You don't need any reproduction. Assuming researchers are using LLMs, you should just see the number of jumps increase as models get better.

I boost it so someone else will see it and prove it wrong.

LLMs most definitely are limited. What’s your position, that LLMs have no limits? That’s obviously wrong, and the fact you hold such an unreasonable position may be why you react so strongly to pieces like this one.