Skip to main content
Chunking endpoints allow you to manage the text processing components of your flows in Graphor. Chunking nodes split documents into smaller, searchable pieces and generate vector embeddings for RAG (Retrieval-Augmented Generation) applications.

​
What are Chunking Nodes?

Chunking nodes are processing components that:
  • Split Documents: Break large documents into smaller text chunks
  • Generate Embeddings: Create vector representations for similarity search
  • Enable Retrieval: Make content searchable for RAG applications
  • Optimize Performance: Balance chunk size and overlap for better retrieval
Chunking component

​
Available Endpoints

List Chunking Nodes

Retrieve all chunking nodes from a specific flow

Update Chunking Configuration

Modify chunking node parameters and settings

​
Chunking Node Structure

Each chunking node contains: 
{
  "id": "chunking-1748287628685",
  "type": "chunking",
  "position": { "x": 300, "y": 200 },
  "data": {
    "name": "Document Chunking",
    "config": {
      "embeddingModel": "text-embedding-3-small",
      "chunkingSplitter": "character",
      "chunkSize": 1000,
      "chunkOverlap": 200,
      "chunkSeparator": "\n\n"
    },
    "result": {
      "updated": true,
      "processing": false,
      "has_error": false,
      "total_chunks": 420
    }
  }
}

​
Configuration Options

​
Embedding Models

  • text-embedding-3-small: Fast, efficient embeddings
  • text-embedding-3-large: High-quality embeddings
  • colqwen: Multimodal embedding model

​
Splitter Types

  • character: Split by character count (fast)
  • token: Split by token count (language-aware)
  • semantic: Split by semantic boundaries (high quality)
  • element: Split by document elements

​
Key Parameters

  • chunkSize: Characters or tokens per chunk (500-2000 recommended)
  • chunkOverlap: Overlap between chunks (10-20% of chunk size)
  • chunkSeparator: Text separator for splitting (default: “\n\n”)

​
Authentication

All endpoints require API token authentication: 
Authorization: Bearer YOUR_API_TOKEN
Learn how to generate API tokens in the API Tokens guide.

​
URL Structure

Chunking endpoints follow this pattern: 
https://{flow_name}.flows.graphorlm.com/chunking[/{node_id}]
Where:
  • {flow_name}: Your deployed flow name
  • {node_id}: Specific chunking node ID (for updates)

​
Response Format

All endpoints return chunking nodes with this structure: 
{
  "id": "string",
  "type": "chunking",
  "data": {
    "name": "string",
    "config": {
      "embeddingModel": "string",
      "chunkingSplitter": "string",
      "chunkSize": "number",
      "chunkOverlap": "number"
    },
    "result": {
      "updated": "boolean",
      "processing": "boolean",
      "has_error": "boolean",
      "total_chunks": "number"
    }
  }
}

​
Error Handling

Common error responses:
Error TypeHTTP StatusDescription
Authentication401Invalid API token
Flow Not Found404Flow doesn’t exist
Node Not Found404Chunking node doesn’t exist
Invalid Config400Configuration validation failed
Error response format: 
{
  "detail": "Descriptive error message"
}

​
Next Steps

List Chunking Nodes

Explore existing chunking nodes in your flows

Update Configuration

Learn how to modify chunking parameters

Deploy Flow

Deploy your updated flows

Chunking Guide

Advanced chunking strategies