what kinda bugs? UI related? in my experience the anthropic models have been great at UI related stuff while gpt did better at complex multistep backend type work. opus would always FIND IT then have to go NO WAIT … THAT’S IT … no not quite … forever on complex tasks… haven’t switched to fable yet because of the output token cost. 5.5 has been my go to these days.