Get Embeddings

Controlled node

Overview

The Get Embeddings node converts text into numerical vector embeddings using AI embedding models. These embeddings represent the semantic meaning of the text as high-dimensional vectors, which can be used for similarity comparisons, semantic search, clustering, or as input features for machine learning workflows.

The node automatically handles both single text strings and arrays of text, outputting either a single embedding vector or an array of vectors respectively. This makes it efficient for batch processing multiple documents or text chunks simultaneously.

Model options

Intellectible provides embedding models optimized for different use cases:

Model	Description	Dimensions
V1	Default lightweight embedding model using Qwen3 0.6B. Fast and cost-effective for most semantic similarity tasks.	1024

Batch Processing

When processing multiple texts, the node automatically batches requests (up to 8 texts per batch for the V1 model) to optimize performance and token usage.

Inputs

Input	Type	Description	Default
Run	Event	Triggers the embedding generation process.	-
Model	Enum	The embedding model to use for generating vectors. Currently supports V1.	V1
Text	Text	The input text content to embed. Accepts either a single string or an array of strings. Must be connected from another node (not exposed in panel UI).	-

Outputs

Output	Type	Description
Done	Event	Fires when the embedding generation is complete.
Embeddings	Data	The vector embedding(s) of the input text. Returns a single array of numbers for single text input, or an array of arrays for multiple texts. Returns null if input is empty or invalid.

Runtime Behavior and Defaults

• Input Handling: The node automatically detects whether the Text input is a single string or an array of strings. For single text, it outputs a single embedding vector. For arrays, it outputs an array of embeddings maintaining the same order as the input.
• Model Default: If no model is specified, the node defaults to the V1 model (Qwen3 0.6B).
• Null Handling: If the Text input is null, undefined, or empty, the node outputs null for embeddings and fires the Done event.
• Type Conversion: All text inputs are automatically converted to strings before embedding generation.
• Token Costs: Embedding generation consumes Intellectible tokens based on the amount of text processed. The V1 model uses the standard embedding token pricing.

Example Usage

Connect the output of a Text node, Read Document node, or Chunk Text node to the Text input of Get Embeddings. Trigger the Run event to generate embeddings, which can then be:

1. Stored in a vector database using the Vector Search Database node for semantic search
2. Compared using similarity metrics in the Compare node or custom code
3. Used as features in machine learning pipelines via Run Python Code or Run JavaScript Code nodes

Example workflow: Read a document → Chunk Text → Get Embeddings → Vector Search Database (to store for later retrieval)