Elon Musk posted on September 2, 2026 that Grok 4.7 comes out in 10 days so September 12.
2.1 trillion parameters — a new, larger pretrain, up from Grok 4.6’s 1.5T base.
Elon has said it will be better than 4.6 in every way except that it will be slightly slower to serve, with better token efficiency.
Grok 4.6 was released Aug 12.
Grok 4.7 comes out in 10 days https://t.co/ZSXmzVFqB1
— Elon Musk (@elonmusk) September 2, 2026
Initial pre-training was finished by mid-August. There has been supplemental training on a large amount of SpaceX company data.
Grok 4.7 will exceed all current models. So better than Fable 5 but not Fable 5.1 Grok 4.7 model will be the best at real-world engineering.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
So far, of the parts of the AI stack, models remain the most commoditized. This is good for the consumer.
I’ve been looking at the mysterious free AIs that have appeared in the past month, especially Ox Alpha in the past week. It works far deeper for me than the free levels of grok and chatgpt, and isn’t that much worse than my paid subscription of Claude. My understanding is that Ox Alpha was unmasked yesterday as a Chinese AI, and all my coding efforts have gone to Beijing. That will confuse them 🙂 Regardless, the paid version is much cheaper than Claude or Grok. The videos I’ve seen on this is that rather than the brute width and depth of grok, they employ some clever bucket reasoning. Polar bears and medieval poetry don’t really need to have much time spent linking the two. Maybe the future winners of AI will be the people with the most cunning optimisation, rather than the people who manage to rent the most gas turbines.
Maybe. But the “Bitter Lesson” is that historically the win went to more compute not more clever human optimization – over and over. Bitter obviously to the clever humans.