version: "3.9"
services:
  localai:
    build: .
    container_name: localai
    ports:
      - "8080:8080"
    environment:
      - MODELS_PATH=/models
      - THREADS=2          # ajustez selon vos cores (léger par défaut)
      - CONTEXT_SIZE=1024  # contexte réduit pour tests
      - DEBUG=false
    command: ["--models-path", "/models", "--address", "0.0.0.0:8080"]
    healthcheck:
      test: ["CMD", "curl", "-fsS", "http://localhost:8080/v1/models"]
      interval: 10s
      timeout: 5s
      retries: 10