Gpt-3.5 concurrent requests limit

[vb]

Take a look at the topic linked below.
It may be a good explanation as to why a increasing number of concurrent requests triggers the rate limit warning.

Rate Limit Advice - update

OpenAI FAQ - Rate Limit Advice - Update Rate limits can be quantized, meaning they are enforced over shorter periods of time (e.g. 60,000 requests/minute may be enforced as 1,000 requests/second). Sending short bursts of requests or contexts (prompts+max_tokens) that are too long can lead to rate limit errors, even when you are technically below the rate limit per minute.