You can run some models pretty decently using CPU inference only, things like Gemma 3 that are built for exactly that use case or some tiny speech to text model

You can run some models pretty decently using CPU inference only, things like Gemma 3 that are built for exactly that use case or some tiny speech to text models via llama.cpp that I have tested out (not so good). Although not the best for "heavy" tasks, if you just need a decent text generator that can produce more or less sensible, generic output you are good to go.