GETTING STARTED
Text Nodes
Text nodes run large language model (LLM) operations on text. Use them to write, rewrite, summarize, extract information, or compose prompts that feed into downstream generation nodes.
What you can do and how to use it
Text to text — give the model an instruction and a block of text; receive a transformed version.
Image to text — connect an image input; the model describes or analyzes the image.
Prompt composition — write a rough idea and have the model expand it into a detailed generation prompt.
Data extraction — pull structured information (names, descriptions, tags) from unstructured text.
How to use it:
Add a Text node to the canvas
Select a model from the model picker
Optionally set a system prompt — a persistent instruction that shapes every output from this node
Type your input or connect an upstream node
Click Generate
The output is a text string. Wire it into a Prompt Helper, Image, Video, or Audio node to use it downstream.
Tips for using Text nodes
Stack a Text node before an Image node to auto-write a detailed visual prompt from a rough brief. Your brief goes in, a polished prompt comes out, and the Image node generates from it directly.
Extract and fan out: Use a Text node to extract a list of product names or descriptions from a document, then pass the output to a List node to fan them out into individual Image generations.
System prompts let you set a persistent persona or formatting rule for the node — for example: “Always respond in plain text with no markdown. Maximum 50 words.”
Quick Navigation