Few questions on perf test numbers being published in litellm docs

[rachitchauhan43]

There are bunch of places in litellm docs where perf numbers have been published:

• This one talks about load testing with 500,000 HTTP connections
• This one talks about 100 requests per second
• This benchmarks for LiteLLM Gateway (Proxy Server) tested against a fake OpenAI endpoint.

Few questions:

1. What are the 1 container's resources (mem and cpu) when these load tests were conducted ? For the last link i see there is a mention of locust and infra setup here.
2. Is there a chart for overhead latency added due to litellm ?