Hello, I would like to know when the updates to be able to communicate again with the chatgpt assistants were going to be deployed?
Since the update of open ai it is impossible to use the assistants because you have to integrate “OpenAI-Beta: assistants=v2” into the header
THANKS
It looks like your topic is missing some important information. Could you provide the following if applicable.
- n8n version:
- Database (default: SQLite):
- n8n EXECUTIONS_PROCESS setting (default: own, main):
- Running n8n via (Docker, npm, n8n cloud, desktop app):
- Operating system:
hello
n8n version 1.4
Database: default
n8n EXECUTIONS_PROCESS setting (default: own, main): i don’t know
Running n8n via npm
- Operating system: Microsfot windows server 2022 datacenter
Thanks @Thim_Ber for posting, moving this to the feature request section! 
Make sure to vote!
1 Like
Hi @Thim_Ber,
As far as I understand it, the latest(OpenAI-Beta: assistants=v2) version of Assistants API shouldn’t break the functionality of v1 but rather enhance the functionality(with some breaking changes), for example the OpenAI Vector Stores.
I’ve just tried to create a new assistant with access to files from both OpenAI interface and n8n. In both cases, I was able to message the assistant and receive appropriate responses, so this functionality should still work.
Could you please elaborate on the issues you’re having with messaging your assistants?
With the new gpt-4o model, it seems the v2 is required for this model to work. Using the OpenAI Assistant node and a gpt-4o based assistant, I get an error directing to this page https://platform.openai.com/docs/assistants/migration.
Both V1 and V2 may be possible for assistants running on earlier models, but for assistants with retrieval (now file_search) etc on the new model, it looks like it is V2 only.
Doesn’t seem possible to use OpenAI Assistant node for assistants using gpt-4o and tools currently.
1 Like
This is fixed now with the latest release.
1 Like
hello, I have just tested and the assistant that I created has instructions and must look for information in the files that it has in its database on openai But when I test, it does not look in the documents for constitute the answer
New version n8n@1.43.0 got released which includes the GitHub PR 9406.
Already there!
If you want to always get the latest releases, use the tag “next” instead of “latest”.
It is not recommended for productions to use “Next,” but personally, I use “next” and take the risks.
It is possible to incorporate the Theadid option under message an assistant? This would be the best way to address assistant memory
1 Like
This is exactly what I’m looking for. I guess we can utilize the OpenAI API to accomplish this in the meantime
Oleg, were you able to make this work? I am trying but unsuccessfully.