Hey!
I was wondering on whether the model usage of phoenix.new is going to be optimized a bit? Right now it seems very very usage hungry. I signed up today and have already had to refill $20 twice which seems a bit too much for a couple of features I tried while on work downtime (not even fulltime using it). Is this expected? Is this a bug? Maybe be able to select different model strengths for different tasks? Or auto select like Cursor/claude code does? As it is it seems to consume credits even faster than api-call based claude code. (about $15-20 every couple of days of heavy usage).
As it is, for a solo developer it’s a bit too much. I love the product and it works super well but I can’t pay this amount of money, so it would be very nice to know what optimizations you have on the roadmap if any.