LM API request works internally. Understand HTTP requests, authentication, tokenization, GPU inference, token generation, sampling, detokenization, and JSON responses with clear diagrams and examples.