Estimate credits

Every convert endpoint has an estimate twin at the same path plus /estimate. It accepts exactly the request body the submit endpoint accepts, reads each source to count its pages, and returns the credits the conversion would cost. It never creates a conversion, never reserves or spends credits, and deletes anything it had to stage to read the source.

Markdown to HTML and Markdown to Text cost no credits, so their estimates always return 0; they exist so one client can price any conversion the same way.

Send the body you would send to the submit endpoint: a multipart file or files, JSON url/urls, text/texts, or references to earlier files and completed uploads. Options such as pages or ocr change the estimate the same way they change the conversion.

curl -X POST https://api.markitdown.ai/v1/convert/pdf/estimate \
  -H "x-api-key: $MARKITDOWN_API_KEY" \
  -F file=@report.pdf

• Plan limits. The per-file size limit, the page cap, the monthly file allowance and storage are checked only when you submit, so an estimate can succeed for a file the submit endpoint then rejects.

• Your balance. The estimate returns balance but does not fail when credits_total is higher; the submit endpoint returns 402 insufficient_credits in that case.

• Scope and subscription do apply: an estimate needs the read scope (a read-only key can price conversions) and an active Lite, Pro or Max subscription; otherwise it returns 403 scope_missing or 403 subscription_required.