[ ABORT TO HUD ]
SEQ. 1

Chain of Thought & The o1 Architecture

🧠 o1 and Reasoning Models 20 min 500 BASE XP

A New Paradigm in AI

The o1 and o3 families of models represent a fundamental shift from standard LLMs. Instead of generating answers token-by-token immediately, they use reinforcement learning to generate a hidden Chain of Thought (CoT) before producing the final output.

How it works

When you give o1 a complex problem (like a math theorem or a massive refactoring task), it:

  1. Breaks the problem down into smaller steps.
  2. Tries different approaches.
  3. Recognizes its own mistakes and backtracks.
  4. Synthesizes a final, highly accurate answer.

Prompting Reasoning Models

Because these models reason internally, traditional prompt engineering techniques like "Think step by step" or few-shot prompting actually hurt performance.

  • Keep it simple: State the problem directly. Do not tell it *how* to think.
  • Provide edge cases: Give it constraints and edge cases to consider.
  • Use Markdown: Structure the input clearly so the model understands the formatting of the problem.
Developer Note: The reasoning models do not support System prompts in the traditional sense; use the developer message role in the API instead of system.
SYNAPSE VERIFICATION
QUERY 1 // 3
How do reasoning models (like o1) differ from models like GPT-4o?
They are faster and cheaper
They generate a hidden Chain of Thought to reason through problems before answering
They can only generate images
They do not understand code
Watch: 139x Rust Speedup
Chain of Thought & The o1 Architecture | o1 and Reasoning Models — OpenAI Academy