server:
  listen: "0.0.0.0:3000"
  request_timeout: 300s
  max_request_body_mb: 10
  session_secret: "${SESSION_SECRET}"
  default_admin:
    username: "${ADMIN_USERNAME}"
    password: "${ADMIN_PASSWORD}"

tokens:
  - name: "open-webui"
    key: "${OPENWEBUI_API_KEY}"
    rate_limit_rpm: 0          # unlimited
    daily_budget_usd: 5.0
  - name: "rayandrew"
    key: "${PERSONAL_API_KEY}"
    rate_limit_rpm: 0          # unlimited
    daily_budget_usd: 10.0

pricing_lookup:
  # url: "https://raw.githubusercontent.com/pydantic/genai-prices/main/prices/data_slim.json"  # default
  refresh_interval: 6h

database:
  path: "/data/gateway.db"
  retention_days: 90

cache:
  enabled: true
  address: "valkey:6379"
  ttl: 3600

providers:
  - name: deepinfra
    base_url: "https://api.deepinfra.com/v1/openai"
    api_key: "${DEEPINFRA_API_KEY}"
    priority: 1
    timeout: 120s
  - name: siliconflow
    base_url: "https://api.siliconflow.com/v1"
    api_key: "${SILICONFLOW_API_KEY}"
    priority: 2
    timeout: 120s
  - name: openrouter
    base_url: "https://openrouter.ai/api/v1"
    api_key: "${OPENROUTER_API_KEY}"
    priority: 3
    timeout: 120s
  - name: groq
    base_url: "https://api.groq.com/openai/v1"
    api_key: "${GROQ_API_KEY}"
    priority: 1
    timeout: 120s
  - name: cerebras
    base_url: "https://api.cerebras.ai/v1"
    api_key: "${CEREBRAS_API_KEY}"
    priority: 1
    timeout: 120s

models:
  - name: "deepseek-v3.2"
    routes:
      - provider: deepinfra
        model: "deepseek-ai/DeepSeek-V3.2"
        pricing: { input: 0.26, output: 0.38 }
      - provider: siliconflow
        model: "deepseek-ai/DeepSeek-V3.2"
        pricing: { input: 0.27, output: 0.42 }
      - provider: openrouter
        model: "deepseek/deepseek-chat-v3-0324"
        pricing: { input: 0.30, output: 0.88 }

  - name: "llama-3.3-70b"
    routes:
      - provider: groq
        model: "llama-3.3-70b-versatile"
        pricing: { input: 0, output: 0 }
      - provider: deepinfra
        model: "meta-llama/Llama-3.3-70B-Instruct"
        pricing: { input: 0.23, output: 0.40 }

  - name: "llama-3.1-8b"
    routes:
      - provider: groq
        model: "llama-3.1-8b-instant"
        pricing: { input: 0, output: 0 }
      - provider: cerebras
        model: "llama-3.1-8b"
        pricing: { input: 0, output: 0 }
      - provider: deepinfra
        model: "meta-llama/Meta-Llama-3.1-8B-Instruct"
        pricing: { input: 0.03, output: 0.05 }

  - name: "qwen-2.5-72b"
    routes:
      - provider: groq
        model: "qwen-2.5-72b"
        pricing: { input: 0, output: 0 }
      - provider: deepinfra
        model: "Qwen/Qwen2.5-72B-Instruct"
        pricing: { input: 0.23, output: 0.40 }

  - name: "qwen-2.5-coder-32b"
    routes:
      - provider: groq
        model: "qwen-2.5-coder-32b"
        pricing: { input: 0, output: 0 }
      - provider: deepinfra
        model: "Qwen/Qwen2.5-Coder-32B-Instruct"
        pricing: { input: 0.07, output: 0.16 }

  - name: "gemma-2-9b"
    routes:
      - provider: groq
        model: "gemma2-9b-it"
        pricing: { input: 0, output: 0 }

  - name: "deepseek-r1"
    routes:
      - provider: deepinfra
        model: "deepseek-ai/DeepSeek-R1"
        pricing: { input: 0.40, output: 1.60 }
      - provider: openrouter
        model: "deepseek/deepseek-r1"
        pricing: { input: 0.55, output: 2.19 }

  - name: "deepseek-r1-distill-llama-70b"
    routes:
      - provider: groq
        model: "deepseek-r1-distill-llama-70b"
        pricing: { input: 0, output: 0 }
      - provider: deepinfra
        model: "deepseek-ai/DeepSeek-R1-Distill-Llama-70B"
        pricing: { input: 0.23, output: 0.69 }

  - name: "deepseek-r1-distill-qwen-32b"
    routes:
      - provider: deepinfra
        model: "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"
        pricing: { input: 0.07, output: 0.16 }