Our API load has not changed in the past few weeks, but starting last night we've been receiving hundreds of "too many request" errors. We are well below published rate limits so just trying to determine what next steps we can take are.
#Was there a change to rate limits made last night?
reef rover
I am this error Faraday::TimeoutError (Net::ReadTimeout with #TCPSocket:(closed)):
which seems like similar behavior. Just started on longer requests.
plucky raven
it does seem like responses have been slower today as well. That said you may want to bump your request timeout on API calls to openai to avoid that error