Question

ChatGPT Assistant Function Calling - Gateway Timeout

  • 14 February 2024
  • 4 replies
  • 86 views

Hey everyone,

I’m having a bit of trouble with Zapier and ChatGPT. When I do simple calls to ChatGPT through Zapier, everything works fine. But, when I try to do function calls, Zapier keeps timing out before getting a response from the GPT Assistant.

Has anyone faced this? How do you stop Zapier from timing out on these function calls? Any tips or tricks to share would be awesome.
 

 


4 replies

Userlevel 7
Badge +14

Hi @Nishan 

Check the OpenAI Status page: https://status.openai.com

Despite the lengthy wait, the issue persists and continues to generate the same error. OpenAI has resolved everything on their side and it is working through Python. I am sure the timeout is caused on the zappier side.

Userlevel 7
Badge +14

@Nishan 

You can try opening a ticket with Zapier Support: https://zapier.com/app/get-help

Userlevel 7
Badge +11

Hi @Nishan 👋

Just came across your posts here and had a look at the related Support ticket but couldn’t tell whether this was resolved. Not sure if you’d seen the suggestion from the Support team but thought I’d share it here in case it’s helpful to you or anyone else that’s running into similar errors with ChatGPT:

I understand you are getting a timeout error when creating a conversation in ChatGPT. 
 
This happens when the calls to the ChatGPT assistant aren't complete in 3 minutes.
 
I notice you are making use of the Conversation ID field in Step 4. The purpose of this Conversation ID field is to load all previous conversation inputs using this Conversation ID and send it OpenAI/ChatGPT one after another, resulting in a longer processing time. So if you have 20 previous conversations under this Conversation ID, and the current run is the 21st conversation, the inputs for the previous 20 conversations will be sent to OpenAI/ChatGPT one by one in separate API calls to re-create the "Thread", only after the 20 previous conversations have been sent to OpenAI/ChatGPT one by one by, will the contents of the current run get sent to OpenAI/ChatGPT. Since this is a serial process rather than a parallel process, it is relatively time consuming if you have a large number of conversations stored under the Conversation ID.
 
I was wondering whether or not you can remove the Conversation ID field in Step 4 to shorten the processing time. By doing so, this will only send the content in the message field in Step 4 in Open/ChatGPT in a single API call.
 
 
0fd718de013ea5fb5fa416c827ccf90f.png


Please do let us know whether you were able to get these errors solved, keen to ensure you’re all set here! 🙂

Reply