Skip to main content

Controlled Routing

Routing is downstream of identity, policy, approval, budget, and audit enforcement.

Syndicate Gate can route among configured providers, but routing is not the control boundary. A request is routed only after the enforcement chain allows execution and records the decision.

Routing Strategies

  • Lowest Latency: Routes to the provider with the fastest historical Time-To-First-Token (TTFT).
  • Lowest Cost: Routes to the cheapest provider capable of serving the model class.
  • Round-Robin: Distributes requests evenly across all configured providers.
  • Weighted Shuffle: Distributes requests based on user-defined percentage weights.

Define A Route After Enforcement

You can define route preferences in the dashboard. Syndicate Gate still evaluates identity, policy, approval, budget, and audit controls before it invokes any target provider.

// Fallback array example
const response = await client.chat.completions.create({
  model: 'anthropic/claude-3-5-sonnet',
  messages: [{ role: 'user', content: 'Process this data...' }],
  // Handled transparently by the Syndicate Gate SDK or custom headers
  fallbacks: ['azure/claude-3-5-sonnet', 'aws-bedrock/claude-3-5-sonnet']
});