I am calling GPT-5 to process math problems via an AI Gateway, and I have noticed that the process is very slow, often failing 2-3 times before a successful response is received.
The bottleneck appears to be the AI Gateway’s response time/reliability. How should this issue be addressed
@yangfan85168-5985 When it fails, what error does it log (if any)? It sounds most like a rate-limiting issue to me, but without more context, it’s hard to tell.
Extra context would be helpful.
Cross-posting this:
Hey there, @yangfan85168-5985! Just checking in on your AI Gateway performance issue. Have you found any solutions, or do you still need assistance? Extra details about the errors would really help us troubleshoot!