Some ChatGPT models deliberate before they answer, working a problem through step by step instead of replying straight away. They are markedly better on hard problems, and slower and dearer on everything else. This course covers how they work, when they earn their cost, and how prompting them differs.
What extended thinking means in practice, how reasoning models differ from standard ones, and what that changes.
The tasks where reasoning models genuinely outperform, the tasks where they do not, and how to decide.
How prompting strategy changes for reasoning models, what to include and what to omit, and how to get more consistent results.
How the reasoning tiers differ in cost and speed, and a working policy for choosing between them.