/TECH

Nathan Lambert disputes claims Chinese labs use Fable for RL distillation

NL

FI

Lambert cites high model costs and noisy RL graders.

Combined views

106.2K

3 posts, first seen 19h ago

590 likes

Reactions from ranked influencers

3 posts

Combined views

106.2K

3 posts, first seen 19h ago

590 likes comments

29 comments

132 saves

33 reposts

Sentiment

Positive50%50%Negative

Positive Read

Positive users request a Stratechery interview with Nathan Lambert on RL, while negative users dismiss Ben Thompson's take on Chinese labs skipping strong models in RL distillation as wrong or uninformed.

Based on 4 visible X reactions from 31 accounts; directional sample.

Digg Deeper

Ask a question below.

Published answers will appear here.

Reactions from the X community

SC scorz@scorzeth

10:16 AM · Jul 21, 2026

@natolambert @xeophon @benthompson I need a Statechery Interview with Nathan Lambert asap @benthompson

JE Jester@lajoiedeslutins

10:28 AM · Jul 21, 2026

@natolambert @benthompson the real distillation happening here is nathan distilling ben's take down to just wrong

SP Sam Parr’s Dealer@samparrsdealer

10:33 AM · Jul 21, 2026

@natolambert @benthompson Yes love Ben, but he clearly does not understand RL. He sounded retarded

DG Dipanshu Gupta@dipshady_

9:16 AM · Jul 21, 2026

@natolambert @benthompson @benthompson Would love a @stratechery interview with Nathan on RL

View all

NL Nathan Lambert@natolambert

8:17 AM · Jul 21, 2026

73TECH

Yo @benthompson I'm sorry but the Chinese labs aren't using Fable / the strongest models as teachers during RL, that's not how distillation works. It wouldn't give that big of a lift (graders during RL are messy) and you cant afford to use Fable like that.

NL Nathan Lambert@natolambert

9:55 AM · Jul 21, 2026

73TECH

@teortaxesTex @benthompson yes, but I think it's not "making distillation more impactful" as Ben said (the crucial point that people will latch to)

FI finbarr@finbarrtimbers

12:56 PM · Jul 21, 2026

656TECH

Co-sign!

29