ChatGPT error: The app did not respond in-time. It may or may not have completed successfully.
Hello everyone,
I am wondering if you had a similar problem as mine. I just wanted to run a test query to ChatGPT via Zapier and I’ve got an error shown below image:
Here is my “Action” setting, I left “Memory field” key empty:
I am wondering what I did wrong because I’ve got the error: Status Code 400 Bad Request.
Best regards, Bartosz
Page 1 / 1
This is ridiculous. A test should have the same time limit as live. how are we supposed to collect the necessary data???
Hi there @J_M,
Thanks for the update here.
Unfortunately, I have exhausted my resources regarding this error. I’d recommend reaching out to our Support Team regarding this issue. Our Support Team has the correct tools to look closer at the Zap and its logs to determine the cause of the issue.
If the error continues, you may need to look at reducing the data you are sending to ChatGPT or try a different model
Hopefully, this helps.
Hi @ken.a .
Ah, that’s really unfortunate, as I’ve got a complex zap and this step is absolutely essential.
It’s a ChatGPT Conversation with Assistant action, and it works in the Open AI Playground. It takes a while there too, but doesn’t time out and succeeds.
Isn’t it possible to extend the timeout limit? With the timeout limit, it significantly reduces the utility of Assistants, which are designed for and capable of long complex tasks.
Also, the earlier response from @SamB says prompt completions during live Zap runs are allowed up to 15 mins to be completed before a timeout error will occur, and this error occurs well before that. So I’m confused as to why I’m getting the error I’m seeing in the first place?
If the error continues, you may need to look at reducing the data you are sending to ChatGPT or try a different model
Hopefully, this helps.
Hi.
I’m currently experiencing this in a live zap. In the zap history it shows up as an error, as in OP’s example:
“The app did not respond in-time. It may or may not have completed successfully.”
Any idea on how to ensure the zap completes?
Thanks for letting us know, @Lorenzo Lione. I can totally understand the need to have the timeout period increased when testing in the Zap Editor as well.
Since the feature request has already been closed I’d recommend reaching out to our Support team to ask that a new feature request be opened up (I don’t have access to do that on your behalf) - specifically to increase the timeout limits when testing ChatGPT actions. You can get in touch with the Support team here: https://zapier.com/app/get-help
Sorry to not be able to provide a more immediate solution here. If that limit does get increased we’ll be sure to update the thread here to let everyone know!
Hi there folks!
Just popping by with a belated update here to confirm that the feature request mentioned here about having the timeout window increased was actioned!
Prompt completions during live Zap runs are now allowed up to 15 mins to be completed before a timeout error will occur. That said, testing in the Zap Editor is still limited to a timeout period of 50 seconds so these may still time out and return a partial response. But when the Zap runs live the response should come through in full.
Happy Zapping!
Hi SamB,
thank you for the update and the efforts to extend the timeout window during live Zap runs to 15 minutes. This enhancement is indeed a significant improvement for production scenarios.
However, I'd like to bring to your attention a critical aspect regarding the testing phase. Currently, with the testing in the Zap Editor being limited to a timeout period of only 50 seconds, it's challenging to obtain the necessary data for subsequent Zaps. This limitation hinders our ability to configure the flow correctly, as the tests do not reflect the real-world conditions we have in production.
For a more effective and seamless integration process, it is essential to have a similar timeout threshold during the testing phase as in the live environment. Without this, the testing phase becomes somewhat ineffectual, as we can't fully validate the Zaps under conditions that mimic live scenarios.
Understanding the constraints and complexities involved, I still believe that aligning the testing timeout more closely with the live environment's timeout would be highly beneficial. It would enable a more accurate and reliable setup, ensuring smoother operations post-deployment.
Thank you for considering this request. Looking forward to a positive response and continued improvements.
L.
Hi there folks!
Just popping by with a belated update here to confirm that the feature request mentioned here about having the timeout window increased was actioned!
Prompt completions during live Zap runs are now allowed up to 15 mins to be completed before a timeout error will occur. That said, testing in the Zap Editor is still limited to a timeout period of 50 seconds so these may still time out and return a partial response. But when the Zap runs live the response should come through in full.
Happy Zapping!
Hey there @ChmieluXx!
As Troy said, it’s likely only a temporary issue. Given the popularity of ChatGPT it’s possible that at the time the action was being tested the OpenAI API was experiencing a high amount of requests which caused it to run a bit slow. If an app’s API doesn’t respond within the set timeframe it’ll cause a timeout error to occur.
I did some checking and it looks like we’ve got a feature request open to increase the timeout window for the Send Prompt action so I’ve added your vote for that. I don’t have a ETA on when that would be implemented but we’ll be sure to email you as soon as it is.
If you’re on a paid Zapier plan and continue to run into this error I’d recommend turning on our Autoreplay feature. That will allow any runs of the Zap that run into that same timeout error to be automatically replayed a bit later, which will hopefully allow it to then go through ok.