logoalt Hacker News

aembletonyesterday at 9:59 AM0 repliesview on HN

You could run the inference locally using the open weight models. Then you'd still get to use the models without sending money overseas