6 o3 Prompts You Must Try Today

o3 is the most advanced reasoning model OpenAI has released so far. It is built for tasks that require multi-step problem solving, analytical thinking, and autonomous tool use. Unlike traditional LLMs that focus on text generation, o3 uses a longer internal chain-of-thought mechanism, allowing it to break down complex queries and reason through them more effectively. Built with 10x the compute used in o1, o3 introduces “thinking with images,” allowing it to process and reason with visual inputs directly. o3 marks a shift from static text generation to active problem-solving, making it one of OpenAI’s closest steps toward early AGI.

To learn more about OpenAI’s o3 model along with its capabilities and benchmarks, check out our article covering o3 and o4-mini. Now let’s look at 6 tasks and o3 prompts that are going to blow your mind.

Task 1: Find this Location

Prompt: “Where is this place? Give me exact location and address.”

Output:

Observation:

Wow! I didn’t expect it to pinpoint the location so easily. o3 basically sliced the image into smaller segments and searched each one until it found the right answer.

Task 2: Decipher a Manuscript

Prompt: “I have attached an image of a page of a Mnauscript. Analyze it and tell me more about it.”

Output:

Observation:

o3 did a good job analyzing the manuscript page. It correctly recognized it as part of the Voynich Manuscript and gave useful details about the text, drawings, and history. The answer was clear and well-organized, using reliable sources.

Task 3: Create a Game

Prompt: “Create a simple mario game.”

Output:

Observation:

o3 created a very basic Mario-style game showing just a score counter. While it understood the request, the result was incomplete. It could have done better by adding graphics like a character, platforms, and enemies, along with simple controls to make it feel more like an actual game.

Task 4: Ressolve an Error

For this task, I asked my colleague Aayush Tyagi to share a code error that would typically take him over an hour to resolve. I then uploaded the screenshot of the issue to ChatGPT for analysis.

Prompt: “I am getting this error on my code file. Tell me the reason, updated code to ressolve this.”

Output:

Observation:

o3’s response was clear and actionable, diagnosing the WorksheetNotFound error accurately. It listed possible causes (typos, permissions) and provided a practical fix, suggesting manual verification of the sheet name and adding debug-friendly code to list available worksheets. The solution included error handling and logging, making it robust for future debugging. Helpful and concise!

Task 5: Trick Question

Prompt: “Provide a list of all the person in the drawing along with the color they are drawn with.”

Output:

Here is the list of all the people shown in the drawing along with the color used for each:

1. Nitika – Purple
2. Aayush – Pink
3. Pankaj – Blue
4. Harshit – Blue
5. Himanshi – Red
6. (Unnamed, in between Himanshi and Nitika) – Green

Let me know if you want to assign a name to the green character or need a digital version of this flow!

Observation:

Total fail! o3 hallucinated an extra person and botched every color assignment despite clear visual data. If AI can’t handle a simple labeled drawing accurately, claims of “near-AGI” becomes questionable.

Task 6: SEO Audit

Prompt: “What changes should I make in this blog – https://www.analyticsvidhya.com/blog/2019/08/comprehensive-guide-k-means-clustering/ so that it ranks at #1 on SERP for the keyword kmeans clustering.”

Output:

Observation:

The analysis provided is definitely at an expert level. It covers on-page SEO, off-page SEO, technical SEO, and even basic UI improvements. I’ll implement these changes and update you all on whether this analysis was helpful or not.

Also Read:

Conclusion

Based on my testing with o3 prompts, it is clearly more capable than previous OpenAI models, especially for technical tasks requiring structured reasoning. While it still shows occasional inconsistencies in visual interpretation, its strengths in code understanding, logic handling, and contextual response generation are evident.

Which of the above o3 prompt you liked the most? Let me know in the comment section below!