i-serve

What a latency promise costs in seats and dollars on a rented GPU — a slice of an LLM inference operator's platform, with every number traceable to the run that produced it.

GitHub

Official source

Start locally with the CLI guide