Can ChatGPT’s o3 Model Build a Usable Engineering Spreadsheet?

Can the o3 model build a clean, usable unit conversion spreadsheet in Excel?

Not generate text. Not just explain formulas. But actually produce a downloadable file, formatted correctly, with working Excel logic built in.

We gave it a fair shot — using a clear, tightly structured prompt.

Prompt setup for the test

The Test: A Simple Unit Conversion Table

The prompt was designed to be as straightforward — and verifiable — as possible.

We asked o3 to:

• Convert 10 common engineering parameters (like pressure, temperature, flow) from English to SI units

• Populate one row of sample input values

• Use standard unit conversion factors (e.g., 1 psi = 6.89476 kPa)

• Include the actual Excel formula used in the table (e.g., =B2*6.89476)

• Return only the Excel file, with no extra text

The structure and expectations were built into the prompt — simple enough that even basic tools should succeed, but strict enough to see where the model might slip.

The Result

ChatGPT o3 followed the prompt fairly well.

• The file included correct headers

• Sample values were inserted as expected

• Each formula was formatted using proper Excel syntax

• All conversions were mathematically accurate

The only cleanup needed? Minor column width adjustments for readability — nothing that affects functionality.

And most importantly: it didn’t hallucinate or try to be clever. It just did the job.

Why This Matters

It’s not about conversions — it’s about how reliably a model can follow structure.

For engineers, analysts, or project teams testing AI tools, this kind of repeatable formatting task is a good benchmark. If a model can’t handle something this simple, it probably can’t be trusted with more complex logic or document automation.

In this case, o3 passed — and did so without overengineering the solution.

📘 Want to test it yourself or learn how to write better AI prompts for real tasks?Check out the course:🔗 AI for Engineers (Udemy)