Jonathan Raa | Nurphoto | Getty Photos
Elon Musk’s xAI on Tuesday unveiled its latest artificial intelligence model, Grok 3, claiming it’ll outperform offerings from OpenAI and China’s DeepSeek according to early testing, which included standardized tests on math, science and coding.
“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 mentioned at an illustration of Grok 3 that was once streamed on his social media platform X.
The team also mentioned it was once launching a modern product known as “Deep Search,” which would act as a “next generation search engine.”
Grok 3 will be rolled out for prime price X subscribers, starting up Tuesday stateside, and ought to restful also be accessible thru a separate subscription for the model’s web and app versions, the xAI team mentioned.
Speaking at The World Governments Summit in Dubai final week Musk had dubbed the model “scary smart,” with highly effective reasoning capabilities, claiming it outperformed all other modern devices in xAI’s inner tests.
“This might be the last time that an AI is better than Grok,” Musk mentioned at the time, in conjunction with that it was once trained on “a lot of synthetic data,” and was once able to reflecting upon its mistakes to help out logical consistency.
The xAI team claimed that an early iteration of Grok 3 had been given greater rankings than modern opponents on Chatbot Areaa crowdsourced web residing that pits a form of AI devices in opposition to every other in blind tests.
Toward the pause of the product demo, Musk mentioned that the firm will address bettering the model.
“We should emphasize 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 mentioned, in conjunction with that the enlighten assistance for the model would be released at a later time.
Intense competition
Musk, who has been barely vocal referring to the functionality dangers of man-made intelligence, started xAI in 2023 entering the generative AI market that involves OpenAI’s ChatGPT.
In September final year, OpenAI launched its most developed model, the o1, which came with reasoning abilities and was once in a space to resolve somewhat advanced science, coding and math projects.
Musk, alongside with Sam Altman, helped develop OpenAI as a nonprofit in 2015.
Nevertheless, in latest years Musk and OpenAI’s management have been feuding. Musk no longer too long previously led an investor neighborhood that submitted a proposal to aquire the AI startup’s nonprofit guardian for $97.4 billion — a proposal OpenAI declined.
Final month, Chinese language start-up DeepSeek vexed the AI market when it released a technical paper that claimed one amongst its start source devices was once in a space to rival the efficiency of OpenAI’s o1 model despite the exercise of a inexpensive, much less energy-intensive direction of.
It carried out the feat in the face of the U.S. restricting leading AI chipmaker Nvidia from promoting its cutting-edge GPUs — historic for practising AI devices — to China.
XAI has a “Colossus supercomputer,” for practising AI, which it mentioned final year was once the exercise of a cluster of 100,000 developed Nvidia GPUs for AI practising. On Tuesday, the firm published that it doubled the dimensions of its GPU cluster for the practising of Grok 3.
While many AI and tech experts have told CNBC that DeepSeek has intensified AI competition, exhibiting what is also achieved with much less developed skills, others are extra skeptical about its affect.
