My point is more it's the wrong tool for the job. Something designed for calculations like wolfram alpha and their API is probably better suited:

https://www.wolframalpha.com/input?i=%28x%5E3%29%2B%28y%5E3%29%2B%28z%5E3%29+%3D+1024

BUT I did ask ChatGPT (so 3.5) to write an app to do it in a couple languages and it gave me a working app first try on each. It's not a very good app as I could optimize it a lot more, but it works. GPT-4 gave a slightly better app in each instance.

Genuine question: Why are you trying to use a language model to do something that you could write a basic app to calculate?

Like I would have asked it to write an app in JavaScript, Java, C#, etc. Some popular language to calculate four perfect cubes to represent a number. That'd probably get me 90% of the way there then I'm just fixing a couple bugs. That seems like the more intuitive use case to me but I'm also a dev by trade.

Just adding on that I used ChatGPT and adding any sort of "write it in the style of X" to the end fools it. Tell it to do some run on sentences, etc. same thing.