AI/ML
LLM Cascades: Escalate Only When the Cheap Model Fails
Cheap model first. The check decides the rest.
