Elon Musk’s xAI unveils new Grok 3 model
.jpeg%3Falt%3Dmedia%26token%3D573bce40-3da7-4b8d-8061-b14b9580bb15&w=3840&q=75)
xAI, Elon Musk’s AI company, has now officially launched Grok 3, the latest version of its AI model. The model is being positioned as a direct competitor to OpenAI’s GPT-4o and Google’s Gemini, two of the most popular AI models currently available. Grok 3 is being made available through Musk’s social media platform X (formerly Twitter), where the AI has already been integrated into the service.
Grok-3 comes with vastly more advanced computational power and was developed using xAI’s supercomputer, Colossus, an AI training cluster that is housed in Memphis, Tennessee. This facility boasts an estimated 200,000 GPUs, making it one of the most powerful AI training clusters in the world. According to Musk, the computational resources allocated to Grok 3 are ten times greater than those used for its predecessor, Grok 2. This increase in processing power has allowed xAI to develop a model with enhanced reasoning abilities, improved contextual understanding, and greater accuracy in handling complex queries.
Grok 3 is available exclusively to Premium+ (a service that recently saw its monthly fee rise to $40) subscribers on the erstwhile Twitter. In addition, xAI has announced a new subscription tier, SuperGrok, which will provide access to enhanced AI capabilities beyond those available in the standard Grok 3 model. While official pricing details for SuperGrok have yet to be confirmed, leaked reports suggest a cost of approximately $30 per month or $300 per year. It remains unclear whether this service will function as a standalone offering or require an existing X subscription.
The newest AI model comes with two modes – Think and Big Brain. Think Mode enables Grok 3 to display its reasoning steps, offering users a transparent view of how it arrives at conclusions. As the name suggests, Big Brain Mode aims to provide answers to more complex queries. It deploys additional computational resources to tackle more complex problems in mathematics, science, and programming. Grok 3 has been evaluated using AIME 2025, a standardized mathematical benchmarking framework. Initial reports indicate that it outperforms OpenAI’s o3-mini-high model in several areas related to logical reasoning and computational problem-solving. Musk notes that certain aspects of the model’s processing steps have been deliberately obscured in order to guard against “distillation”, a technique used by rival AI developers to extract knowledge from existing models.
Alongside Grok 3, xAI has introduced Grok 3 Mini (which comes with a dip in accuracy) and DeepSearch, a research-focused AI tool that can analyze and synthesize information from the internet and X’s internal data. Unlike conventional search engines that return a list of links, DeepSearch aims to provide contextually relevant summaries by leveraging Grok 3’s enhanced reasoning capabilities. It seems to be a competitor to OpenAI’s AI-powered research tools, which similarly employ advanced language models to improve information retrieval.
Ilya Sutskever,Ilya Sutskever Safe Superintelligence,Safe Superintelligence,Safe Superintelligence $1 billion funding,Safe SuperIntelligence funding
AI startup Safe Superintelligence (SSI) is now in the process of raising over $1 billion in new funding, thereby taking its valuation to nearly $30 billion. This development places SSI among the most valuable private AI companies in the world, despite the fact that it has yet to introduce a commercial product or generate revenue (unlike OpenAI or Anthropic, which have introduced widely used AI models).
Leading the latest infusion of capital is Greenoaks Capital Partners, a San Francisco-based venture capital firm which has backed several tech companies, including Scale AI and Databricks Inc. Greenoaks is reportedly set to contribute $500 million to this round, according to reports that state sources familiar with the matter. The specifics of the deal are yet to be finalized.
Prior to this latest funding round, SSI had already raised over $1 billion in September 2024, which placed its valuation at $5 billion. The company’s initial backers included prominent venture capital firms such as Sequoia Capital, Andreessen Horowitz, DST Global Ltd., SV Angel, and Northern Four Dragon Gate Capital. The valuation exceeds SSI’s prior goal of seeking a valuation of at least $20 billion, since investor interest appears to have driven the valuation even higher.
It is to be expected, though, since recent years have seen a series of massive investments in AI startups. OpenAI’s ChatGPT had kicked off the AI race, and since then, the industry has seen a surge in capital flowing into AI firms, with several major players securing multi-billion-dollar deals in recent months. For instance, last month, AI startup Anthropic received $1 billion in funding from Big Tech behemoth Google.
SSI was established in June 2024, just one month after Ilya Sutskever departed from OpenAI, the organization he co-founded and where he served as Chief Scientist. His exit followed internal conflicts related to AI safety, including his involvement in the temporary ousting of Sam Altman as OpenAI’s CEO in November 2023. He was joined by Daniel Gross, a venture capitalist and former AI lead at iPhone-maker Apple, and Daniel Levy, a researcher who previously worked at OpenAI. The company maintains offices in Silicon Valley and Tel Aviv (which is a recent expansion) and operates with a focus on developing safe AI.
Unlike other AI firms, Safe Superintelligence does not aim to generate revenue or roll out AI products until it achieves its primary objective. Last year, Sutskever had revealed that SSI is “special in that its first product will be the safe superintelligence, and it will not do anything else up until then. It will be fully insulated from the outside pressures of having to deal with a large and complicated product and having to be stuck in a competitive rat race.”