Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

What's a good/affordable GPU to run these projects locally?

It seems like building anything on top of these runs into either a big GPU cost for yourself or a big compute cost if you scale for others.



3090 can run 30b parameter models, 2x can run 65b parameter models.

4090 can run the same, very slightly faster for much more money.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: