[meisel]

I think only a small percentage of users care that much about running LLMs locally to pay for extra hardware for it, put up with slower and lower-quality responses, etc. . It’ll never be as good as non-local offerings, and is more hassle.