Throwing a tiny little LLM at it helped me assess that it was far too slow for me to reasonably use at the scale I needed. So it didn’t really matter how accurate the prompt was. I was more just pointing out that I didn’t know if would be too slow without trying it. I maybe could have done some simple math in retrospect, but trying it out was easy enough