@TSBurkhardt i'm opposed on the general principle that more diversity in models is good and the cost of training these llms is so high that we don't get very many samples of the space of minds possible with this architecture. ofc i also want more open models and i do run local models at home…