6 o3 Prompts to Strive Right now

o3 is probably the most superior reasoning mannequin OpenAI has launched to date. It’s constructed for duties that require multi-step downside fixing, analytical pondering, and autonomous software use. Not like conventional LLMs that target textual content era, o3 incorporates an extended inner chain-of-thought mechanism, permitting it to interrupt down complicated queries and purpose by way of them extra successfully. Constructed with 10 instances the compute utilized in o1, o3 introduces “pondering with photographs,” enabling it to course of and purpose about visible inputs immediately. This multimodal integration, mixed with its means to proactively invoke instruments like canvas, browser search, and file evaluation, marks a shift from static era to lively problem-solving, making o3 one of many closest approximations of early AGI capabilities in OpenAI’s mannequin lineup.

To study extra about OpenAI’s o3 mannequin together with its capabilities and benchmarks, try our in-depth article protecting o3 and o4-mini. Now let’s take a look at 6 duties and o3 prompts which might be going to blow your thoughts.

Job 1: Discover this Location

Immediate: “The place is that this place? Give me precise location and deal with.”

Output:

Statement:

Wow! I didn’t count on it to pinpoint the placement so simply. o3 principally sliced the picture into smaller segments and searched every one till it landed on the appropriate reply.

Job 2: Decipher a Manuscript

Immediate: “I’ve connected a picture of a web page of a Mnauscript. Analyze it and inform me extra about it.”

Output:

Statement:

o3 did a superb job analyzing the manuscript web page. It appropriately acknowledged it as a part of the Voynich Manuscript and gave helpful particulars concerning the textual content, drawings, and historical past. The reply was clear and well-organized, utilizing dependable sources. Nonetheless, it might have been clearer about which elements are guesses (just like the plant identification) as a substitute of details.

Job 3: Create a Sport

Immediate: “Create a easy mario sport.”

Output:

Statement:

o3 created a really fundamental Mario-style sport exhibiting only a rating counter. Whereas it understood the request, the consequence was incomplete. It might have completed higher by including graphics like a personality, platforms, and enemies, together with easy controls to make it really feel extra like an precise sport.

Job 4: Ressolve an Error

For this job, I requested my colleague Aayush Tyagi to share a code error that might sometimes take him over an hour to resolve. I then uploaded the screenshot of the difficulty to ChatGPT for evaluation.

Immediate: “I get this error on my code file. Inform me the rationale, up to date code to ressolve this.”

Output:

Statement:

o3’s response was clear and actionable, diagnosing the WorksheetNotFound error precisely. It listed doable causes (typos, permissions) and supplied a sensible repair, suggesting handbook verification of the sheet identify and including debug-friendly code to record accessible worksheets. The answer included error dealing with and logging, making it strong for future debugging. Useful and concise!

Job 5: Trick Query

Immediate: “Present a listing of all of the particular person within the drawing together with the colour they’re drawn with.”

Output:

Right here is the record of all of the folks proven within the drawing together with the colour used for every:

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

Let me know if you wish to assign a reputation to the inexperienced character or want a digital model of this circulate!

Statement:

Whole fail! o3 hallucinated an additional particular person and botched each shade project regardless of clear visible information. If AI can’t deal with a easy labeled drawing precisely, claims of “near-AGI” turns into questionable.

Job 6: web optimization Audit

Immediate: “What modifications ought to I make on this weblog – https://www.analyticsvidhya.com/weblog/2019/08/comprehensive-guide-k-means-clustering/ in order that it ranks at #1 on SERP for the key phrase kmeans clustering.”

Output:

Statement:

The evaluation supplied is certainly at an skilled degree. It covers on-page web optimization, off-page web optimization, technical web optimization, and even fundamental UI enhancements. I’ll implement these modifications and replace you all on whether or not this evaluation was useful or not.

Additionally Learn:

Conclusion

Primarily based on my testing with o3 prompts, it’s clearly extra succesful than earlier OpenAI fashions, particularly for technical duties requiring structured reasoning. Whereas it nonetheless reveals occasional inconsistencies in visible interpretation, its strengths in code understanding, logic dealing with, and contextual response era are evident. For builders and AI practitioners, o3 is a dependable assistant for prototyping, debugging, and deep technical evaluation, and a critical step towards extra autonomous AI techniques.

Which of the above o3 immediate you favored probably the most? Let me know within the remark part under!