From the archive
OpenRouter adds PDF support across its model catalog
Document requests can use native model support, paid OCR for scans, or free text extraction for simpler PDFs.
OpenRouter introduced PDF support across its model catalog, including documents made from scanned images. The feature works through its application programming interface (API) and chat interface.
Choose how a document is read
Where a provider supports PDFs directly, OpenRouter uses that native route. Other requests can use one of two processing options:
- Mistral OCR extracts text and embedded images, including content in scanned pages.
- The free text-only engine extracts existing text without optical character recognition (OCR) or image support.
The choice matters for image-heavy reports and scanned files, which cannot be treated like a PDF with selectable text.
Document processing has a separate price
OpenRouter announced a rate of $2 per 1,000 pages for its Mistral OCR processing option. The resulting content is then used in the model request. The OpenRouter profile covers the gateway, while the Mistral AI profile covers the underlying document and model services.