I'm running a totally usable 13b parameters llama model in my macbook air, which seems to give outputs equivalent to what I was getting from GPT3 in June 2022.
How much more hardware would it really be needed for GPT-4 level outputs natively? Perhaps software optimizations alone could do most of the trick.
Most of the point of skipping electron is memory usage, which is a major issue with having a full separate installation of Chrome just to handle a webpage, which is what Slack is.
People from outside Brazil are missing a lot for not having contact deAssis' work.
It is so hard to break the barrier that separates Brazilian literature to the outside. Fortunately Posthumous Memoirs of Brás Cuba was somewhat successful on this task after two centuries.
Hope others follow.