Elon Musk has released an AI that is smarter than ChatGPT

Elon Musk’s artificial intelligence startup Accounts is unveiled Grok 3is its latest AI model, which the company claims outperforms other leading AI models across key technical benchmarks. The announcement marks an important escalation of the race to develop even more powerful AI systems.

This launch comes just a few days after Musk’s $97.4 billion failed bid to acquire OpenAI, the company he founded with Sam Altman in 2015, was the company he cofounded with Altman. Musk described Grok 3 during a livestreamed demo on X as “an order-of-magnitude more capable than Grok 2″and emphasized the ability to reason complex problems.

Initial testing seems to support xAI claims. The model topped influential Leaderboard of Chatbot Arena: higher scores than OpenAI GPT-4O Google’s Gemini DeepSeek V3 model (19459084) in blind user testing. Published benchmarks show Grok 3, achieving superior scores on mathematics (AIME ’24), scientific reasoning tasks (GPQA), and coding tasks.

Inside Grok 3’s massive computing infrastructure : 200,000 GPUs, and a new Data Center

Former OpenAI researcher wrote “Grok 3 has around state-of-the-art thinking capabilities.” Andrej Karpathy (19459084) in an X-post after early-access testing. “Few models do this reliably. OpenAI’s top thinking models also get it, but DeepSeek-R1, Gemini Flash Thinking 2.0, and Claude don’t.

Developing the model required massive computational resources. xAI doubled the number of Nvidia chips in its GPU cluster, which is housed in a Memphis data center. This infrastructure investment highlights the growing computational demands of advanced AI as companies race to create more capable systems.

Grok 3 was released early today. I believe I am one of the few people who were able to run a quick vibration check.

Think
Grok 3 has a state-of-the-art thinking model (“Think” ) and performed well on my Settler’s Catan game. pic.twitter.com/qIrUAN1IfD

– Andrej Karpathy (@karpathy)””https://twitter.com/karpathy/status/1891720635363254772?ref_src=twsrc%5Etfw””> February 18, 2025 (19659016)

DeepSearch and advanced reason: How Grok 3 hopes to outsmart ChatGPT

and Google Gemini. Grok 3’s “DeepSearch”which combines web search with reasoning capabilities, analyzes information from multiple sources. The system includes specialized modes to solve complex problems, such as a “Think”which shows the reasoning process, and a Big Brain mode that allocates extra computing power for difficult tasks.

The thing to pay attention to when it comes to AI is the learning speed. Tech industry veteran: “And @xai learns way faster than anyone else.” Robert Scobs is citing a conversation between Apple Siri cofounder Tom Gruber.

Grok 3 benchmarks.

In AI, the thing to pay attention to is learning speed. And @xai learns faster than anyone else.

Who is that?

Apple Siri cofounder Tom Gruber. He told me a decade earlier at dinner that this is the most important aspect to pay attention to. pic.twitter.com/yWCiJsN9pU

— Robert Scoble (@Scobleizer) During testing, however, some limitations were discovered. Karpathy noted that this model struggles with certain types humor and ethical reasoning tasks and sometimes fabricates references. These challenges are present in all current AI systems, and they highlight the difficulty of developing a truly human-like AI.

CEO of Scale.ai Alexandr Wang tweeted his praise for the release. He noted the superior performance of the model on various benchmarks, and expressed excitement for future collaboration.

– Alexandr Wang (@lexandr_Wang)””https://twitter.com/alexandr_wang/status/1891714169629524126?ref_src=twsrc%5Etfw””> February 18, 2025 (19659027)

AI industry competition heats-up: What Grok 3’s release means for OpenAI’s DeepSeek, and the future of Artificial Intelligence

This model will be made available through X Premium+ ($40/month), and a standalone ” SuperGrok service($30/month). Enterprise API access will be available in the next few weeks. This launch intensifies the competition in AI, especially as Chinese startups are launching their own products. DeepSeek has recently demonstrated similar performance with lower computational requirements. This development raises concerns about the sustainability and the long-term viability of the AI computational arms race, as companies continue to invest billions into ever more powerful hardware infrastructure.

Musk emphasized that Grok 3 remains in beta, with improvements expected “ “Almost every day ” The company plans to open-source Grok 2 once it stabilizes and add voice interaction capabilities.

But perhaps the most telling thing about Grok 3’s debut, is not its technical specifications or benchmark results, but what it represents – the mounting tension between Musk’s former colleagues at OpenAI. Musk unveiled his new model just days after his failed $97.4 Billion bid to acquire OpenAI. This shows that even a rejected suitor could become a formidable competitor in the high-stakes race to dominate AI.

VB Daily provides daily insights on business use-cases

Want to impress your boss? VB Daily can help. We provide you with the inside scoop about what companies are doing to leverage generative AI. From regulatory shifts and practical deployments, we give you the information you need to maximize your ROI.

Read our Privacy Policy.

Thank you for subscribing. Click here to view more VB Newsletters.

An error occured.