n8n
n8n copied to clipboard
OpenAI Assistant & gpt4-o
Bug Description
I'm getting an error while requesting the OpenAI Assistant using the gpt-4o model as an assistant with APIv1, because gpt-4o cannot be used with the APIv1 schema.
Problem in node ‘OpenAI Assistant‘ 400 The requested model 'gpt-4o' cannot be used with the Assistants API in v1. Follow the migration guide to upgrade to v2: https://platform.openai.com/docs/assistants/migration.
To Reproduce
To reproduce this problem you can use this public example: https://n8n.io/workflows/2098-chat-with-openai-assistant-by-adding-a-memory/
And please update your model as below :
Expected behavior
The expected behavior is a working assistant.
Operating System
Ubuntu Linux 22.04
n8n Version
1.39.4
Node.js Version
18.16.0
Database
MySQL
Execution mode
main (default)
Hey @mbakgun,
Using the example provided with a GPT4-O model it appears to be working as expected, Looking at your screenshot though rather than using an existing assistant are you trying to make a new one?
I have also tried this and for me it is working in 1.41.0, Are you able to provide an example that reproduces this?
Hi @Joffcom, I'm trying to create a new assistant, you can find the whole settings.
https://github.com/n8n-io/n8n/assets/8726393/9d7ab7ae-fd14-4ae1-b288-1b3f4051f6b9
Hey @mbakgun,
Even following that I am not able to reproduce the issue, Is your OpenAI account fairly new? We are going to need to update the node so I have created NODE-1350
as the dev ticket to get this updated.
Nope @Joffcom, I've been around for more than 2 years. I believe @OlegIvaniv's pr addresses the required changes. I'll update here as soon as I get that update.
@mbakgun it does indeed, We were chatting about it internally. It is mix of old node and newer n8n version.
Alright, thanks :) I'm looking forward to test it
Edit : With the version of 1.41.0, it worked well 🚀
Opening until the PR is merged and released 🙂
This is working fine. I have tested it. The only thing that it is still a bug is that every message sent via chat, ends up creating a new Thread. But this is a separate bug
Fix got released with [email protected]