A service tier that uses faster, cheaper models for simple requests.
It is the AI express checkout lane. Tiny questions get scanned and bagged fast.
You meet it in chats, support bots, and AI agents. It saves money and time, then sends hard problems to a bigger model.
Routing
Routing decides if a request starts in the fast tier or goes to a stronger model.
SLM
An SLM often runs the fast tier for common light tasks.
Fallback model
The fast tier can pass hard problems to a stronger fallback model.
Cost-aware AI
Fast and slow tiers are a common way to control AI cost.