You look at the comments here to see if it’s any different or getting rave reviews. Most model are crap, the new mistral one from yesterday seem to be pretty g

You look at the comments here to see if it’s any different or getting rave reviews. Most model are crap, the new mistral one from yesterday seem to be pretty good. But most models are not very practical or useful for anything other than amusement right now. I imagine in a year we’d get close to gpt4 models locally and with low spec requirements where a 2 series NVidia card can run.