We use multiple other LLM API providers and I’ve never seen anything like this before. I actually cannot believe that even though we’re under our rate limit, we’re getting back success rates of single-digit percentage points. And just because we’ve sent 532 requests when the rate limit is a thousand, we’re basically getting back no results. 
We’re Paid tier 1, I’d rather be Paid tier 2, but I guess I have to get enough throughput to justify that, but I’m not sure how to get the throughput required if the actual API does the results to consume enough tokens. What is going on?
