Companion videos

Prompt engineering for reasoning models (o3, R1, Claude extended thinking) — companion videos

The article's core point is that reasoning models flipped the prompting playbook: you stop coaching the chain of thought and start writing a tight problem spec. These videos give you the why and the where-it-came-from — a clear explainer of what makes reasoning models a different kind of model, plus a short team conversation where the people who built o1 describe the "aha" moment that made standard prompting tricks redundant.

Primary pick

27:51

o1 - What is Going On? Why o1 is a 3rd Paradigm of Model + 10 Things You Might Not Know

AI Explained

The clearest non-hype explanation of why o1 (and by extension o3, R1, Claude extended thinking) layers reinforcement learning on top of next-token prediction to reward correct multi-step answers, and what that means for how you talk to it. Watch this and the article's "stop saying think step by step, start writing the spec" advice stops feeling arbitrary.

What you should get from this: Understand why reasoning-model prompts should specify the problem, constraints and success criteria instead of asking for visible chain-of-thought.

Watch or know first: Basic prompting experience and familiarity with model names such as o-series, Claude extended thinking and R1.

AI Expert note: Treat the o1 terminology as historical context. The useful lesson is the prompting shift toward tight task specifications; current model names, product limits and best practices will keep changing.

Open video page

Also worth watching

03:16

Building OpenAI o1

OpenAI

Three minutes with the o1 team describing the moment the model started questioning its own reasoning during RL training. Useful as the primary source for the article's claim that the chain-of-thought is now happening inside the model, not in your prompt.

What you should get from this: See the original product/research framing that made reasoning models different from ordinary chat models.

Watch or know first: Watch the primary pick first or read the companion article's reasoning-model section.

AI Expert note: Keep this as historical source material, not as current implementation guidance. It explains why the category emerged, but production choices should be based on the current model lineup and API docs.

Open video page