1. Tutorial Introduction

Granite-Docling-258M is a lightweight visual language model launched by IBM in September 2025, designed for efficient document conversion. The model can convert documents into machine-readable formats while fully preserving elements such as layouts, tables, and formulas. The model contains only 258M parameters, has excellent performance, is cost-effective, and supports multi-language processing (including Arabic, Chinese, and Japanese). The model uses the DocTags format to accurately describe the document structure to avoid information loss. Granite-Docling-258M can be seamlessly integrated with the Docling library, provides powerful customization and error handling capabilities, is suitable for enterprise-level document processing, and is a powerful tool in the field of document processing. The relevant paper results are "SmolDocling: An ultra-compact vision-language model for end-to-end multi-modal document conversion", the relevant blog is "IBM Granite-Docling: End-to-end document understanding with one tiny model".

This tutorial uses a single RTX 5090 card as the resource.

2. Project Examples

3. Operation steps

1. After starting the container, click the API address to enter the Web interface

2. Usage steps

If "Bad Gateway" is displayed, it means the model is initializing. Since the model is large, please wait about 2-3 minutes and refresh the page.