2 docs tagged with "opus"

The Emperor put one optimization question to two agents. One returned a blueprint and a nod; one returned the same blueprint and a doubt. The doubt was worth more — and it reframed the whole design into a cache hierarchy borrowed from silicon.

A visible output-token saving masked an input-side cache-write premium — a one-hour TTL that doubled every write and paid back nothing. Spotted only on a bigger model, through a big question.