Those efficient open weight models still need 500k of hardware to run the good ones for a few concurrent users (single digit amount). They still require a ton of resources for training.
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments