We need to talk about Jev

Most likely Jev is and will be a perfect fit for robotics

Were not gonna need as many data centers as we thought

Hi Matthew, thanks for this... hmmm... it’s interesting to see how AI is evolving thanks to human creativity. I wonder how Jev will perform in a private, local environment as an "open-source" model - assuming one is ever released - and whether its capabilities depend on hardware computing power or stem from a novel coding approach. If it’s the former, that would imply the solution is energy-intensive and perhaps not quite as groundbreaking, though improving its efficiency-to-energy-consumption ratio would certainly be an attractive prospect. However, if the latter factor is the key, things get really interesting; it would mean we’ve reached a pivotal moment thanks to RLCD, given that the "Middle Kingdom" boasts some truly talented programmers... and in that case implementing Jav in machines / robots making decisions.... "May God have mercy on us". Boston Dynamics do you smell what I smell?!

so why hasn't @Figure_robot purchased this yet? This would be great for robotics.

This is a great video. So much potential for Jev. Especially being the middle provider.

banger video

I need a t-shirt with their slogan: (best response to everything going on right now)

Speed is the headline, but the real question: does Jev complement LLMs or replace them? If it's a classification/routing layer, this is massive for agent architecture.

Real question: is Jev a tool you use or a mind you're raising? Very different conversations.

bro you are life saver

Thanks Matthew. Nice compilation of examples.

See diffusion open weight models, been out in use for a while now. Not as fast, reliable as newer sliding window attention (SWA) models, like gemma4 and mistral. Advances in AI happen exponentially fast, too fast for humans to keep up with. Nothing is revolutionary anymore.