What 50,000 Runs of a 5-Line Eval Taught Us

Findings from running a 5-line evaluation 50,000 times on how AI coding models calibrate effort, token cost, and tool use.