curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json" -d '{
"model": "Qwen/Qwen3-8B",
"messages": [
{"role": "user", "content": "Give me a short introduction to large language models."}
],
"temperature": 0.7,
"top_p": 0.8,
"top_k": 20,
"max_tokens": 8192,
"presence_penalty": 1.5,
"chat_template_kwargs": {"enable_thinking": false}
}'