Skip to main content

​
Overview

Configuration options control how Fintelite AI processes your documents. These settings can be applied at two levels:
  • Template Level: Set defaults when creating templates
  • Request Level: Override template settings in individual predict requests

​
Available Configurations

Parser Configuration

Control document parsing behavior and parser modes

Chunking Configuration

Split large documents into chunks for better processing

Citations

Track which document blocks extracted data came from

​
Key Configuration Options

​
Preset Mode (Quick Setup)

Use preset configurations for common scenarios:
PresetBest For
FASTSpeed-critical applications, simple documents
BALANCEDGeneral use, good quality/speed balance
PRECISEHigh accuracy requirements, complex documents
BUDGETCost optimization, self-hosted deployment

​
Manual Configuration

Configure parser and extraction modes individually:

​
Configuration Priority

When both template and request have configuration:
  1. Request config takes highest priority
  2. Template config used if request doesn’t specify
  3. System defaults used if neither specifies

​
Quick Reference

​
Configuration Options

ConfigurationPurposeDefaultAvailable In
use_parserEnable document parsingfalseTemplates, Requests
parser_modeParser quality level (LITE, PLUS, PRO)LITETemplates, Requests
extraction_modeAI model identifierNEXUSTemplates, Requests
preset_modeQuick preset (FAST, BALANCED, PRECISE, BUDGET)-Templates, Requests
use_chunkEnable chunkingfalseTemplates, Requests
chunking_configChunking strategy & size (template)-Templates
chunk_strategyChunking strategy (request override)-Requests
chunk_sizeChunk size (request override)-Requests
enable_citationsTrack data sourcesfalseTemplates, Requests

​
Request-Only Options

OptionPurposeDefault
force_urlForce URL output typefalse
Template vs Request Config: Templates use nested chunking_config object, while request overrides use flat chunk_strategy and chunk_size fields. See API Reference for complete examples.

​
Common Configurations

Quick setup for speed-critical applications:
{
  "preset_mode": "FAST"
}
Equivalent to: parser_mode: LITE, extraction_mode: FREYA
Recommended for general use:
{
  "preset_mode": "BALANCED"
}
Equivalent to: parser_mode: PLUS, extraction_mode: ODIN
Best for complex documents requiring verification:
{
  "use_parser": true,
  "parser_mode": "PRO",
  "extraction_mode": "ODIN_PRO",
  "enable_citations": true
}
For long contracts and reports:
{
  "use_parser": true,
  "use_chunk": true,
  "chunking_config": {
    "strategy": "PAGE",
    "target_chunk_size": 2
  }
}
Fast processing without parsing step:
{
  "use_parser": false
}
Note: Citations require use_parser: true. Only PAGE chunking strategy is available without parser.

​
Learn More

Parser Configuration

Parser modes and settings

Chunking Configuration

Chunking strategies explained

Citations

Citation format and usage