ChatGPT - Reasoning Models: Choosing and Prompting Best Practices

**The document distinguishes between OpenAI's reasoning models (o-series) and GPT models, highlighting their unique strengths.** **Reasoning models excel at complex problem-solving, strategic planning, and tasks requiring high accuracy, while GPT models are better suited for straightforward, fast, and cost-effective execution.** **The guide suggests using reasoning models for ambiguous tasks, extracting key information from large datasets, understanding complex relationships in documents, multi-step planning, visual reasoning, code review, and model response evaluation.** **It also provides guidance on prompting reasoning models, recommending simple, direct prompts and emphasizing the use of delimiters and specific guidelines.** **The document suggests a combination of both model types for optimal AI workflows, leveraging the strengths of each for different task components.** **It is suggested that developers now use developer messages rather than system messages when using reasoning models.**