Streaming Responses

Instead of waiting for the complete response, modern LLM APIs can send tokens to the user as they're generated. This reduces perceived latency. What is this delivery method? Flag format: CONGRESS{method_in_snake_case}