Model cascade

Route a query through a chain of increasingly capable and costly models, stopping at the first whose output passes a confidence or verification check, so most requests are served cheaply while quality is preserved on the hard tail.

grounded in: router_stats.py / the system's tiered-router capability, plus the trend theme 'Model migration for cost/speed: teams porting production agents to newer models chasing measurable speed and cost wins' —

Connected concepts

Tiered LLM router, Inference economics, Verification loops, Cost-aware agent eval