Chutes icon

Chutes

Interact with Chutes.ai for AI text generation, image generation, and inference

Actions16

LLM (Text Generation) → Chat

AI-generated

Summary

Send a chat conversation to a selected Chutes.ai Large Language Model chute and receive the generated assistant response.

Inputs

  • Chute URL — Optionally select a specific chute URL for the chat model; can use a static selection or expressions to dynamically specify the chute endpoint. Defaults to 'https://llm.chutes.ai'.
  • Messages (required) — Provide a sequence of chat messages forming the conversation. Each message must specify a role ('system', 'user', or 'assistant') and the message content text.
  • Additional Options — Configure generation parameters such as temperature (randomness), max tokens, nucleus sampling (top_p), frequency and presence penalties, stop sequences, response format (text or JSON), whether to stream responses, seed for deterministic outputs, and timeout in seconds.

Output shape

a single JSON object representing the chat completion response from the chute API

The response object contains the generated assistant message or completion according to the selected chute and options. The response format can be plain text or structured JSON as requested. If streaming is enabled, note that n8n does not currently support real-time SSE streaming, so full responses are returned. If a timeout is set and exceeded, the node will error and can trigger retry workflows.

Examples

Example 1: Basic chat with default chute

Use default chuteUrl. Provide messages with roles 'user' and their content. Leave additional options as default.

Example 2: Use custom chute URL with specific temperature and response format JSON

Set chuteUrl to a custom expression or URL. Set temperature to 1.0 for more randomness. Set responseFormat to JSON to get detailed structured output.

Links

Discussion