Developer Resources

Every conversion runs as an asynchronous job. Submit a file and an operation to POST /api/v1/jobs and the API immediately answers with a job ID and a status of processing, rather than waiting for the conversion to finish. That keeps the connection short even for large files or slower operations like OCR.

From there, poll the job status endpoint at your own pace until the status changes to completed, failed, or rejected. A completed job includes a download URL and an expiry time; a failed or rejected job includes an error message explaining what went wrong, such as a corrupted file or an unsupported operation.

Once a job is completed, the download endpoint returns the converted file with the right content type and filename headers, ready to save or stream straight into your own storage. Download links stop working after their expiry window, so pipelines should fetch the result promptly instead of caching the URL for later.

Every API request needs an API key, sent in the X-API-Key header. Keys are created and revoked from your account's API Keys page, and each one carries its own per-minute and per-day credit allowance, maximum file size, and maximum page count.

There is no separate login step beyond the header: a request with a missing or invalid key is rejected before any file is read, so a bad key fails fast instead of burning a conversion slot.

Every plan enforces its own requests-per-minute and requests-per-day ceiling, along with a maximum file size and page count; going over it returns a rate-limit error until the window resets. Because conversion is asynchronous, one slow OCR job never blocks your other requests — you can submit several jobs back to back and poll each one independently.

For batch or scheduled work, treat the API as fire-and-forget: create the jobs, store their IDs, and check status later instead of holding a connection open while a job runs.

The command-line tool wraps this same API for one-off conversions and quick scripts: it uploads a file, polls for you, and downloads the result in a single command, with no HTTP client to write. That's the faster path when a person is running conversions from a terminal or gluing them into a shell script.

Call the API directly when you're integrating conversion into a backend service, need webhook notifications on completion, or want full control over retries, concurrency, and error handling. The two aren't exclusive — many teams script the CLI for local tooling and CI, and call the API from the application that serves their own users.