Model router

Model router: A component routing each request to a model according to task, cost, capability, latency or operating rules. It can combine small and frontier models without always using the most expensive. Request classification and fallbacks must be evaluated end to end.

In italiano: Model RouterLegge IV · Loss, gradiente e ottimizzatori

It can combine small and frontier models without always using the most expensive. Request classification and fallbacks must be evaluated end to end.

Fonti