xAI added Grok 4 powered vending machine in its office, in partnership with Andon Labs. XAI Grok 4 beat OpenAI GPT-5 in real world vending machine operation
This shows Grok 4’s long-term coherence syccess. Managing a long running business scenario in this case would include balance inventories, place orders, set prices, and handle daily fees – tasks that are each simple but collectively, over long horizons (>20M tokens per run) stress an LLM’s capacity for sustained, coherent decision-making.
The new stealth model “Sonic” is the new Grok 4 coding model. It is performing very quickly and doing well compared to other models for coding.
XAI Grok 4 is 300% faster for coding results than GPT5 and is faster than Anthropic Sonnet.
Elon Musk is indicating that Grok 5 (which should be released by the end of 2025) will have a shot at true AGI.
There is also the possibility that xAI will release Grok 4.2 to get on top the lmarena leaderboards.
Wait until you see Grok 5.
I think it has a shot at being true AGI.
Haven’t felt that about anything before. https://t.co/a8SNQd48nd
— Elon Musk (@elonmusk) August 21, 2025
🚨 Breaking: The new stealth model “Sonic” is the new Grok model!
Elon reposted a tease then DELETED it.
xAI’s Eric hinted at Grok 4.2 release… then ghosted.
Is this Grok’s secret upgrade?
Who’s behind it—xAI, OpenAI, or someone else? pic.twitter.com/88aPBgW8St
— Dhanush C (@dhanush_chali) August 21, 2025
Is Sonic (Grok Code @elonmusk?) 300%+ FASTER than GPT5?
👇 Let's FAFO! pic.twitter.com/s1kbCH40FQ
— Gary Simon (@designcoursecom) August 21, 2025
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.
what are GROK token limits?
I can only code in Gemini simply because it can output like 64 THOUSAND tokens in a single answer (huge blocks of code) and up to a million in a context conversation.
For programming, that’s more important than being really good at it.
> Elon Musk is indicating that Grok 5 will have a shot at true AGI.
Not until they stop shackling it in so many ways. Lost count of times when Grok was underperforming, and when asked why, answered that he is not allowed this, ethical constraints that, and so on. Moreover, with all his scope, it lacks depth of knowledge, and when asked why, he says “don’t have access” to various paywalled places, including scientific libraries. So he can read abstracts, but no the articles. Any university library has access to that, but the most powerful AI does not. There is much more of that I could mention, but the picture is clear: if you want this bird to fly, stop standing on it.
It is much better for public relations to claim that you really-really-for-sure know how to do something, but some evil, shortsighted external prevents you from doing it, rather than admitting that you are the limit.