SpaceXAI Delays Grok 4.7 Launch as 2.1-Trillion-Parameter Model Undergoes Final Refinement
Sign in✓ Signed in
SpaceXAI announced a delay to the release of Grok 4.7, its 2.1-trillion-parameter large language model, just one day before the scheduled September 12 launch, as the team applied additional refinements to improve task completion and reasoning rigor. Elon Musk indicated on September 11 that the model "needs a few more days to cook," citing over-correction in reinforcement learning that had excessively penalized longer responses, causing the system to prematurely give up on complex problems it could otherwise solve.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...