Better? I am not sure. A parent comment [1] was suggesting better LLM performance using completion than using chat. UX wise it is probably worse except for power users.

[1] https://news.ycombinator.com/item?id=46113298