5.5 high is great...
I've been running ChatGPT Plus 4 accounts, but I mainly used 5.5 low because I was running out of tokens.
Low definitely seems to have a tendency to do only what I ask.
With high, it proceeds while looking around the edges a bit. It feels like there are fewer gaps.
It consumes more tokens and is slower, though...
There are fewer debugging needs, and it debugs well in one go.
It has a tendency to fix things on its own even without deliberately providing logs.
I'm curious to see what 5.6 will be like when it comes out soon hehe
