Researcher questions OpenAI's anti-distillation blog: conclusions may not hold at scale

maksym_andr · x · 2026-11-22

A reply in the discussion around OpenAI's anti-distillation blog post. The quoted tweet argued the post is a big deal: it appears to concede that at medium reasoning effort — which most people use — there is nothing OpenAI can do to prevent distillation and protect the product; only very high reasoning effort can be limited, and that's rarely used.

Maksym Andriushchenko (EPFL) pushes back: he's not sure strong conclusions should be drawn from that plot — it depends heavily on how exactly distillation is implemented, and he doubts the blog's authors spent as much effort on it as those who leverage distillation at industrial scale.

Related event: Researcher Questions OpenAI's Anti-Distillation Claims(4 posts)→

Original post →

More from Models

Models channel →