Upstage Document ParsingInstall

Summary

Submit a document file asynchronously for parsing by uploading binary file data to the Upstage Document Parsing API, initiating an asynchronous processing request.

Inputs

• Binary Property (required) — Name of the input binary property containing the document file to upload.
• Model (required) — Choose the AI model for parsing; options are 'document-parse' (recommended) or 'document-parse-nightly'.
• OCR (required) — Configure OCR behavior: 'Auto' applies OCR only to image documents; 'Force' applies OCR to all documents.
• Base64 Encoding Categories — Select categories ('figure', 'table', 'equation', 'chart') for which cropped base64 images should be returned.
• Merge Multipage Tables — Optionally merge tables spanning multiple pages into a single table.
• Output Formats (required) — Select one or more output document formats to include in the parsing result, such as HTML, Markdown, or Text.
• Include Coordinates (required) — Whether to include bounding box coordinates for each layout element in the response.
• Chart Recognition (required) — Enable chart recognition to convert charts into tables in the parsed output.

Output shape

a JSON object confirming the submission containing at least a request_id and a submitted status flag

The operation returns immediately with a request ID used to retrieve results later. To get the parsed document, use the 'Async Get Result (By Request ID)' operation with the returned request_id. Errors during submission are included in the output if 'Continue On Fail' is enabled.

Examples

Example 1: Submit PDF document for asynchronous parsing

Provide binary property 'data' containing the PDF file; select 'document-parse' model; leave OCR as 'Auto'; no base64 images returned; do not merge multipage tables; output as HTML only; include coordinates; enable chart recognition.