Every new generation of LLMs can write code better than the previous ones. It's really just a matter of time as the models keep improving. They can already generate basic solutions in one shot if simple enough and if enough examples were present during training.

Complexity is the problem. This is what humans, and now LLMs, struggle with when it comes to designing and coding solutions. Managing complexity is why we've attempted to invent paradigms such as OOP, which some claim only makes things worse.