LLM Council: Cost

Purpose

Cut the cost and latency of a multi-model pipeline that already exists — a council, ensemble, debate, or plain "we call N models per request" setup — without giving up the quality it was built for. The lever is almost never "pick a cheaper model everywhere"; it's making the expensive path fire only when it's actually earning its keep.

When to use / when NOT to use

Use this skill when the pipeline is already running and the complaint is cost, latency, or token spend — including when the user never says "council" ("we call 4 models per request and the bill exploded", "cut multi-agent orchestration cost").

Not for: