version: "3.8"
services:
  ai-gateway:
    image: ai-gateway-test
    security_opt:
      - no-new-privileges:true
    read_only: true
    tmpfs:
      - /tmp
      - /root/.cache
    build:
      context: .
      dockerfile: Dockerfile
      target: test-builder
    command: make test
    tty: false
    stdin_open: false
    volumes:
      - ./aichat:/app/aichat
      - ./scripts:/app/scripts
      - ./test:/app/test
      - ./tools:/app/tools
    ports:
      - "8001:8000"
    environment:
      ENV: "local"
      MODELS: >
        {
          "mixtral-8x7b-instruct": {
              "backend": "vllm",
              "upstream_model": "casperhansen/mixtral-instruct-awq",
              "address": "http://localhost:8888/v1",
              "friendly_name": "Mixtral 8x7B",
              "maker": "Mistral AI",
              "max_tokens": 2000,
              "conversation_token_limit": 8000,
              "conversation_token_limit_premium": 8000,
              "free_min_api_version": 2,
              "rate_limit":  30,
              "rate_limit_interval_seconds": 10800,
              "free": true,
              "tool_support": false,
              "image_support": false,
              "audio_support": false,
              "video_support": false,
              "key": "chat-leo-expanded"
          },
          "llama-2-13b-chat": {
              "address": "http://localhost:8080",
              "friendly_name": "Llama 2 13B",
              "maker": "Meta",
              "upstream_model": "TheBloke/Llama-2-13B-chat-GPTQ",
              "max_tokens": 2000,
              "conversation_token_limit": 8000,
              "conversation_token_limit_premium": 8000,
              "rate_limit":  30,
              "rate_limit_interval_seconds": 10800,
              "free": true,
              "tool_support": false,
              "image_support": false,
              "audio_support": false,
              "video_support": false
          }
        }
      MODEL_TRIAGING: >
        {
          "coding": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "file": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "vision": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "language": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "default": {
            "premium": "llama-2-13b-chat",
            "non-premium": "qwen-14b-instruct"
          },
          "long_context": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "content_agent": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          },
          "summary": {
            "premium": "llama-2-13b-chat",
            "non-premium": "llama-2-13b-chat"
          }
        }
      RATE_LIMITING_ENABLED: "0"
      TITLE_MODEL: "llama-2-13b-chat"
      NEAR_API_KEY: "key"
      DEEP_RESEARCH_URL: "http://host.docker.internal:8080"
      DEEP_RESEARCH_ENABLED: "false"