Muhammed Selim Korkutata | Anadolu | Getty images
The XAI of Elon Musk unveiled its latest artificial intelligence model, Grok 3 on Tuesday, claiming that it can surpass Openai and Chinese depth offers depending on the early tests, which included standardized tests on mathematics, Sciences and coding.
“We are very happy to present Grok 3, who is, we think, an order of magnitude more capable than Grok 2 in a very short period of time,” said Musk during a demonstration of Grok 3 which was broadcast on His social media platform X.
The team also said that it launched a new product called “Deep search”, which would serve as a “new generation search engine”.
Grok 3 will be deployed for Premium X subscribers, from Tuesday in the United States, and will also be accessible thanks to a separate subscription for the web and app versions of the model, said team XAI.
Speaking at the World Government Summit in Dubai last week, Musk had nicknamed the “intelligent” model, with powerful reasoning capacities, saying that it has surpassed all other existing models in XAI internal tests.
“It may be the last time that AI is better than Grok,” said Musk at the time, adding that it was trained on “a lot of synthetic data” and was able to think about its mistakes to achieve logical consistency.
The XAI team said that an early iteration of Grok 3 had received better marks than existing competitors Chatbot arenaA Crowdsource website that offers different models of AI against each other in blind tests.
Towards the end of the product demo, Musk said that the company will continue to improve the model.
“We must emphasize that this is a kind of beta, which means that you should expect certain imperfections at first, but we will improve it quickly, almost every day,” he said , adding that vocal assistance for the model would be published in a later.
Intense competition
In September of last year, Openai launched its most advanced model, the O1, which came with The capacities and was able to solve relatively complex science, coding and mathematics.
Musk, with Sam Altman, helped to create Openai as a non -profit organization in 2015.
However, in recent years, the leaders of Musk and Openai have been in Queregue. Musk recently directed a group of investors who submitted a proposal to buy the non -profit parent from the AI startup for $ 97.4 billion – an Openai offer has decreased.
Last month, the Chinese start-up Deepseek shocked the AI market when it published a technical document which said that one of its open source models was able to compete with the performance of the O1 model of Openai Despite the use of a cheaper and less intensive energy process.
He accomplished the feat in front of the hand Nvidia From the sale of its advanced GPUs – used for the formation of AI models – to China.
XAI has a “Colossus Supercompleut,“For AI training, which, according to last year, used a group of 100,000 NVIDIA GPUs advanced for the formation of AI. On Tuesday, the company revealed that it had doubled the size of its cluster GPU for the formation of Grok 3.
While many IA and technology experts told CNBC that Deepseek had intensified IA competition, showing what can be done with less advanced technology, others are more skeptical about its impact .
