llm-bench 0.2.2

Send the same prompts to OpenAI, Anthropic or any OpenAI-compatible server and compare p50/p99 latency, tokens per second and cost in one table