Yeah, at this point I think it’s somewhat of a useless distraction. They do the validation work to prevent any problems from it, but an immense amount of drama seems to arise every time the topic comes up, and I’m sure Anthropic (and many insufferable AI bros) have been crowing to high heaven that Linus Torvalds uses it.

They’ve clearly chosen to allow it at the highest level for anyone competent enough to validate the work and stand by it. Claude/Gemini/ChatGPT won’t be answering reviewer questions about the implementation, that’s up to the human standing by the work, so they only need to know about the human, from a practical perspective.

This is truly horrible, amazing work.

This just in: Doing something takes longer than not doing that thing. Only Denuvo stupid enough to act surprised.

It is the world’s first open 3T-class model

Good lord that’s a lot of VRAM. Hopefully soon there’s another phase of finding good ways to distill this recent progress into smaller models I can actually run locally.