when do you switch from a fast cheap model to a heavy reasoning model mid-session?
by•
for standard UI components and basic CRUD endpoints, lighter fast models feel instant and keep my momentum going without waiting thirty seconds per reply.
the moment a tricky async bug or state loop shows up, those same fast models start going in circles until i switch over to a heavier reasoning model.
do you stick to one model for the whole build or swap them depending on the task?
12 views
Replies
Switching happens the second an edit requires importing from another file. A fast model lost context mid-task and silently rebuilt an existing helper two files away under a new name, leaving duplicate logic that went unnoticed until code review.