What counts as a page

Every document is measured in pages. The page count decides both the cost and whether the document fits under the parser's page limit.

Counting

The count is read when the document is received and returned as metadata.page_count by Get a document.

Cost

cost = pages × the parser's per-page rate

A 10-page PDF costs ten times as much as a single image. With has_multiple_entries=true, the document is still charged by its pages, however many entries it produces. See Plans and credits.

Page limit

Each parser has a Maximum Pages setting (default 100). Documents over the limit are rejected before processing, so they're never charged:

{
  "status": "error",
  "code": "doc:unknown",
  "message": "Document exceeds maximum page count limit of 100.Provided: 140 pages.",
  "data": null
}

The limit protects you from accidentally processing a huge file. Raise it in the parser's settings if your documents are legitimately long, or split the file before sending it.

Keeping costs down

• Send only the pages you need. If the data is on page 1 of a 30-page contract, extract that page first.
• Test in the dashboard on a handful of documents before running a batch.
• Store the document_id of every async submission so a retry after a network error doesn't send (and pay for) the same document twice.