Smart Routing Policies

Create visual routing rules and automated fallback cascades to ensure 99.99% availability and cost optimization for your AI workloads.

• Portal Page: /projects/[projectId]/routing-policies

graph TD
    Rule["Routing Policy: Production Flagship"]
    Rule --> P1["Primary: Anthropic claude-3-5-sonnet"]
    Rule -->|Fallback on 5xx / Overload| P2["Secondary: OpenAI gpt-4o"]
    Rule -->|Fallback on Timeout| P3["Tertiary: Google gemini-2.5-flash"]

Creating a Routing Policy #

1. Navigate to Routing Policies in your active Project.
2. Click + Create Routing Policy.
3. Select a strategy:
   • Priority Fallback Cascade: Ordered list of models.
   • Lowest Cost: Dynamic routing to the lowest $/token provider.
   • Lowest Latency: Dynamic routing to the fastest TTFT provider.
4. Define health thresholds and cooldown timeouts.
5. Click Deploy Policy.

Next Steps #

• View execution traces: Provider Attempt Accounting
• Review FinOps controls: FinOps & Budget Controls