Tips for Using the GPT-5 Family of Models

What is GPT-5 and why should I use it?  

GPT-5 is OpenAI’s latest generative AI system, released in August 2025. While there may be significant advantages to using the model, you may also need to adjust your approach given these changes.

Adjusting your Approach  

If you utilize Dartmouth Chat for things like coding or displaying mathematical equations, you will notice that GPT-5 will not by default return results in Markdown. This was intentional decision made “in order to preserve maximum compatibility with developers whose applications may not support Markdown rendering.” cookbook.openai.com.

The way around this is to utilize a system prompt to ask the model to format code and formulas as it has in the past. The example system prompt provided by Open AI reads as follows:

- Use Markdown **only where semantically correct** (e.g., `inline code`, ```code fences```, lists, tables).
- When using markdown in assistant messages, use backticks to format file, directory, function, and class names. Use \( and \) for inline math, \[ and \] for block math.

These instructions can be entered under Chat Controls > System Prompt from within Dartmouth Chat.

Further, you may need to periodically “remind” the model of these instructions or it could default back to responding without Markdown. As stated at the same link above: “Occasionally, adherence to Markdown instructions specified in the system prompt can degrade over the course of a long conversation. In the event that you experience this, we’ve seen consistent adherence from appending a Markdown instruction every 3-5 user messages.”

Users of Dartmouth Chat (and the GPT-5 model API in general) will need to balance the strengths and improvements offered by the new model with the need to (periodically) add system prompts to ensure expected formatting.