Nathan Lambert estimates RL training for inkling model cost $5 million

/TECH

The estimate assumes 1,000 GB300 GPUs ran for one week.

Combined views

26.8K

4 posts, first seen 3h ago

314 likes

Source Posts on X

NL

Nathan Lambert@natolambert

Let's estimate the scale of the inkling final RL run? An initial guess O(1K GB300s) ~1 week ~$5M? Would be a fun interview question for a post-training lab. Glad to see some info like this shared publicly.

11:27 AM · Jul 15, 2026

Reply

RA

Rishabh Agarwal@agarwl_

The scaleRL paper did 15M rollouts to understand the science of scaling!

12:23 PM · Jul 15, 2026

Reply

Combined views

26.8K

4 posts, first seen 3h ago

314 likes10 comments

10 comments

178 saves

18 reposts

Sentiment

Positive0%0%Negative

Guardrails removed spam, off-topic, unclear, or duplicate replies.

Digg Deeper

Ask a question below.

Published answers will appear here.

Source Posts on X

NL

Nathan Lambert@natolambert

Let's estimate the scale of the inkling final RL run? An initial guess O(1K GB300s) ~1 week ~$5M? Would be a fun interview question for a post-training lab. Glad to see some info like this shared publicly.

11:27 AM · Jul 15, 2026

Reply

RA

Rishabh Agarwal@agarwl_

The scaleRL paper did 15M rollouts to understand the science of scaling!

12:23 PM · Jul 15, 2026

Reply

FI

finbarr@finbarrtimbers

@natolambert Checkpoint spacing is interesting, wonder why it varies

11:56 AM · Jul 15, 2026

Reply