services:
  llama-cpp:
    image: ghcr.io/ggml-org/llama.cpp:server
    container_name: llama-cpp-qwen
    restart: unless-stopped
    ports:
      - "8000:8000"
    volumes:
      - ./models:/models
    command:
      - -m
      - /models/qwen2.5-0.5b-instruct-q4_k_m.gguf
      - --host
      - 0.0.0.0
      - --port
      - "8000"
      - --ctx-size
      - "4096"
      - --threads
      - "4"