Skip to main content
The V2 configuration format will eventually be deprecated. All V2 API calls will still be functional in the meantime. Please migrate to the V3 version for the latest features and improvements.

Overview

The 2025-10-14 release introduces a restructured configuration format (v3) that provides better organization and clarity. The redesign is mainly a structual change, as underlying API calls should function the same, however new features post V3 will be released on the V3 version. This guide will help you migrate from the Legacy (v2) configuration format to the new format.

Convert your V2 config to V3

Key Changes

1. Input Parameter

The document_url parameter has been renamed to input for clarity: Legacy (v2)
2025-10-14 (v3)

2. Configuration Structure Reorganization

The configuration options have been reorganized into more logical groupings:
  • enhance: AI-powered enhancements (agentic modes, figure summarization)
  • retrieval: RAG-focused settings (chunking, filtering, embedding optimization)
  • formatting: Output format controls (tables, page markers, markup)
  • spreadsheet: Spreadsheet-specific settings
  • settings: General settings (OCR system, timeouts, passwords)

Complete Mapping Reference

Parse Configuration

Basic Options → Multiple Categories

Advanced Options → Multiple Categories

Experimental Options → Multiple Categories

Extract Configuration

Extract Response Format

The extract response format has changed significantly: Legacy (v2)
2025-10-14 (v3)

Migration Examples

Example 1: Basic Parse with Agentic OCR

Legacy (v2)
2025-10-14 (v3)

Example 2: Parse with Multiple Configurations

Legacy (v2)
2025-10-14 (v3)

Example 3: Extract Configuration

Legacy (v2)
2025-10-14 (v3)

Example 4: Spreadsheet Processing

Legacy (v2)
2025-10-14 (v3)

Async Configuration

The async configuration structure remains similar but uses the async parameter: Legacy (v2)
2025-10-14 (v3)

Breaking Changes Checklist

When migrating your code, make sure to:
  • Replace all document_url with input
  • Move ocr_mode="agentic" to enhance.agentic=[{"scope": "text"}]
  • Update ocr_system values (highres/multilingual → standard)
  • Replace table_summary.enabled with retrieval.embedding_optimized
  • Move figure/table enhancements to enhance.agentic
  • Convert boolean flags to list entries where applicable (e.g., enable_change_trackingformatting.include=["change_tracking"])
  • Update spreadsheet clustering values (default → fast, intelligent → accurate)
  • Restructure extract response handling to use nested value/citations format
  • Move extract schema and system_prompt into instructions object
  • Update citation handling in extract to use the new nested format

Need Help?

If you encounter issues during migration:
  1. Check the API Reference for the 2025-10-14 version
  2. Review the configuration examples in the new version
  3. Contact support at support@reducto.ai