PDF parser FAQ for developers and teams

PdfParse is a PDF parsing API and web app. You send a PDF or a scan and get back JSON, Markdown and CSV with tables, form fields, headings and reading order kept, through one REST endpoint, SDKs for Python and Node, webhooks and batch jobs.

Backend and AI engineers who build document intake, RAG pipelines or internal assistants, and operations teams that process invoices, statements, forms and contracts every month. Engineers use the document parsing API, operations teams use the same engine in the web app.

PDFs with a text layer, scanned PDFs, PNG, JPG and TIFF images, and multi page documents up to the file size and page count of your plan.

Yes. The demo on the PDF parser homepage parses page one of any PDF or scan up to 10 MB without an account and without a card. The demo is a one time preview, not a plan.

No. Operations teams upload files in the web app, define fields with a form and export CSV. Developers get the API and SDKs on every plan.

Tables return as rows and columns with merged cells resolved, multi line cells kept together and tables split across pages joined. Each table is also written as its own CSV. See PDF table extraction.

Pages without a text layer go through OCR automatically, including rotated and skewed pages, and return the same structure as digital PDFs. You can force or disable OCR per request. See the OCR API.

Every block and every schema field carries a confidence score and the page it came from. Your code can accept high confidence values and send the rest to a person for review.

Yes. Layout detection orders blocks the way a person reads the page, so two column papers and reports come out as continuous text instead of interleaved lines.

Headings become Markdown headings, lists stay lists and tables become Markdown tables. Page headers and footers are separated from the body. See PDF to Markdown.

Official SDKs for Python and Node.js, with typed results. Any other language works with one multipart HTTP request. See the Python PDF parser and the Node package.

Yes. From the Growth plan you create extraction schemas with field names, types and descriptions, and every document returns exactly those keys in the fields object. See PDF data extraction.

Small files return in the same request. Larger files return a job id right away, and you either poll the job or receive a signed webhook when the result is ready.

Yes. From the Growth plan a batch job takes up to 500 files, and an idempotency key makes retries safe.

Yes. From the Growth plan the response includes chunks split by section, each with page numbers, section path and bounding box. See PDF parsing for RAG.

Starter keeps files for 7 days. Growth lets you choose a period up to 30 days. Scale and Enterprise can turn on zero retention mode, where files and results are removed as soon as the result is delivered. Files in the website demo are parsed in your browser and never uploaded.

No. Files and results are processed only to return your result and are never used to train models.

Yes. Files are encrypted in transit and at rest. Access is controlled per API key, and Scale and Enterprise include an audit log. Details are on the security page.

Enterprise customers choose an EU or US processing region and receive a data processing agreement.

One page of a PDF or one image file counts as one page. Pages count as usage in the billing period, so deleting a document never returns them.

Parsing pauses and you see an Upgrade button. Nothing is charged automatically. An upgrade charges the difference right away and the new allowance applies at once.

Plans start at $49 per month, or $24 per month billed yearly. Every plan and limit is on the PDF extraction API pricing page.

Yes. Cancel in billing settings and access runs to the end of the period you paid for. Payments are final, with the exceptions in the refund policy.