The paper "CARAT: Do Materials LLMs Reason or Recite?" raises the question of whether large language models for materials science are reasoning or reciting.