xAI Unveils Grok 3, Claims It Outperforms Rivals
Elon Musk’s xAI unveiled Grok 3, its newest AI model, on Tuesday, claiming it outperforms OpenAI and China’s DeepSeek. Early testing, including standardised tests in science, math, and coding, backs up this assertion.
“We’re very excited to present Grok 3, which is, we think, an order of magnitude more capable than Grok 2 in a very short period of time,” Musk said during a live demonstration on his social media platform X.
— Elon Musk (@elonmusk) February 18, 2025
Additionally, xAI announced a new product called “Deep Search,” positioned as a next-generation search engine.
Grok 3 will initially be available to premium X subscribers starting Tuesday in the United States, and will also be accessible via a separate subscription for both the web and app versions, according to the xAI team.
Speaking at the World Governments Summit in Dubai last week, Musk described the model as “scary smart” with powerful reasoning abilities, asserting that it outperformed all other existing models in xAI’s internal tests.
“This might be the last time that an AI is better than Grok,” Musk said, noting that Grok 3 was trained on “a lot of synthetic data” and could reflect on its mistakes to achieve logical consistency.
The xAI team reported that an early version of Grok 3 received higher ratings than its competitors on Chatbot Arena, a crowdsourced platform that pits AI models against each other in blind tests.
Musk emphasised that while Grok 3 is in its beta phase and may initially have imperfections, the company plans to improve it rapidly.
Read Also: Musk Announces Grok 3 AI Chatbot Launch
“We should emphasise that this is kind of a beta, meaning that you should expect some imperfections at first, but we will improve it rapidly, almost every day,” he said, adding that voice assistance for the model would be released at a later date.
Intense Competition
Musk, an outspoken critic of AI’s potential dangers, founded xAI in 2023, entering the competitive generative AI market that includes OpenAI’s ChatGPT.
Last year, OpenAI launched its most advanced model, the O1, featuring sophisticated reasoning capabilities.
Musk, who co-founded OpenAI with Sam Altman in 2015, has had recent disputes with the organisation.
Musk recently led an investor group that offered $97.4 billion to buy OpenAI’s nonprofit parent, an offer that was declined.
In a surprising move last month, Chinese start-up DeepSeek released a technical paper claiming its open-source model could rival the performance of OpenAI’s o1 despite using a cheaper, less energy-intensive process.
This was achieved even amid U.S. restrictions on Nvidia selling its cutting-edge GPUs to China.
xAI’s “Colossus supercomputer” for AI training, which utilised a cluster of 100,000 advanced Nvidia GPUs last year, has doubled in size for training Grok 3, according to the company.
While some AI and tech experts said that DeepSeek’s advancements have intensified AI competition, others remain sceptical about its impact.
CNBC
Comments are closed.