Azure OpenAI Chat Model try to use responses API instead of chat completions API

Questions
1

Describe the problem/error/question

I upgraded n8n, and my “Azure OpenAI Chat Model” nodes no longer work (HTTP 404 error; the full error message is below).
I’ve done some testing, and the issue seems to occur with versions above 2.20.0. Everything works correctly with versions up to 2.19.5.
I noticed that this release upgraded the OpenAI dependency from 6.9.0 to 6.34.0, so the issue might be related.
From my tests, it looks like the node is trying to call an endpoint that uses the Responses API instead of the Chat Completions API.
With versions < 2.20.0:
https://myurlazure.net/openai/deployments/mydeploymentname/chat/completions

With versions >= 2.20.0:
https://myurlazure.net/openai/deployments/mydeploymentname/responses

The “OpenAI Chat Model” node allows the use of the Responses API through a checkbox option, but the “Azure OpenAI Chat Model” node does not provide this option:

Is it planned to add Responses API support to the “Azure OpenAI Chat Model” node, or is this behavior considered a bug that will be fixed?

What is the error message (if any)?

The resource you are requesting could not be found

404 status code (no body) Troubleshooting URL: MODEL_NOT_FOUND - Docs by LangChain

Please share your workflow

(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)

Share the output returned by the last node

The resource you are requesting could not be found404 status code (no body) Troubleshooting URL: MODEL_NOT_FOUND - Docs by LangChain

Information on your n8n setup

  • n8n version: 2.20.0
  • Database (default: SQLite): Postgresql
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app): Docker
  • Operating system:
2

Hi @SEBASTIEN_LANDRE

Your diagnosis is spot on, this is a regression in 2.20.0. The OpenAI SDK bump flipped the Azure node to the Responses API endpoint, and unlike the OpenAI Chat Model node it has no toggle to switch back to chat/completions, so a deployment that doesn’t support Responses just 404s.

It’s a bug. It’s already on GitHub: there’s an open issue asking for the Azure node to get the same Use Responses API toggle the OpenAI node has (n8n-io/n8n #30964). Worth adding your version pinpoint there.

For now, pin back to 2.19.5, which you confirmed works. On Docker just set the image to n8nio/n8n:2.19.5 until the toggle lands.