Extract any PDF into CSV, editable Word, RAG-ready Markdown, or a searchable PDF with one API call. Scanned or form-heavy? AI rebuilds it faithfully — layout, tables and checkboxes intact, every value verified. (And yes — convert files to PDF too.) Plug it straight into Claude, Cursor & other AI assistants over MCP. Files are never stored.
No credit card. No setup. First PDF in under 60 seconds.
🎁 Invite a friend — you both get free pages.
MCP Use it from your AI assistant — Claude, Cursor & more →See it in action — AI-rebuilt scanned tables, bulk PDFs → clean data, and RAG-ready Markdown. PDFjet.dev.
Why PDFjet
Extract tables to CSV, rebuild documents into editable Word, or get a searchable PDF — one API call. AI handles the scanned and form-heavy ones. Convert files to PDF and merge/split/encrypt too. Zero data retention.
Tables to CSV, documents to editable Word or RAG-ready Markdown, or a searchable PDF — same endpoint, you pick the format.
Scanned pages, forms, and checkboxes get OCR'd and rebuilt by a vision model — layout, labels, and values intact, every number verified against the source.
Your files are never stored or used to train models. Stateless processing — data in, results out, nothing kept. Standard extractions run entirely on our servers; only optional AI rebuild sends pages to a third-party vision model — opt out with ?engine=off.
Every plan gets the full toolkit. Your files are never stored.
Pricing
Billed per page processed. A 10-page PDF = 10 pages. Every plan includes every feature — Convert, Extract (CSV/Word/Markdown/Searchable PDF), Premium AI & PDF tools. You just choose your monthly volume.
Start for free — no credit card required. Full usage logs from day one.
Cancel anytime. All plans include zero data retention.
Why PDFjet stands out
Pull tables, documents, and scanned forms into CSV, Word, and searchable PDF — then merge, split, and encrypt at sub-second speed, with zero data retention and better pricing than every alternative.
3–5× faster than alternatives
2–5× faster than industry avg
1.5–3× faster (industry: 2–5s)
Fully stateless — competitors store files
| Capability | PDFjet | Adobe | pdfRest | ConvertAPI | PDFShift | PDF.co |
|---|---|---|---|---|---|---|
| PDF → CSV & editable Word | ||||||
| AI document rebuild (vision LLM) | ||||||
| HTML to PDF | 0.5s | |||||
| Office docs (DOCX, XLSX, PPTX) | 1.6s | |||||
| Images (JPG, PNG, TIFF) | 0.5s | |||||
| PDF merge | 0.3s | |||||
| PDF split | 0.4s | |||||
| Password encryption | 0.3s | |||||
| Zero data retention | ||||||
| Per-request audit logs | ||||||
| Direct PDF bytes response | ||||||
| Mac app | Coming soon | |||||
| Free tier | 100/mo | 500/mo | 300/mo | 1,500 | 50/mo | 200/mo |
| Pricing | Free, then $19/mo | enterprise | from $15/mo | from $15/mo | from $9/mo | from $99/mo |
| Operation | Avg | Min | Max |
|---|---|---|---|
| Merge 2 PDFs | 0.31s | 0.29s | 0.35s |
| Split 10p → all | 0.37s | 0.31s | 0.41s |
| Encrypt 1-page PDF | 0.32s | 0.31s | 0.33s |
| PDF pass-through (5p) | 0.34s | 0.29s | 0.41s |
| JPG → PDF | 0.41s | 0.32s | 0.57s |
| PNG → PDF | 0.53s | 0.51s | 0.57s |
| HTML → PDF | 0.58s | 0.36s | 0.92s |
| TIFF → PDF | 1.06s | 1.02s | 1.13s |
| DOCX → PDF | 1.58s | 1.55s | 1.62s |
| XLSX → PDF | 1.56s | 1.55s | 1.57s |
| PPTX → PDF | 1.67s | 1.57s | 1.73s |
Integration
One API key, zero SDKs needed. POST a file over HTTP and get back CSV, Word, Markdown, a searchable PDF — or a PDF. Copy-paste code examples for every language live in your dashboard.
