Anthropic has a 2-hour engineering take-home test. It says its new Claude 4.5 model outscored every human who took it.

Anthropic is solidifying its dominance in AI coding, with a new release that performed better on its engineering test than ...