Your First AI Gateway

Why Use a Gateway

Instead of calling OpenAI directly, your application calls SBproxy. SBproxy adds auth, rate limiting, model fallbacks, spend tracking, and guardrails, all in config, no code changes required.

Install SBproxy

brew tap soapbucket/tap && brew install sbproxy

Not on macOS? See the install page for Linux, Docker, and binary downloads.

Create Your Config

Create a file called sb.yml:

proxy:
  http_bind_port: 8080

origins:
  "ai.example.com":
    action:
      type: ai_proxy
      providers:
        - name: openai
          api_key: ${OPENAI_API_KEY}
          models: [gpt-4o, gpt-4o-mini]
      default_model: gpt-4o-mini
      routing:
        strategy: fallback_chain

Start the Gateway

export OPENAI_API_KEY=sk-...
sbproxy serve -f sb.yml

Send a Request

curl -H "Host: ai.example.com" http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-4o-mini", "messages": [{"role": "user", "content": "Hello"}]}'

Your existing application code does not change. Only the base URL changes from https://api.openai.com to your SBproxy address.

Add a Second Provider

Add Anthropic as a fallback. If OpenAI is down, requests automatically route to Claude:

proxy:
  http_bind_port: 8080

origins:
  "ai.example.com":
    action:
      type: ai_proxy
      providers:
        - name: openai
          api_key: ${OPENAI_API_KEY}
          models: [gpt-4o, gpt-4o-mini]
        - name: anthropic
          api_key: ${ANTHROPIC_API_KEY}
          models: [claude-sonnet-4-20250514]
      default_model: gpt-4o-mini
      routing:
        strategy: fallback_chain

Add API Key Auth for Users

Give each team or application its own API key:

origins:
  "ai.example.com":
    action:
      type: ai_proxy
      providers:
        - name: openai
          api_key: ${OPENAI_API_KEY}
          models: [gpt-4o, gpt-4o-mini]
      default_model: gpt-4o-mini
      routing:
        strategy: fallback_chain
    authentication:
      type: api_key
      api_keys:
        - name: frontend-team
          key: sk-frontend-abc123
        - name: backend-team
          key: sk-backend-def456

Add Per-Key Rate Limits

Prevent any single team from exhausting your provider quota:

    policies:
      - type: rate_limiting
        requests_per_minute: 60
        key: 'request.headers["x-api-key"]'

Next Steps

• Model Routing - weighted, cost-optimized, and latency-based routing strategies
• Guardrails - content policies, PII redaction, and token budgets
• Providers - full list of 200+ models across 72 native providers