· Updated
Anthropic redesigned the exam three times. The current version uses Zachtronics-inspired puzzles with constrained instruction sets - problems sufficiently novel that models can't draw on training data. The takeaway: domain-specific problems fall to models with extensive training coverage. Novel, constrained puzzles are where human reasoning still wins.
The original exam is open-sourced. Score below 1,487 cycles and email performance-recruiting@anthropic.com with your code and resume.
New on the Anthropic Engineering Blog: We give prospective performance engineering candidates a notoriously difficult take-home exam. It worked well—until Opus 4.5 beat it. Here's how we designed (and redesigned) it: https://t.co/3RZVyhpVij
More like this
AnthropicApr 30
bubble boiApr 8
ClaudeMay 1
ClaudeDevsApr 28