Please, anybody may help us in avoiding bad ChatGPT API performances dependong on servers we used over time?
Which are the most critical server parameters we need to check since on some servers ChatGPT API does not even respond randomly.
Thanks.
Nicola - Milano, Italy
Using the OpenAI API uses almost none of your own server’s resources. You send some text, and you get some text back. It can literally run on a Commodore 64.
Any significant load would be from managing your client sessions and their backend. That may require more memory and instances.
You don’t really have choice of the server datacenter you connect to. You are routed by Cloudflare geographically.
OpenAI was giving higher tier users preference, slowing the output of tokens of tier 1 API users that had paid less than $50 total to OpenAI in the past. That’s about the only thing that might affect you under your control, where OpenAI made a problem and the solution is to pay them more money.
3 Likes
Agree with @_j
We don’t have any choice or say in the server locations. What do you mean by this?
Thanks a lot for your prompt and useful reply. Yet, in aug23, I had huge problems on my Web Writer application due to random / no responses from ChatGPT API using a technosware.com server. Those problems immeditely disappeared when we started running it on DigitalOcean.com servers. Same good responses on https://www.tvmserver.com/ servers. Now, on new myWindowsHosting.com server… AGAIN random / no responses. It really seems to be a server correlated issue. Any futher and hugely appreciated hint? Thanks oo!! Nicola
Thanks a lot for your prompt reply. As i just replied to @_j , it really seems to be a server correlated issue. Any futher and hugely appreciated hint? Thanks oo!! Nicola
1 Like
You got me on Commodore 64, and it is true. Now reminiscing the great days and nights of my C64 