Elon Musk’s AI company, xAI, released its latest flagship AI model, Grok 3, late Monday night, along with new capabilities in the Grok apps for iOS and the web.
Grok, xAI’s answer to models like OpenAI’s GPT-4o and Google’s Gemini, can analyze images and respond to 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 slated for release in 2024, but missed that deadline.
xAI has been using an enormous data center in Memphis — a data center containing around 2,00,000 GPUs — to train Grok 3. In a post on X, Musk claimed that Grok 3 was developed with “10x” more computing than Grok 2, its predecessor, and with an expanded training data set that ostensibly includes filings from court cases.
“Grok 3 is an order of magnitude more capable than Grok 2,” Musk said during a live-streamed presentation Monday. “[It’s a] maximally truth-seeking AI, even if that truth is sometimes at odds with what is politically correct.”
Grok 3 is a family of models, to be precise — not just one. A smaller version of Grok 3, Grok 3 mini, responds to questions more quickly at the cost of some accuracy. Not all models are available as of yet (and some are in beta), but the rollout begins Monday.
xAI's Grok 3 model is claiming superiority over GPT-4 on various benchmarks like AIME, which tests mathematical problem-solving, and GPQA, which evaluates performance on PhD-level science questions.
An early version of Grok 3 also performed well in Chatbot Arena, a crowdsourced AI model comparison.
Grok 3 comes in two variants: Grok 3 Reasoning and Grok 3 Mini Reasoning. These models are designed to “think through” problems by fact-checking themselves before giving results, making them more reliable than typical models. xAI asserts that Grok 3 Reasoning outperforms OpenAI’s o3-mini-high on benchmarks, including the new AIME 2025 math test.
Grok 3 models are available through the Grok app, where users can ask it to “Think” or activate “Big Brain” mode for more complex queries. The reasoning models are ideal for math, science, and programming questions.
A new feature, DeepSearch, allows Grok to scan the internet and X for information, providing research summaries. Premium+ subscribers can access Grok 3, while SuperGrok, a $30/month plan, unlocks additional features like unlimited image generation and advanced queries.
In the coming weeks, Grok will introduce voice mode, allowing users to interact with a synthesized voice, and Grok 3 will be integrated into xAI’s enterprise API.
xAI also plans to open-source Grok 2 once Grok 3 stabilizes. Musk had originally promoted Grok as unfiltered, willing to engage in controversial topics, but earlier versions exhibited political bias. Musk aims to make Grok more politically neutral, though it remains unclear if that has been achieved.
Source: TechCrunch
Bd-pratidin English/Fariha Nowshin Chinika