You can run Llama models for free locally, pretty much every AI player release stuff out there for everyone to use for free. It’s surprisingly “open” like that.
And you cannot run the real R1, that still requires a huge investment. It can run reduced quants, barely any different to the other current public offerings.
It seems you can run the full R1 on a 96GB ram gaming rig even without GPU. I have little practical experience because of my old PC but it seems both improvements to efficiency to run locally (at least for a group of people) and research about bias or poisoning is being done.
You can run Llama models for free locally, pretty much every AI player release stuff out there for everyone to use for free. It’s surprisingly “open” like that.
And you cannot run the real R1, that still requires a huge investment. It can run reduced quants, barely any different to the other current public offerings.
It seems you can run the full R1 on a 96GB ram gaming rig even without GPU. I have little practical experience because of my old PC but it seems both improvements to efficiency to run locally (at least for a group of people) and research about bias or poisoning is being done.
BRB gotta buy another 64gig RAM kit.