Document OCR
The general reader behind the others. Bounding boxes in the pixels of your own page, tables as rows rather than a run-on string, and markdown you can put straight into a document. Streams as it reads, or answers in one body.
POST /api/v1/ocrDrop a PDF or image, paste, or click to choose
Up to 5MB · PDF, PNG, JPEG, WebP, TIFF
Contextual Privacy: Your document may contain personal information.
POST /api/v1/ocr/stream Text appears as it is read. Server-sent events.
Put this in your own product
One POST. The endpoint above is the one you would call.