# you can start and stop backends by running `docker-compose up -d <service name>`
version: '3.8'
services:
  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    ports:
      - "11434:11434"  # Ollama default
    volumes:
      - ./models:/models
      - ./scripts:/scripts # needed for import script
    environment:
      - OLLAMA_MODELS=/models/.ollama
    restart: unless-stopped

  text-generation-webui:
    image: atinoda/text-generation-webui:default-cpu
    container_name: textgen-webui
    init: true
    environment:
      - EXTRA_LAUNCH_ARGS="--listen --verbose" # Custom launch args (e.g., --model MODEL_NAME)
    ports:
      - "7860:7860"  # Web UI default
      # - "5000:5000" # API Default
      # - "5005:5005" # Streaming API default
    volumes:
      - ./models:/app/user_data/models
    restart: unless-stopped

  # llamacpp server can only run one model at a time; set it below
  llamacpp:
    image: ghcr.io/ggerganov/llama.cpp:server
    container_name: llamacpp-server
    ports:
      - "8000:8000"  # llama.cpp server default
    volumes:
      - ./models:/models
    environment:
      - MODEL_DIR=/models
    restart: unless-stopped
    command: |-
      --port 8000
      --no-webui
      --metrics
      --jinja
      --ctx-size 8192
      --alias "Home-3B-v3"
      --model "/models/Home-3B-v3-fixed.q4_k_m.gguf"

  localai:
    image: localai/localai:latest
    container_name: localai
    ports:
      - "8080:8080"  # LocalAI default
    volumes:
      - ./models:/models
    environment:
      - MODELS_PATH=/models
    restart: unless-stopped