AvailablePDF Split to Markdown
A no-backend PDF-to-Markdown tool that sends pages to Gemini in parallel, estimates cost, and reruns only the pages that need it.
The Gemini key stays in the browser, and the OCR controls (parallel pages, image detail, timeout, and page range) are visible before any conversion runs.
Parallel where it helps, ordered where it matters
PDF Split to Markdown extracts pages lazily and sends a bounded number of OCR requests at once. The default concurrency is four and can be changed before a run. Results can finish out of order, but the final Markdown cannot, so page assembly remains serialized by document position.
The app estimates cost before conversion and replaces the estimate with actual
input and output token usage as results arrive. Automatic retries are limited
to transient 429 and 503 responses. A bad page can be rerun on its own
without paying to process the rest of the document again.
No backend does not mean no network
No application or Folio backend receives or stores the Gemini API key or PDF
pages. The browser sends both directly to Google's Gemini API, with the key in
the x-goog-api-key request header. The pages and key therefore leave the
device for Google to process. That boundary is stated plainly because it is the
meaningful privacy fact.
The source is public.