As of: 2026-11-13 13:49 CET.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/p/x-elonmusk-2120088070443655190/

# “We have dumped JAX/SGLang and rewritten the entire RL and inference stack from…”

Elon Musk (@elonmusk), X, 2026-11-10.
Source: https://x.com/elonmusk/status/2120088070443655190

## Archived text

> We have dumped JAX/SGLang and rewritten the entire RL and inference stack from scratch in C/C++, which took some extra time. 
> 
> Training (aka pre-training) is done. Lot of cleanup was needed on the SFT data to get that mostly right. Now, 4.8 is in RL and the evals are starting to climb. We will do a soft rollout to Bot users later this week and see how it goes.
> 
> Grok 4.8 will bear no resemblance to 4.7 (thankfully). It is not yet incredible, but, this time, the foundations are built right, so I expect it to evolve to be something special.

Archived 2026-11-12 via fxtwitter (unofficial).
Counts: 4,066,390 views, 29,159 likes, 1,583 reposts, 1,675 replies (at fetch time)

## Cited in

- 2026-11-10: [Musk says Grok 4.8 is 'not as good' as Claude Opus 5.5, 'a masterpiece', and that xAI rewrote its RL and inference stack in C/C++](https://postcutoff.com/e/2026-11-10-musk-opus-5-5-masterpiece-grok-4-8/)
- Coming up: [xAI expected to release Grok 4.8](https://postcutoff.com/upcoming/#2026-10-01-grok-4-8-release)