Musk’s reply on X said xAI was improving the 0.5T Grok foundation model V8, shipped publicly as Grok 4.3, every few days; that the 1.5T V9 had just finished training and would next take Cursor data in supplemental training; and that release was due in three to four weeks (post). It updates his February claim that Grok 4.2 would improve every week through continuous post-training, and lands three months after an xAI co-founder resigned warning that live self-improvement loops were a year away. Within the recursive-self-improvement cluster it is the point where the cadence of improvement drops below the cadence of release: the model users are running is not the model they ran last week.
Development
A shipped model is reported improving every few days
A release stops marking a finished artifact: the frontier model in production is itself a moving target, updated on a cadence of days while its successor trains behind it.
Claim
Elon Musk reported the half-trillion-parameter Grok 4.3 is now improving every few days while a 1.5-trillion-parameter successor has finished pre-training and is about to start mid-training on data from a coding-tool deal, with release due in three to four weeks.
Reported metric: 0.5T → 1.5T
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to May 17, 2026 (theinnermostloop.substack.com)
- Elon Musk on improving the 0.5T Grok foundation model every few days (X reply) (x.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.