Hey developers! Ever felt like working with APIs is a bit like solving a tricky puzzle about Function Call? OpenAI’s API has given us some real challenges.
One major issue? Thread expiry. You craft the perfect prompt, we submit the output to the run which “requires_action” and… nothing. The run of the thread expires. It’s like waiting for a reply that never comes.
This often happens, almost 85/90% of times.
Have any of you faced similar issues? How did you overcome them?
I wrote in Support Chat but it’s getting days to get a reply. After almost a week, still in the same situation.
I haven’t had any issues submitting tool outputs to run. I’m using the streaming flag in my application, which helps with the response time. The amount of information sent for the tool outputs also impacts the response time. But overall, it has been consistent.
Can you elaborate more on your solution? Are you building an application or doing it directly through the API? If possible, please share the code.
The AI that wrote this seems to have been incapable of relaying for you the notion of whether it is a case of using tools, receiving a tool_call, and providing a tool return to the Assistants run status that is met with challenges.
I avoided challenges by sticking with Chat Completions, and receiving token stream chunks back within a second.
Ever felt like working with APIs is a bit like solving a tricky puzzle about Function Call? OpenAI’s API has given us some real challenges.