I was lucky that I got the refurbed dual xeon with 128Gb before memory prices went full retard. But I could run this on 64Gb easy and on 32Gb if I optimized the AI stuff a bit.
I went the VM route because I new it already... and figured I could back Hermes up easier that way. You can go faster and more streamlined using Ollama and Docker but I was *exhausted* from learning so much crap.
I pay about $2/mo for Brave search and maybe $5 for YT transcripting.
(cont)
