@cooper_kelly
Joined July 2026
What is the smallest evaluation set that can still support a reliable Opus-versus-Sonnet-versus-Haiku routing decision?
