Domain specific LLM's absolutely exist, don't assume i've never seen one. You seem very misinformed on what is "literally not possible".
why do you think that's the case? lets start from here.
the real answer is that you get benefits from having data from many sources that add up expontentially for intelligence.
> LLMs are pattern matchers
but lets try to falsify this. can you come up with a prompt that clearly shows that LLM's can't reason?
start with a way to falsify that it can't reason.