← back
The Model Isn't Wrong—You're Just Bad at Prompting
Original: The Model Isn’t Wrong—You’re Just Bad at Prompting
Takeaway
Prompt engineering is still the cheapest lever — but reasoning models invert the old rules, preferring minimal prompts with encouragement to think longer.
Summary
- PromptHub's Dan covers core prompt engineering: chain-of-thought, few-shot, meta-prompting, and how reasoning-model prompting differs.
- Chain-of-thought zero-shot ('think step by step') or few-shot reasoning chains improve performance; Auto-CoT and AutoReason use LLMs to generate the chains automatically.
- Few-shot returns flatten or degrade beyond ~1-2 diverse examples; meta-prompting tools (Anthropic, OpenAI, PromptHub) tailor prompts per provider.
- For reasoning models (o1, R1): minimize examples, give clear task description, encourage more thinking — Microsoft's MedPrompt paper found few-shot hurt o1 performance.
prompt-engineeringchain-of-thoughtreasoning-models
Original description
LLMs aren’t failing—you just need better prompts. This talk covers Chain of Thought, Few-Shot, and Meta Prompting, plus key differences when working with reasoning models. Learn when simple beats complex, when to avoid few-shot prompting, and how to fine-tune your approach with free tools. Stop blaming the model—fix your prompts. RESOURCES Prompt Engineering Substack https://prompthub.substack.com PromptHub Blog https://www.prompthub.us/blog Prompt Engineering with Reasoning Models https://www.prompthub.us/blog/prompt-engineering-with-reasoning-models Chain of Thought Prompting Guide https://www.prompthub.us/blog/chain-of-thought-prompting-guide The Few Shot Prompting Guide https://www.prompthub.us/blog/the-few-shot-prompting-guide PROMPT TEMPLATES AND COLLECTIONS AutoReason https://app.prompthub.us/community/playground/5316 DeepSeek-R1 training template https://app.prompthub.us/community/playground/5788 Reasoning prompts collection https://app.prompthub.us/community/group/361 TOOLS PromptHub Prompt enhancers https://www.prompthub.us/features/enhancers