Hi everyone,
I’m trying to connect an Opus 4.5 model deployed on Microsoft Azure AI Foundry to the AI Agent node in n8n, but I haven’t been able to get it working.
I tried reusing the Azure OpenAI node, assuming it would be compatible, because they both require the same fields:
-
Endpoint
-
Model version
-
Deployment name
-
API key
The model works fine from other tools/SDKs with this configuration, so the Azure AI Foundry deployment itself seems correct. However, in n8n I only manage to get things working when using plain OpenAI; whenever I try to use anything that’s not “direct OpenAI” (in this case Anthropic Opus 4.5), the AI Agent node fails or doesn’t respond as expected.
My questions are:
-
Does the AI Agent node currently only support direct OpenAI? Why?
-
Is there any “official” or recommended workaround to use Azure AI Foundry models (not OpenAI) with the AI Agent node?
-
Is there any update planned to support others Azure Foundry models?
If you need logs, screenshots, or my exact node configuration, I’m happy to share them.
Thanks in advance!

