Technique Guide
Chain of Thought Prompting
Chain of Thought (CoT) showed that models answer better when they work through a problem before concluding. Today's reasoning models do that on their own, so the technique has split in two: direct the thinking on frontier models, and trigger it on smaller ones.
What is Chain of Thought Prompting?
Chain of Thought (CoT) prompting encourages an AI model to articulate its reasoning process before giving a final answer. Instead of jumping straight to a conclusion, the model “thinks out loud” through intermediate steps.
There are two classic forms: few-shot CoT, where you provide worked examples that show step-by-step reasoning, and zero-shot CoT, where you simply add “Let's think step by step” to your prompt. Both significantly improve accuracy on models that don't reason on their own.
What changed with reasoning models
GPT-6, Claude Fable 5.1 and Opus 5.5, and Gemini 3.x reason in a hidden phase before answering. You can't see that reasoning, only an optional summary. For these models, give the task, constraints and output format, and control depth with the effort setting instead of adding “think step by step.”
Zero-shot CoT example
Q: If a store has 48 apples and sells 3 bags of 6 apples, how many are left?
Let's think step by step.
When to Use Chain of Thought
- ✓Small, fast or open-weight models without a built-in thinking phase
- ✓Reasoning turned down or off to save cost and latency
- ✓Apps where visible steps are the product, like tutoring
- ✓Multi-step math, logic or planning on any of the above
- ✓Few-shot examples that teach a reasoning pattern for classification
Articles
How to Prompt Reasoning Models
Effort settings, hidden reasoning and what to put in the prompt now that models think on their own.
Chain of Thought Prompting: The Complete Guide
Everything you need to know about CoT prompting — from basic step-by-step reasoning to zero-shot CoT.
What is Prompt Engineering?
Understand the broader skill set that CoT is part of — essential context for beginners.
Zero-Shot vs Few-Shot Prompting
CoT works differently with zero examples vs. a few demonstrations. Here's when to choose each.
Why Your AI Results Are Mediocre
Skipping reasoning steps is one of the top causes of bad AI output. Learn how CoT fixes it.
The Iteration Loop: Refining Your Prompts
How to test and improve CoT prompts systematically.
Related Lessons
Go deeper on CoT and the surrounding techniques with these structured curriculum lessons.
Related Guides
Practice CoT in the Curriculum
The Intermediate track covers CoT, few-shot, system prompts, and more in a structured order.
Start Intermediate Track