Glossary · 1 · Foundations: how AI works
Reasoning model
Also known as: Thinking model, Extended thinking
A reasoning model is a large language model trained to work through a problem in intermediate steps before giving its final answer, spending more computation at inference time to improve results on complex tasks such as math, code and multi-step analysis.
- Intermediate
- Technical writers
- Technical project managers
- Developers
In one sentence
Reasoning models explained: LLMs that think in steps before answering — when they are worth the extra time and cost.
Example
Asked whether a product change requires updating six related safety topics, a reasoning model checks each dependency step by step instead of answering at once.
Why it matters on your learning path
- Technical writers: Useful for consistency checks across large content sets and for complex rewrites; overkill for simple edits.
- Technical project managers: Reasoning takes longer and costs more tokens. Match the model to the task.
- Developers: Many APIs expose a reasoning or “thinking” budget you can tune per request.
Reasoning model vs. chain-of-thought prompting
Chain-of-thought prompting asks any model to show its steps. A reasoning model has been trained to do this on its own. Several brands offer both fast and reasoning modes.