Best Practices

1. Enable Response Caching

Cached responses cost $0 and return instantly. Enable for all repeated queries.

{
  "model": "gpt-3.5-turbo",
  "messages": [...],
  "enableCache": true  // Default: true
}

2. Use Cheaper Models When Possible

3. Use Cost-Based Routing

Let AIRouter automatically select the cheapest provider

{
  "routingStrategy": "cost"
}

1. Use Latency-Based Routing

For real-time applications, route to the fastest provider

{
  "routingStrategy": "latency"
}

2. Monitor Rate Limits

Check response headers to avoid hitting rate limits

const remaining = response.headers.get('X-RateLimit-Remaining');
const reset = response.headers.get('X-RateLimit-Reset');

if (remaining < 10) {
  console.warn(`Only ${remaining} requests remaining`);
}

1. Secure API Key Storage

2. Use Appropriate Permissions

Create separate keys with minimal required permissions

• Read-only keys for analytics dashboards
• Full access keys for production apps
• Set expiration dates for temporary keys

3. Rotate Keys Regularly

Implement key rotation every 90 days for production environments

1. Use Fallback Routing for Critical Operations

{
  "routingStrategy": "fallback"  // Automatic failover
}

2. Implement Error Handling

try {
  const response = await chat({...});
} catch (error) {
  if (error.status === 429) {
    // Rate limited - wait and retry
  } else if (error.status === 402) {
    // Insufficient credits
  } else if (error.status === 503) {
    // Service unavailable
  }
}