← Concept IndexDEFINITION WHY IT MATTERS COMMONLY CONFUSED WITH
Reasoning models
Also called: thinking models, test-time compute
Models trained to spend extra computation 'thinking' before they answer — generating and weighing internal steps — which improves hard, multi-step tasks at the cost of more time and money per answer.
They shift the trade-off: better on genuinely hard problems, overkill and slow for simple ones. Knowing when the harder mode earns its cost is a practical skill, not just a technical detail.
A different kind of truth-teller. More deliberation reduces some errors but does not make a model factual or immune to hallucination.