I’m running into a weird issue with a GPT I created. The GPT has an action that connects to an API I maintain which has information about concerts.
If I query the API from the preview, it always works and returns correct data.
Once I save the GPT and then query it from the normal GPT interface, it fails with a generic error message. When I ask for more info it tells me “The error encountered was a ClientResponseError …”
The API itself has no issues. I have testing set up via Postman and it always returns results.
A few weeks back I had this working fine in “production”, but then it stopped working after I made some cosmetic changes to the GPT.
Any tips here on how to debug? Seems like a bug with OpenAI if it works in preview but not in production.
not to be that guy, but have you saved the most up-to-date version of your GPT? If so, maybe you could try and connect a different, additional action to your GPT and see if it runs into the same problem. If it doesn’t then maybe do some replacing, where you modify the new Action until it is identical to your first one, maybe with a different operationID for the routes so they don’t get confused. 
all I can think of, personally.
Hi @geoff.maddock ,
I’m seeing this exact behavior in my custom GPT. Did you figure out your issue?
GPT action to an external API works in preview but not the normal GPT interface and usually it chalks it up to authentication error. I’m using an API key which works in Preview, so not sure what’s going on here.
The version history doesn’t appear to show my last several updates and there is a perennial ‘Updates pending’ status in the upper right corner. I’m not sure if that’s a correlation or a cause.
2 Likes
@sarbojitmishra, I have also been having the same issue since yesterday. My API is perfect. I thought it was on my side. For some reason my custom GPT is getting hung down and I do not get my analysis when I ask GPT to look and study the data.
1 Like
