Commit

What model do you use the most? I'm 95% on DeepSeek V4.1 Flash, cost effective and very fast.

so @paulinevos was suggesting that model yesterday, need to try it out;