Parsing Modes

Parse4ai offers different parsing modes to suit your needs.

Model Versions

Pipeline Mode (Default)

• Standard parsing pipeline
• Supports OCR, formula, and table recognition
• Best for general document parsing

VLM Mode

• Vision Language Model based parsing
• Advanced understanding of document structure
• Ideal for complex layouts

Parsing Options

OCR (Optical Character Recognition)

Enable OCR for scanned documents or image formats.

Formula Recognition

Extract and recognize mathematical formulas from documents.

Table Recognition

Detect and extract tables with structure preservation.

Language Support

Specify document language for better accuracy (default: Chinese).

Page Ranges

Parse specific pages using page range syntax:

• "1-5": Pages 1 to 5
• "2,4-6": Page 2, pages 4 to 6
• "2--2": From page 2 to second-to-last page