Supported file types

File types

Sensible supports the following file types:

For more information, see the following tables.

Operation context

	PDF	Microsoft Word	Spreadsheet formats	JPEG, PNG	TIFF
Sensible app's Extract tab	✅	✅	✅	✅	❌
Single-file extraction with SDKs or API	✅	✅	✅¹	✅	✅
Portfolio extraction	✅	✅	❌	✅²	❌
Email bodies	✅	✅	✅	✅	n/a
Email attachments	✅	✅	✅	✅	✅
Classification by type with SDKs or API	✅	✅	✅³	✅	✅

SenseML extraction method

	PDF	Microsoft Word	Spreadsheet formats	JPEG, PNG	TIFF
Methods that render non-text pixels⁴	✅	✅	❌	✅	❌
NLP Table method, Fixed Table method⁵	✅	✅	❌	✅	❌
Extraction of text that requires OCR	✅	✅	❌	✅	✅

For XLS, XLSM, and XLSB files, use an asynchronous extraction endpoint, such as Extract from URL or Generate an upload URL. If you use the synchronous Extract data from a document endpoint, base64-encode the file in the request body's document field instead of sending the file bytes as the request body. Specify spreadsheet content types in lowercase, for example, application/vnd.ms-excel.sheet.binary.macroenabled.12. For CSV files, the asynchronous extraction endpoints require the content_type body parameter, text/csv.
If you're processing portfolios and expect a mix of image- and non-image files, then you can process single-page image files as portfolios. Sensible returns the first document it identifies in the image file.
Sensible doesn't classify CSV files. To classify XLS, XLSM, or XLSB files with the API, specify the XLSX content type, application/vnd.openxmlformats-officedocument.spreadsheetml.sheet. Sensible detects the file's actual format. The Python SDK classifies these files without you specifying the content type.
Methods that render non-text image pixels include pixel-based methods, such as Box, Checkbox, Nearest Checkbox, and Signature methods, multimodal LLM-based methods, and image coordinates returned by the Document Range method.
As alternatives to these Table methods, use the Fixed Table method or the List method.

Sensible supports the following file sizes:

Operation	Size limit for `/extract/{doc-type}` API endpoint	Size limit for asynchronous calls
Single-document file extraction	under 4.5MB, or under 30 seconds processing time	6 GB
Portfolio extraction	n/a	6 GB
Classification	4.5 MB	4.5 MB

When you submit a document to Sensible, it performs the following file conversions:

Word documents: Sensible converts the document to PDF before processing it.
Email bodies: Sensible converts the body to PDF before processing it.
Spreadsheet documents: All OCR settings are inapplicable for this file type. Sensible extracts text directly from the file without OCR. Sensible represents the text both internally and in the Sensible app's editor as follows:
- Standardizes the formatting of all text in the file. Each cell contains exactly one line.
- Standardizes cell height at 0.25'' tall and cell width at 1''. Overflow text in a cell is still available for extraction but isn't viewable in the Sensible app editor unless you click on a line in the rendered document to view its details.
- Standardizes the maximum page height at 15 inches. Sensible splits longer sheets into consecutive pages.