5 prompts to try first in ChatGPT o-3 mini-high

ChatGPT app on iPhone
(Image credit: Shutterstock)

Sometimes you just need an extra boost, and in the case of ChatGPT, this is where the o-3 mini-high model comes in. The o3 mini-high model takes more time to analyze problems, offering higher intelligence and more thorough responses — particularly for STEM-related topics.

Whether that’s crafting blocks of code or solving a logic puzzle, your best bet at achieving high-quality results is by using this model that’s built for science, math, and coding. While ChatGPT-4o remains perfectly suitable for most everyday tasks, it’s useful to know there are more powerful options out there.

If you feel like you’re not getting the precision and accuracy you’re striving for when inputting complex prompts, o3 mini-high is a game-changer. Here are five prompts I tested myself which you can either use as is, or tweak according to your preferences.

Switch on ChatGPT o-3 mini-high

(Image: © Tom's Guide )

After you log in to your account, start by switching on ChatGPT o-3 mini-high. You can do so by clicking on whatever model you’re currently running, visible at the top left of your screen.

From the drop-down menu select ChatGPT o-3 mini-high. You’ll need a paid Plus account for this feature.

1. Design a website

(Image: © Tom's Guide )

Are you ready to create your own website? With a simple prompt, you can ask ChatGPT o-3 mini-high to use its coding skills to create a website using HTML and CSS, two languages used to structure the content and style of a webpage.

For example you can use this prompt: Create the HTML and CSS code for a sleek portfolio website to showcase the photos I take as a wedding photographer.

The output...

(Image: © Tom's Guide )

2. Finding flaws in math puzzles

(Image: © Tom's Guide )

Asking ChatGPT-4o to find a 5-digit number that, when multiplied by 5, yields the original number with its digits reversed yields a reply: 17985. Which is, unfortunately, incorrect as multiplying that figure by 5 results in 89,925.

On the other hand, after 2 minutes of intense reasoning, ChatGPT o-3 mini-high correctly concluded that there is no 5‑digit number that satisfies the condition.

The puzzle does work when it’s a four-digit number multiplied by 4. With the original number needing to be 2178.

You can try it out yourself with the prompt: Find a 5-digit number that, when multiplied by 5, yields the original number with its digits reversed.

3. Create the code for your own game

(Image: © Tom's Guide )

Who said coding can’t be fun? What better way to make use of this model’s coding capabilities than asking it to create an original game for you?

To create your own game using AI try type:  Design a novel, interactive game using Python and Pygame.

4. Enhancing ChatGPT

(Image: © Tom's Guide)

Since Plus users of ChatGPT have limited usage of o3‑mini‑high (unlike the uncapped usage for Pro users), it may be worthwhile to refine prompts with less powerful models before asking your final queries to o3-mini.

We first asked ChatGPT-4 to create a business model for a bookstore. With this model in hand, we then turned on o3‑mini‑high and asked for a detailed SWOT analysis.

The prompt you can use is: You will be provided with a business model for a bookstore. Your task is to critically analyze it and provide a detailed SWOT analysis.

The output...

(Image: © Tom's Guide )

5. Studying for a test

(Image: © Tom's Guide )

The standard ChatGPT-4o model remains great to handle most tasks, including determining the correct answer to questions testing your logic. However, if you want to be sure you’re studying correctly, switching to o-3 mini-high may be a wise decision.

For our prompt we used a sample LSAT question from a database that was used to evaluate the logical reasoning ability of GPT-4. GPT-4o gave us the incorrect answer but ChatGPT o-3 mini-high solved the puzzle correctly on the first try.

The prompt is as follows: The organizer of a book club must select at least five and at most six works from a group of nine works. These works include: Three French novels, three Russian novels, two French plays, and one Russian play. The selection must follow these rules: No more than four French works can be selected. At least three but no more than four novels must be selected. At least as many French novels as Russian novels must be selected. If both French plays are selected, the Russian play cannot be selected.

You can then ask ChatGPT o-3 mini-high to find out which of the following statements could be true about the organizer's selection of works.

  1. No Russian novels are selected
  2. Exactly one French novel is selected
  3. All three plays are selected
  4. All three Russian novels are selected
  5. All five French works are selected

The output...

(Image: © Tom's Guide )

Spoiler alert: The correct answer is 1.


Now that you've learned how to try ChatGPT Search, why not check out our other AI articles? ChatGPT can help you craft the perfect cover letter to make you stand out and even write a better resume. And if you're new to ChatGPT's reasoning model o3-mini, here are 5 prompts to try first.

Christoph Schwaiger

Christoph Schwaiger is a journalist who mainly covers technology, science, and current affairs. His stories have appeared in Tom's Guide, New Scientist, Live Science, and other established publications. Always up for joining a good discussion, Christoph enjoys speaking at events or to other journalists and has appeared on LBC and Times Radio among other outlets. He believes in giving back to the community and has served on different consultative councils. He was also a National President for Junior Chamber International (JCI), a global organization founded in the USA. You can follow him on Twitter @cschwaigermt.

You must confirm your public display name before commenting

Please logout and then login again, you will then be prompted to enter your display name.