xAI places 6-trillion and 10-trillion parameter Grok versions into training
xAI has placed 6-trillion and 10-trillion parameter versions of Grok into training as part of wider industry moves to expand AI training runs by orders of magnitude in compute. The effort follows Anthropic’s March scale-up that produced Mythos and is expected to continue across labs. Cursor AI and SpaceXAI are jointly training a significantly larger model from scratch on Colossus 2’s million H100-equivalent GPUs, using ten times the compute of prior runs together with pooled data and training methods.
#517Andrew Curran@ANDREWCURRAN_Elon also has a 6T and 10T version of Grok in training now. The step-change that emerged when Anthropic had enough compute to train Mythos in March will start appearing from dozens of other places this year.
#517Andrew Curran@ANDREWCURRAN_
Sualeh Asif@sualehasif996Big model coming soon ... !
#1733Ryo Lu@RYOLU_…and more to come
#1749Sualeh Asif@SUALEHASIF996Big model coming soon ... !