/extract/premium
Pro & Business
Smart AI rebuild for scanned & form-heavy docs — vision model reconstructs layout, tables & checkboxes, every value verified. Deterministic-first; ?engine=ai forces AI, ?engine=off keeps it on-server (no third-party AI).
file(multipart, PDF)X-Engine header/extract/csv
All plans
Detect tables in a PDF and pull them into clean CSV rows. Scanned PDFs are OCR'd first.
file(multipart, PDF)/extract/docx
All plans
Rebuild a PDF as an editable Word document — headings, paragraphs and tables, layout preserved.
file(multipart, PDF)/extract/md
All plans
Get clean, chunkable Markdown — headings, lists and tables — ready to embed for RAG / AI pipelines.
file(multipart, PDF)/extract/searchable
All plans
Turn a scanned PDF into a searchable one — looks identical, with an invisible OCR text layer. Digital PDFs pass through unchanged.
file(multipart, PDF)/convert
All plans
Convert any supported file to PDF. Send the file as multipart form data and receive raw PDF bytes.
file(multipart)passwordencrypt output PDF/merge
All plans
Combine multiple PDFs into a single document. Upload 2+ PDF files and receive one merged PDF.
Tip: Need to merge non-PDF files? Convert them to PDF first via /convert, then merge the results.
files[](multipart, 2+ PDFs)ordercomma-separated indices/split
All plans
Extract specific pages from a PDF. Specify page ranges and get back a new PDF with just those pages.
file(multipart, PDF)pagese.g. "1-3,5,8-10"/encrypt
All plans
Password-protect an existing PDF. Upload a PDF and set a user password to restrict access.
file(multipart, PDF)passworduser open passwordPDFs go in via /extract and come out as CSV, Word, or searchable PDF. Everything else converts to PDF via /convert. More formats added regularly.
Install the PDFjet MCP server and Claude, Cursor, or any MCP client can read PDFs directly — “extract this contract to Markdown,” “pull these tables to CSV,” “make this scan searchable.” Standard extractions stay fully on-server.
View on npm →Pass your API key in the Authorization header as a Bearer token. Keys start with pj_.
Success returns 200 with raw PDF bytes. Errors return JSON with an error field and appropriate status code.
Free tier included — no credit card required. Free: 100 pages/month. Pro: 1,000 pages/month. Business: 10,000 pages/month. Every plan has every feature. Limits reset on billing cycle. 429 when exceeded.
Works with any language that supports HTTP. No SDK required — just POST a PDF, get clean data back.
FAQ
PDFjet extracts any PDF into clean CSV, editable Word, RAG-ready Markdown, or a searchable PDF with one API call — and also converts Word, Excel, PowerPoint, images, and HTML to PDF. AI rebuilds scanned and form-heavy documents.
Yes. Scanned PDFs are OCR'd automatically, and the Premium AI engine rebuilds messy or form-heavy scans into clean output — layout, tables, and checkbox states intact, every value verified against the source.
No. Files are processed in memory and deleted immediately — never stored, never used to train models. Standard extractions run entirely on our servers; only optional Premium AI sends pages to a vision model.
Free includes 100 pages/month. Pro is $19/mo for 1,000 pages (10× the Free tier) and Business is $99/mo for 10,000 pages (100× the Free tier), with annual plans (2 months free). Every plan includes every feature; you're billed per page processed.
Yes. The pdfjet-mcp MCP server lets Claude, Cursor, and other MCP-capable AI assistants extract and convert PDFs directly using your PDFjet API key.
CSV (tables), editable Word (.docx), clean Markdown for RAG and AI pipelines, and a searchable PDF with a selectable OCR text layer.
CSV, editable Word, RAG-ready Markdown, and searchable PDF — with AI for the messy ones. Zero data retention. Full audit logs on every request. Start for free — no credit card required.