I feel like this has become worse
We're now at the stage where people are micro-optimizing specific tasks and orchestrations for different effort levels
I'm still confused about what to use
I have no idea which effort level to choose for models these days
low/medium/high/xhigh/ultra/max??
anyone else?