Back to guides

n8n Integration

Connect n8n chat workflows to LLM Gateway with an OpenAI Chat Model, then verify a real execution.

Connect an n8n OpenAI Chat Model to LLM Gateway to use an accessible model in an AI workflow. This walkthrough uses a chat trigger and an AI Agent.

Video walkthrough

Create a chat workflow

  1. Add On chat message as the first step. The canvas names it When chat message received.
  2. Connect it to an AI Agent node.
  3. Keep the agent's prompt source connected to the chat trigger. Its message expression is {{ $json.chatInput }}.
  4. Click the plus under the agent's Chat Model connector and choose OpenAI Chat Model.

For a first connection test, leave memory and external tools disconnected.

Add the gateway credential

In OpenAI Chat Model, click Connect to OpenAI or create a new OpenAI credential from the credential selector.

Field Value
API Key Your LLM Gateway workspace key
Organization ID Leave blank
Base URL https://api.llmgateway.io/v1

Save the credential and select it on the model node. The OpenAI credential selects the API format; the base URL directs requests to LLM Gateway.

Keep the key in n8n's credential store, rather than in a prompt or workflow expression.

Configure the chat model

  1. Turn Use Responses API off for this Chat Completions walkthrough.
  2. Change the model selector to By ID and enter the exact ID from the live catalogue.
  3. For an agent that calls tools, choose a model with tool support. Set any output token limit within that model's supported range.

With a DevPass key, choose a canonical model included in your plan. Upstream provider prefixes pin routing and are not supported on coding plans.

Run and inspect

Return to the canvas, click Open chat, and send a short prompt. Check the answer and the execution log for successful AI Agent and OpenAI Chat Model steps.

The gateway request belongs to the workspace that issued your key. Review its usage in that workspace's dashboard. n8n's execution details help identify which workflow step made the request.

Once the connection works, add memory or approved tools as needed. Review permissions before connecting actions that change external systems.

Troubleshooting

Authentication fails: check the saved credential, active key, and exact base URL, including /v1.

The model is absent from the picker: use By ID with an accessible model from the live catalogue.

The model request fails: confirm Use Responses API is off for this setup, then inspect the failing model step's error and token settings.

The agent has no prompt: connect the chat trigger and check the chatInput expression.

See the official OpenAI Chat Model reference for node options.