For me, at least it's that the newer Claude models seem optimised for one-shotting things, which is not what I want. As the amount of code per turn increases, I have a harder job keeping up and ensuring that it's doing what I want.

That being said, I had to nope out of a similar thing from GPT 5.6 today, so it appears to be a US frontier lab issue. Claude is particularly bad though, as it produces far too much code even when I tell it not to, unlike GPT (and Kimi) which at least listen to me a little better.

More generally, I want a usable human review experience, and Claude code doesn't deliver that for me.

Quality is hard to measure and I would not say the concern is so much cost but the intersection of cost and time. Often I am jamming on something and I like being somewhat in the loop. So maybe same level of quality, I am using Anthropic modela for both harnesses, but I get to the output quicker and at a drastically lower cost.