A model trained to show explicit step-by-step reasoning and problem-solving logic before producing final answers, rather than jumping directly to conclusions.