Text extraction

Extract text to convert high-quality business documents into a simpler file format that can be used by AI models or to find and isolate key pieces of information from documents such as contracts.

Text extraction is powerful for use cases where you want to extract specific entities or categories of information from a document based on the document structure.

Compatibility and specifications

Cloud platforms

Supported input file types

You can extract text from documents in different languages, or from a document that has a mix of multiple languages. Extract text from the following file types:

Supported document formats:

File type Extension Text extraction support Semantic key-value pair extraction
PDF file .pdf Yes Yes
Powerpoint .pptx .ppt Yes Yes
Word document .docx .doc Yes Yes
Excel .xlsx Yes No

Supported image formats:

File type Extension Text extraction support Semantic key-value pair extraction
BMP image .bmp Yes Yes
HEIC/HEIF image .heic .heif Yes Yes
JFIF image .jfif Yes Yes
JPEG image .jpeg .jpg Yes Yes
PNG image .png Yes Yes
TIFF image/multi-image .tif .tiff Yes Yes

Other supported formats:

File type Extension Text extraction support Semantic key-value pair extraction
HTML .html Yes Yes
Markdown .md Yes No
Note: You cannot use the text extraction API to extract key-value pair data from XLSX documents.
Supported output file types

You can store the extracted text in the following file formats:

  • JSON
  • Markdown
  • HTML
  • TXT

For details about the contents of the extracted result in each output file type, see Specifying the output format.

Supported storage types

You can store your input documents in the following connected storage types:

  • IBM Cloud Object Storage

  • Amazon S3

  • Any generic Amazon S3-compatible storage

  • Box

  • IBM watsonx.data SharePoint

  • IBM FileNet P8

    Note: The IBM FileNet P8 connection is only available in the Toronto data center and for a managed cloud service provider (MCSP).

You can store the text extraction output files in the following connected storage types:

  • IBM Cloud Object Storage

  • Amazon S3

  • Any generic Amazon S3-compatible storage

  • Box

    Note: The text extraction API is certified for use with the generic Amazon S3-compatible MinIO object storage.

For details about how to create a connection to the various types of data stores in your project, see Connectors for watsonx.ai.

Supported foundation models

The text extraction API is certified to use the mistral-small-3-1-24b-instruct-2503 model for key-value pair extraction and image verbalization. You can also use alternative models that can process visual input and respond in a JSON format such as:

  • llama-4-maverick-17b-128e-instruct-fp8
  • mistral-medium-2505

For foundation model details, see Supported foundation models.

Supported languages
  • For scanned PDF documents and images, text extraction uses Optical Character Recognition (OCR). Supported languages depend on the OCR language models.
  • For programmatic documents, such as HTML, Markdown, and Microsoft Word, PowerPoint, and Excel files, text extraction supports all languages.

Image verbalization is supported for documents in all languages. However, the generated image descriptions are returned in English, regardless of the language of the input document.

Semantic key-value pair extraction is officially supported for the following languages:

  • Chinese
  • English
  • French
  • German
  • Italian
  • Japanese
  • Portuguese
  • Spanish

When using the API, specify the corresponding language code or script supported by OCR.

Restrictions

  • You can extract text from specific input file types and store the extracted output in certain file types. Every input file type cannot be extracted into every supported output format. The following table provides details about which input file type is compatible with the various output formats:

    Input file type and extracted output format compatibility for the text extraction API
    Input file type Compatible output file formats
    Programmatic PDF All formats
    Scanned PDF All formats
    Image All formats
    Microsoft PowerPoint file All formats
    Microsoft Word file All formats
    Markdown All formats
    Microsoft Excel file Markdown, JSON, plain text
    HTML file Markdown, JSON, plain text

  • Key-value pair extraction results are returned only in the assembly output format, so key-value pair data is not included in HTML, Markdown, or plain text output formats.

Ways to work

You must generate credentials to authenticate with watsonx.ai APIs. For details, see Generating a bearer token.

You can extract text from documents stored in your watsonx.ai project with these programmatic methods:

REST API

You can extract text from files in IBM watsonx.ai programmatically by using the text extraction method of the watsonx.ai REST API.

For details about how to customize a text extraction request, see Text extraction parameters.

For API method details, see the watsonx.ai API reference documentation.

Python

You can extract text from files in IBM watsonx.ai programmatically by using the Python library.

See the TextExtractionsV2 class of the watsonx.ai Python library.

Try the sample notebook: Use the watsonx.ai Text Extraction V2 service to extract text from file.

Node.js

You can extract text from files in IBM watsonx.ai programmatically by using the Node.js SDK. For more information, see the following resources:

To learn more, see the code example.

Learn more