Curious how an LLM request and response cycle works? Follow one prompt from tokenization through inference to streaming, step by step and in plain English.