What is a reasoning model?

A reasoning model, also called a thinking model, is an AI model that works through a problem step by step before it gives an answer. It writes out its own working, checks it and corrects it, which is why many AI apps show a “thinking” message and a pause. This makes it slower and more expensive, and better at problems with several dependent steps. Many AI models now offer it as a setting rather than as a separate model.

Why it matters for a business

The working costs time and money. It is billed at the output rate, so a short answer with long thinking behind it can cost several times the answer alone. It pays off where a wrong middle step would ruin everything after it: a calculation with dependencies, a delivery note reconciled against a contract, code that has to run. It buys nothing on drafting an email, and it cannot supply a fact the model never held. More working on a missing fact produces a better-argued wrong answer.

Go deeper

AI vs. machine learning, and where LLMs fit

Related terms

• Large language model (LLM)
• Token pricing
• Inference
• AI hallucination
• RLHF (reinforcement learning from human feedback)

All terms, A to Z