Reasoning Model
Also known as · reasoning models · thinking model
A model trained to deliberate at length before answering hard problems.
A reasoning model is an LLM specifically trained to spend extra computation 'thinking' before it answers — generating a long internal chain of reasoning, exploring approaches, and checking itself, rather than responding immediately. This makes it markedly better at math, coding, and multi-step logic.
The trade-off is time and cost: producing all that intermediate reasoning uses more tokens and takes longer. For simple questions it's overkill; for genuinely hard problems it can be the difference between a right and wrong answer.
Reasoning models essentially bake chain-of-thought into the model itself, and many offer a way to dial the amount of 'thinking effort' up or down depending on the task.