B-TIER technique
GPT-5 Prompt Anatomy — OpenAI's six-part role-to-stop-condition template
100k ft — What is this?
GPT-5 Prompt Anatomy — OpenAI's six-part role-to-stop-condition template
50k ft — Why care?
A worked example of OpenAI's GPT-5 prompting structure: Role, Task, Context, Reasoning, Output format and Stop conditions, shown on a 'find three under-the-radar hikes near San Francisco' prompt. The useful bits are the explicit self-vetting step and a stop condition that defines when the answer is done.
10k ft — Key takeaways
• Ask the model to internally verify facts against named sources before answering.
• Define a concrete completion test so the model knows when to stop.
• Pull OpenAI's official GPT-5 prompting guide as a reference doc
• Split prompts into role, task, context, reasoning, output format and stop condition blocks.
5k ft — Tools & techniques
ChatGPT, GPT-5, OpenAI
chain-of-thought, context loading, prompt engineering, role prompting, stop conditions, structured output
1k ft — Deep-dive / try it
Most people are still prompting the wrong way.
OpenAI just shared their own framework for crafting prompts that make GPT-5 feel like magic.
Their key advice? Be crystal clear with your instructions, load the model with context, and give it a specific role to play.
Swipe to see the full breakdown.
➡️ Want to stay ahead in the world of AI? Grab your coffee, relax, and let our newsletter deliver the biggest updates and trends straight to your inbox.
Don’t miss out! click the link in our bio and subscribe for free today!
thinkbase.ai
41w
Thanks