Skip to content
View original post on X: Leandro von Werra
Published
34/100

Labs will keep their most capable models internal, Leandro von Werra says

AILeandro von Werra says the most capable cyber, math and bio models are already kept internal by AI labs. He argues labs will keep more intelligence for their own purposes under safety and internal research, beyond what they sell as tokens.
Leandro von WerraVerified on X
@lvwerra
View on X

And it will not stop at AI research. Already the most capable cyber, math and bio models are kept internal.

Under the umbrella of safety and internal research the labs will keep more and more intelligence that creates value beyond tokens for their own purposes.

elie@eliebakouch
with this, i think it's quite likely that anthropic will never publicly release (or at least not in the coming months) a model more capable than opus 5.5 at llm training
View quoted post on X
Source details

Leandro von Werra · x.comPublished · added here

Original title: And it will not stop at AI research. Already the most capable cyber, math and bio models are kept internal.

Topics

Data & trainingExpert opinion