{"model":"deepseek-chat","stream":true,"stream_options":{"include_usage":true},"temperature":0.2,"max_tokens":2200,"messages":[{"role":"system","content":"You are a rigorous technical diagnostic assistant."},{"role":"user","content":"Write a detailed technical analysis in Simplified Chinese about how to diagnose why LLM streaming output can be slower on a server than on a local machine even when both use the same public egress IP. Write multiple paragraphs and try to use the full output limit. Do not use tables."}]}