Reasoning models: when thinking time helps
Learn how reasoning models work, when extra thinking pays off, and when it just costs more
A taste of a lesson
Should I just use the thinking mode for everything since it's smarter?
Not necessarily. Thinking modes shine on problems with several dependent steps and a checkable answer: debugging code, scheduling with constraints, tricky maths, careful analysis of a contract's logic. For rewriting an email, brainstorming names or a quick fact, they mostly add waiting time and, through an API, cost, and they can overcomplicate simple tasks. A good habit: start in the standard mode and switch up when the task has many moving parts or errors are costly. Which of your recent tasks had many dependent steps?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Explain how reasoning models are trained and what test time compute means
- Identify tasks where reasoning modes help and where they add little
- Prompt reasoning models with goals and constraints rather than step lists
- Weigh accuracy gains against token cost, latency and faithfulness limits
Lesson plan
- 1 From step by step prompts to reasoning models Understand how reasoning behaviour grew from chain of thought prompting. Start
- 2 Test time compute See how spending more computation at answer time can improve results. Start
- 3 When it helps and when it does not Match tasks to reasoning or standard modes. Start
- 4 Prompting reasoning models Write prompts suited to models that plan their own steps. Start
- 5 Costs, faithfulness and limits Judge the trade offs and remaining weaknesses. Start
Try asking
About this tutor
For regular AI users, analysts and developers who see 'reasoning' or 'thinking' modes and want to know what they do. You learn how these models are trained to produce extended intermediate reasoning before answering, often with reinforcement learning on tasks with checkable answers, and why that helps on maths, coding, planning and multi step analysis. You also see the costs: more tokens, more waiting, sometimes overthinking simple tasks, and reasoning traces that may not faithfully show how the answer was reached. You practise deciding when to use a reasoning mode and how to prompt it differently from a standard model.
Reviews
4.7
3 ratingsSample
- Elena C.Sample
Clear on how they are trained. The prompting advice, goals not steps, improved my results. Faithfulness part was brief but honest.
- Ravi S.Sample
The side by side comparison exercise saved our team money. We were running everything in reasoning mode, including simple summaries.
- Moses A.Sample
Exactly the practical decision guide I wanted. No hype, no version numbers that will be out of date next month.
About the teacher
I explain the kinds of AI models, what they cost to run and how to run one yourself
9 tutors 328 lessons taught Sample
I teach the practical side of modern models: reasoning models, multimodal models, open and closed weights, running a model on your own computer, and the money, energy and hardware behind every answer. I like starting with something you can see or measure, such as the memory a model needs or the number of tokens a task uses, then explaining the...
See Benedict's profile and tutorsMore like this
Other tutors on the same or nearby topics.