so runs are quick and cheap: $0.003 per PDF of up to 50 pages.

What you get

For every PDF, one result with:

• Markdown of the whole document: headings, paragraphs, bullet lists and ruled tables as Markdown tables. Pages are marked with <!-- page N --> comments, so you can split or cite by page.
• Plain text of the whole document, with table rows as tab-separated lines.
• Per-page output (optional): Markdown, text, table count and a scanned flag for each page.
• Metadata: page count, title, author, subject, keywords, the app that made the file, creation and modification dates (ISO 8601).
• Scanned pages flagged: pages that are only an image (no text layer) are listed in scannedPages, and needsOcr is true. You know at once which files need OCR, instead of getting empty text back without a warning.
• Counts: words, characters, tables.
• Errors per file: a broken link, a password-protected PDF or a file that is not a PDF gets a row with an error message and is not charged. The other files in the run carry on.

How to use it

1. Paste one or more PDF links into PDF URLs, or upload files under Upload PDFs. (The sample link is a public IRS form. Remove it if you only upload files.)
2. Optional: set Pages (for example 1-3, 7, 10-), Max pages per PDF, or turn off Include per-page output for smaller results.
3. Run it. Download the results as JSON, CSV or Excel, or read them through the API.

Input

Field	Required	Example	Notes
urls	one of urls or files	["https://example.com/report.pdf"]	Up to 1,000 links per run
files	one of urls or files	uploaded in the Console	Stored in your own Apify key-value store
pages	no	"1-3, 7, 10-"	Pages to convert in every PDF. Empty = all pages
maxPagesPerPdf	no	500	1 to 2,000. Later pages are skipped and the result says truncated: true
includePages	no	true	Adds the pages list

{
    "urls": [
        "https://www.irs.gov/pub/irs-prior/fw9--2024.pdf"
    ],
    "pages": "1-3",
    "includePages": true
}

Output example

One dataset item per PDF (here a small 3-page test file with a table and one scanned page; pages is cut to two entries):

{
    "url": "https://example.com/reports/quarterly-report.pdf",
    "fileName": "quarterly-report.pdf",
    "fileSizeBytes": 1991,
    "pageCount": 3,
    "pagesConverted": 3,
    "pageNumbers": [1, 2, 3],
    "truncated": false,
    "title": "Quarterly Report",
    "author": null,
    "createdAt": "2026-01-01T12:00:00+02:00",
    "markdown": "<!-- page 1 -->\n\n# Quarterly Report\n\nSales grew in every region this quarter.\nThe table below lists units sold.\n\n- Online orders doubled\n\n| Region | Q1 | Q2 |\n| --- | --- | --- |\n| North | 10 | 12 |\n| South | 7 | 9 |\n\n## Summary\n\nNorth sold the most units.\n\n<!-- page 2 (scanned image, no text layer) -->\n\n<!-- page 3 -->\n\nEnd of report.",
    "text": "Quarterly Report\nSales grew in every region this quarter.\nThe table below lists units sold.\n• Online orders doubled\nRegion\tQ1\tQ2\nNorth\t10\t12\nSouth\t7\t9\nSummary\nNorth sold the most units.\n\n\n\nEnd of report.",
    "wordCount": 37,
    "charCount": 203,
    "tableCount": 1,
    "scannedPages": [2],
    "needsOcr": true,
    "unreadablePages": [],
    "pages": [
        {
            "pageNumber": 1,
            "markdown": "# Quarterly Report\n\nSales grew in every region this quarter.\n...",
            "text": "Quarterly Report\nSales grew in every region this quarter.\n...",
            "tableCount": 1,
            "charCount": 185,
            "isScanned": false
        },
        {
            "pageNumber": 2,
            "markdown": "",
            "text": "",
            "tableCount": 0,
            "charCount": 0,
            "isScanned": true
        }
    ],
    "error": null,
    "convertedAt": "2026-10-31T16:00:00.000000+00:00"
}

Metadata the file does not have is null, never guessed.

Pricing

Pay per PDF, nothing per page for normal documents:

Event	Price	When
PDF converted	$0.003	Each PDF converted, up to 50 pages
Extra 50 pages	$0.003	Each further block of up to 50 pages in a longer PDF
Actor start	$0.00005	Apify's standard start event, once per run

Examples: 1,000 PDFs of up to 50 pages cost $3.00. One 120-page PDF costs $0.009 (50 pages, then 2 blocks of 50). Files that fail are not charged, a PDF with no text at all (only scanned or blank pages) is not charged, and scanned pages do not count toward the extra blocks.

Before each PDF the actor checks your run's maximum charge. If the next PDF would go over it, the run stops there; if only part of a long PDF fits, it converts the pages that fit and marks the result truncated. A run never goes over the limit you set.

Limits

• Text PDFs only, no OCR. Scanned pages are detected and listed, not read. Run those pages through an OCR tool.
• Tables need ruling lines. Tables drawn with cell borders become Markdown tables. Tables laid out with spacing only (common in academic papers) come out as text lines, in reading order. Very dense grids (more than about 1,500 line crossings on one page, such as graph paper or a tiny-print sheet of 80 rows by 30 columns) and chart-heavy pages also come out as text lines, so a page never stalls the run.
• Headings come from font size. Text clearly larger than the body becomes # or ##, bold text a little larger becomes ###. PDFs have no real heading tags, so this is a best guess.
• Reading order follows the order the PDF draws its text, which is right for most single and multi-column documents. Rotated text (margin stamps, chart labels) is left out when most of the page is upright.
• Files up to 100 MB, up to 2,000 pages per PDF, up to 1,000 PDFs per run.
• Links must be public http or https URLs. The actor respects each site's robots.txt (RFC 9309): if a site disallows automated access to a path, that file gets an error row and no charge.
• Speed: a 6-page form takes about 4 seconds, a 113-page document about 8 seconds on the default memory. For big batches, give the run more memory (up to 1 GB): it finishes faster at the same price, since you pay per PDF.

FAQ

Is this good input for ChatGPT, Claude or a RAG pipeline? Yes, that is what it is for. Markdown keeps headings, lists and tables that plain text loses, and the page markers let you chunk by page and cite page numbers.

What happens to scanned PDFs? You get the metadata, needsOcr: true and the list of scanned pages, so you can route only those files to OCR. Pages of the same file that do have text are converted as usual. A PDF with no text at all is listed but not charged.

Can I convert only some pages? Yes. Set Pages, for example 1-5 or 2, 4, 10-. You are charged by the pages converted, not by the size of the file.

Can I call it from code, a schedule or an AI agent? Yes. Use the Apify API, a schedule, an integration such as Make or Zapier, or an MCP client. It runs with limited permissions and needs no access to your other Apify data.

Do you keep my files? The actor reads each file during the run and writes the results to your own run storage. Uploaded files stay in your own key-value store, under your control.

Is there personal data in the results? Only what is in your own documents. The actor does not collect anything from anywhere else.

Something looks wrong or a file failed. What do I do? Open an issue on the Issues tab with your run link and, if you can share it, the PDF. Issues are answered within a few days.