A reasoning model is an AI model optimized for multi-step problem solving, planning, verification, or deliberative computation.
The term describes behavior and training emphasis, not a guarantee that private internal reasoning is exposed or that outputs are always correct.