Elon Musk’s artificial intelligence startup xAI has debuted its updated Grok-3 model chatbot technology which the company claims outperforms its rivals.
The new Grok-3 has “more than 10 times” the computing power of its predecessor, claimed Musk.
The startup said that across math, science and coding benchmarks, Grok-3 beat OpenAI’s GPT-4o, Google Gemini, DeepSeek’s V3 model and Anthropic’s Claude.
xAI claims Grok 3 beats GPT-4o on benchmarks including AIME which uses a sample of math questions and GPQA which uses PhD-level physics, biology, and chemistry problems.
Two models in the Grok 3 family, Grok 3 Reasoning and Grok 3 mini Reasoning, are reported to even be able to “think through” problems, similar to “reasoning” models like OpenAI’s o3-mini and Chinese AI company DeepSeek’s R1.
A new search engine has been introduced by way of Grok-3 called DeepSearch. DeepSearch acts as a reasoning chatbot that expresses its process of understanding a query and how it plans its response using research, brainstorming and data analysis.
xAI used an enormous data centre in Memphis containing around 200,000 GPUs to train Grok 3, according to Tech Crunch.
“Grok 3 is an order of magnitude more capable than Grok 2,” said Musk during a livestreamed presentation announcing the latest model. “[It’s a] maximally truth-seeking AI, even if that truth is sometimes at odds with what is politically correct.”
He noted that part of the training set used for the chatbot included filings from court cases.
xAI is starting a new subscription called SuperGrok for the bot’s mobile app and Grok.com website, and plans to open-source versions of Grok models.
Musk recently offered to buy OpenAI’s nonprofit arm for $97.4 billion, a bid that was rejected by OpenAI’s chief executive Sam Altman.
In December, xAI closed its Series C funding round of $6 billion (A$9.43 billion), backed by the likes of Blackrock, Kingdom Holdings, and Sequoia Capital, among others.
Chinese chatbot DeepSeek which was unveiled its chatbot in January sent shockwaves through the tech industry when it claimed that it developed its model with a budget of around A$9.6 million, a tiny number compared to the billions that some of its competitors are spending to develop similar generative AI services.
