An AI large language model (LLM) that shows the intermediate steps it takes to generate the answer. Reasoning models are used to handle very complex problems that require more than just recognizing patterns, which is the primary approach in most AI systems. Advanced math and software coding are the types of problems reasoning models are used for. OpenAI's o1 and o3 are AI reasoning models.
Why Not Both?
Known as a "hybrid AI reasoning model," in 2025, Anthropic's Claude 3.7 Sonnet is a single model that lets users choose between a quick answer or step-by-step reasoning. See
large language model.