Reasoning models are AIs that "think out loud" before giving you an answer. Instead of blurting out the first thing that comes to mind, they work through the problem step by step, like showing their work on a math test. This makes them much better at hard problems like coding, logic puzzles, and complex analysis — but it also means they take longer and use more resources to respond.