Dockerfile

FROM vllm/vllm-openai:latest

WORKDIR /app

EXPOSE 7860

CMD ["vllm", "serve", \
     "Qwen/Qwen3-0.6B", \
     "--port", "7860", \
     "--dtype", "float32", \
     "--enable-reasoning", \
     "--reasoning-parser", "deepseek_r1"]