How to Add a Custom ChatGPT Model in AI Agent Workflow in n8n?

Questions
1

I have created a workflow in n8n using the AI Agent node and added OpenAI. However, I want to use my custom ChatGPT model instead of the default OpenAI models. How can I integrate my custom model into the AI Agent workflow in n8n?

If any additional configurations or API settings are needed, please provide guidance.

2

Hey there! If you’ve created a custom GPT Agent directly on OpenAI’s platform (like an Assistant with specific instructions) and want to use it in n8n, the integration is straightforward. Here’s how:

Step-by-Step Guide to Connect n8n to Your OpenAI Agent:

1. Get Your Agent’s (Assistant) ID from OpenAI

  • Go to the OpenAI Playground or the “Assistants” section in your dashboard.
  • Find your custom Assistant’s ID (e.g., asst_xyz123).

2. Use the “OpenAI” Node in n8n

  • Add the OpenAI node to your workflow.
  • Select the action “Create Thread and Run” (or “Create Assistant” if configuring from scratch).
  • In the Credentials section, add your OpenAI API key.

3. Configure the Node to Use Your Agent

  • In the Assistant ID field, enter your Assistant’s ID (asst_xyz123).
  • Add your Agent’s specific instructions in Instructions (if needed):
    "You are a marketing expert. Always respond in English."

4. Run and Test

  • Add input (e.g., a user query) via a Webhook or Manual Trigger node.
  • Connect everything and execute the workflow.

Example Workflow:

Webhook (User Input) → OpenAI (Create Thread and Run) → Agent’s Response → Output

Additional Configurations (Optional):

  • Base Model: If your Agent uses a customized base model (e.g., gpt-4-turbo), select it in the Model field.
  • Tools: Enable tools like retrieval or code_interpreter directly in the node’s settings.

Key Tip:

If your Agent uses knowledge files (e.g., PDFs for context), upload them via OpenAI’s API first and reference their IDs in the node (e.g., file_abc123).

Need help finding your Assistant ID or structuring instructions? Share more details! :blush:

Cheers,
Dandy

Example Node Configuration:
(Replace with actual screenshot if possible)

Troubleshooting:

  • Use the HTTP Request Node to debug API calls if responses aren’t as expected.
  • Check OpenAI’s Assistant API Docs for advanced settings.
3

It doesn’t look like suitable answer.
I need just to understand how importing external library in N8N.
For example, I am going to import ai image generator product to n8n workflow that made by me.
Currently, n8n looks like only support already imported library now. I am going to add my ai product to n8n. But it is not on the list of n8n.
At this time, what is the solution to solve this problem.

4

Adding to the chain here- for those of us that custom build our own models, is there a way to integrate in n8n? Dandy_Dow answered based on a custom agent built with one of the proprietary LLM models.