---
title: LLM Embeddings
description: Convert text into a vector embedding for search, similarity, or retrieval workflows.
source: https://www.edgaze.ai/docs/builder/nodes/llm-embeddings
section: builder
---
# LLM Embeddings

> Convert text into a vector embedding for search, similarity, or retrieval workflows.

## Overview

LLM Embeddings turns text into a numeric vector using OpenAI embedding models. The output is infrastructure data , not something customers read directly.

Use it inside larger workflows for semantic search, memory, indexing, or comparing how similar two pieces of text are.

```docsgraph
llm-embeddings
```

## Ports

- **Input , Text:** The string to embed , connected upstream content or fallback text from the inspector.
- **Output , Embedding:** An array-like vector for downstream similarity or storage steps.

## Inspector

- **Text (fallback):** Used when the Text input is not connected.
- **Model:** Default small model is cheap and strong for most retrieval jobs; large model suits premium memory workflows.

## Tips

- Feed clean, meaningful text , garbage in produces weak vectors.
- Keep this block inside the graph; pair it with Workflow Output only if customers need to see raw vectors.
- Do not confuse embeddings with visible copy from LLM Chat.
- Requires an OpenAI key in the run modal (or Edgaze-hosted run where applicable).

## Related

- [Workflow Studio](/docs/builder/workflow-studio)