Elon Musk's xAI launches its latest model, the Grok 3

Elon Musk's AI company, xAI, late Monday released its latest artificial intelligence model, Grok 3, and unveiled new capabilities for Grok's iOS and web apps.

Grok, xAI's answer to models like OpenAI's GPT-4o and Google's Gemini, can analyze images and answer questions, and powers a number of features on Musk's social network, X.

Grok 3, which has been in development for several months, was optimistically scheduled for release in 2024, but missed that deadline. Monday was an ambitious launch.

xAI has used a large data center in Memphis containing about 200,000 GPUs to train Grok 3.

In a post on X, Musk claimed that Grok 3 was developed with “10x” more computing power than its predecessor, Grok 2, using an expanded training dataset that allegedly includes court case files.

"Grok 3 is an order of magnitude more capable than Grok 2," Musk said during a live presentation on Monday, the Telegraph reports.

“[It is] an AI that seeks the truth to the fullest, even if that truth sometimes contradicts what is politically correct,” he added.

The Grok 3 is a family of models, to be exact. A smaller version of the Grok 3, the Grok 3 mini, answers questions faster at the cost of some accuracy.

Not all of Grok 3's models and related features are available yet (some are in beta), but they started rolling out on Monday.

xAI claims that Grok 3 beats GPT-4o on benchmarks including AIME (which evaluates a model's performance on a sample of math questions) and GPQA (which evaluates models using PhD-level physics, biology, and chemistry problems).

An early version of Grok 3 also scored competitively in Chatbot Arena, a multi-source test that pits different AI models against each other and has users vote on their favorite responses, according to xAI.

Two models in the new Grok 3 family, Grok 3 Reasoning and Grok 3 mini Reasoning, can carefully “think through” problems, similar to “reasoning” models like OpenAI’s o3-mini and Chinese AI company DeepSeek’s R1.

Reasoning models try to check themselves before they produce results, which helps them avoid some of the pitfalls that typically hinder models.

xAI claims that Grok 3 Reasoning outperforms the best version of o3-mini – o3-mini-high, on several popular benchmarks, including a newer math benchmark called AIME 2025.

These reasoning models can be accessed via the Grok app. Users can ask Grok 3 to “Think” or – for more difficult questions – use the “Big Brain” reasoning mode that uses additional calculations.

xAI describes reasoning models as best suited for math, science, and programming questions.

Musk said that some of the reasoning models' "thoughts" are hidden in the Grok app to prevent distillation, a method used by AI model developers to extract insights from other models.

Recently, DeepSeek was accused of distilling OpenAI's models to create its own.

Grok's reasoning models support a new feature in the Grok app called DeepSearch, xAI's answer to AI-powered research tools like OpenAI's deep search.

DeepSearch scans the internet and X to analyze information, and provide an abstract in response to a query.

Subscribers on X's Premium+ tier ($50 per month) will initially have access to Grok 3, and other features will be rolled out after a new plan that xAI calls SuperGrok.

For a price of $30 per month or $300 per year (if the information is to be believed), SuperGrok unlocks additional reasoning questions and DeepSearch throws in unlimited image generation.

In the future – about a week from now – the Grok app will gain a “voice mode,” Musk said, which will give Grok models a synthesized voice.

A few weeks after that, Grok 3 models will be available through xAI's enterprise API, along with the DeepSearch capability.

"xAI plans to open source Grok 2 in the coming months," Musk said.

“Our general approach is that we will open source the latest version [of Grok] when the next version is fully released,” he continued.

"When Grok 3 is mature and stable, which is probably within a few months, then we will open source Grok 2," Musk emphasized.

The American billionaire announced Grok nearly two years ago, presenting the AI ​​model as tense, unfiltered, and anti-“woke” – in short, ready to answer controversial questions that other AI systems won’t.

He kept some of that promise. It's said to be vulgar, for example, Grok and Grok 2 would happily oblige, spewing colorful language you probably wouldn't hear from ChatGPT.

But Grok models before Grok 3 were protected by political entities and would not cross certain boundaries.

In fact, a study found that Grok leaned toward the political left on topics like transgender rights, diversity programs, and inequality.

Musk has blamed the behavior on Grok's training data - public websites - and vowed to "move Grok closer to political neutrality."

It is not yet clear whether xAI has achieved this goal and what the consequences might be. /Telegraph/