Skip to content
System One

Diogo Almeida's Jev launch post (X)

The founder's launch post, around 76k likes, arguing that RLCD-trained decision models are a shorter path to economic value than chat models. Submitted separately to HN as item 49716682.

Still from Diogo Almeida's Jev launch thread (X)Play the video in the post below
Still from x.com

Post on X

Views
39,488,883
Likes
75,897
Reposts
8,214

Read from X on . Counts change daily.

Reported by the author

Speed
20-200x faster20-200x fasterx.com
Cost
40-400x cheaper40-400x cheaper (w/ output tokens free)x.com

Diogo Almeida, who writes that he co-invented ChatGPT, opens with the question he kept asking himself: why have superhuman chat models not led to AGI? The post’s answer is two years of work in stealth on RLCD, a new way to train models, and Jev, which Almeida calls a new type of frontier AI model.

The post lists the launch claims directly: 20 to 200 times faster, 40 to 400 times cheaper with output tokens free, and “frontier composable intelligence optimized for decisions”. It ends by calling this, as far as Almeida can tell, the shortest path to an AI-based economic revolution. A launch video is attached.

Open on X

PrimitivesChoiceScoreNoul

More like this

Keep browsing