Grok 3: xAI's Latest AI Model Shakes Up the Chatbot Landscape

What is Grok: This chatbot is brimming with attitude

Created by Elon Musk's startup xAI, Grok is an AI chatbot that seeks to differentiate itself from its competitors by being more than happy to answer "spicy questions that are rejected by most other AI systems." As a result, it's often sassy, snarky and sarcastic, injecting humour into conversations. It's also unfiltered which has proven rather controversial - but then it does largely reside within X which is much more of a free-for-all these days. Let's check it out. This article was correct as of February 2025. AI tools are updated regularly and it is possible that some features have changed since this article was written. Some features may also only be available in certain countries. Originally set to be called TruthGPT, Grok was initially released by xAI in November 2023 as a feature usable by Premium+ users of X. It's since been made available to all and it's becoming well known for adopting an edgier approach to its chats. And while it's very much integrated within X, it's also been stepping out of the social media bubble in recent times with its own app. It's certainly become one to watch. As with rivals, it makes use of a large language model but, in this instance, it's also draws upon the millions of posts on X (as well as data pulled from Tesla). It uses those posts for training and, crucially, to help provide answers to all manner of questions. Grok can access web pages too, which allows it to stay abreast of current affairs. Since it's a chatbot, Grok is able to answer questions on just about any topic. You can request all manner of things from recipes and code to news summaries and recommendations. You can also use it to analyze PDFs and images, summarize texts and check out the latest trends. The chatbot can get personal. It will draw upon your own tweets and work out your interests - great for, say, generating book and podcast recommendations but it may make you a little wary about privacy. There's a "roast me" feature too which again draws on your interests to make you the butt of a host of jokes - great if that's your thing. Grok lets you create images using a video model called Aurora. It's capable of generating photorealistic images and there are no restrictions on using real-life people or characters. This has already led to deepfakes featuring the likes of Taylor Swift, Donald Trump and Mickey Mouse. It will draw the line at fully nude images, though. Grok is mainly embedded within Twitter and while it's beginning to branch out a bit, it's not yet being widely integrated into other apps - you can't call upon it with a word processor app such as Word or Pages, for instance. Grok's personality and unfiltered answers also make it relatively unsuitable for business or educational use - its potential for causing offence can also be off-putting. And while you will see citations when Grok draws upon a particular source, the service has been known for sharing misinformation. Grok has a free tier and it allows you to make 10 requests every couple of hours along with four image generations and three image analyses. If you subscribe to the Premium and Premium+ tiers of X, however, there are no limitations to how you use Grok. A Premium subscription costs $8 / £8/ AU$13 each month while Premium+ is $22 / £17/ AU$35 each month - which one you opt for will depend on how you intend to use X as a whole. For example, Premium+ gives you early access to newer versions of Grok. Grok is available by clicking Grok in X's menu and also directly from a post - if you select the Grok logo on a post, Grok will lend some context and allow you to start asking questions based on it. Grok also has a standalone website at grok.com and there are apps for iOS and Android. Grok is particularly good at keeping you up to date with current trends and goings on - being integrated with X and having web access means it's a rather decent search engine. It has good general knowledge and it's certainly good at drawing on your interests to create relevant answers. Whether or not you like your X posts feeding Grok's large language model is another question. Grok's image generation capabilities are exceptional so long as you don't start pushing the boundaries of taste. Grok has humour too and while it may not be to everyone's liking, it does at least lend personality. Its coding abilities are on a par with GPT-3.5 and the whole package shows promise. Grok 3 (currently only for Premium+ subscribers) is taking everything to another level too so it'll be interesting to see where this AI chatbot goes. You should use Grok if you want an AI that can react quickly to real-time topics and be reflective of internet culture and trends. You'll also use Grok if you're looking for AI to generate images that involve people including yourself because the results are great. Grok is also useful if you simply desire an AI that doesn't take life seriously - it's fun and unfiltered. But is that always a good thing? You want an AI for work or for education - it's unfiltered nature means it's not always suitable. Don't use Grok if you want to produce deepfakes either - yes, it can produce them but that doesn't mean you should (ok, preachy preachy). Don't use Grok if you want app integration either - for that, the likes of Copilot and Apple Intelligence make more sense. ChatGPT does much of what Grok achieves other than being entirely up-to-date but if that doesn't matter then you'll find this one does a very professional job. Flux AI is also great at generating images in multiple styles using natural language descriptions.

What is Grok? -- everything you need to know about xAI's chatbot

Grok, the AI assistant developed by xAI, has quickly evolved from an early prototype into a competitive player in the AI space. Founded by Elon Musk in July 2023, xAI aims to advance artificial intelligence with the goal of deepening human understanding of the world. Since its launch, Grok has undergone rapid development, positioning itself alongside established AI models from companies like OpenAI and Google. The name Grok comes from the 1961 science fiction novel Stranger in a Strange Land by Robert A. Heinlein. In the book, "grok" is a Martian word that means to deeply understand something in a way that transcends language -- essentially to fully and intuitively grasp a concept as if becoming one with it. Elon Musk and xAI likely chose the name Grok to reflect the AI's goal of deeply comprehending and reasoning about the world, rather than just responding with surface-level information. The name aligns with xAI's mission of advancing AI's understanding of the universe. xAI was established with a stated mission to "understand the true nature of the universe." With Grok, the company set out to create an AI model that provides informative and engaging responses while integrating real-time data from X (formerly Twitter). Launched in November 2023 as a beta chatbot for X Premium users, Grok was designed to offer insightful, context-aware answers, with an emphasis on reasoning and adaptability. Despite being introduced as an early-stage product, Grok demonstrated rapid iteration. Developed in just two months, its initial beta phase laid the foundation for significant improvements, supported by a team of AI researchers and engineers with experience from DeepMind, OpenAI, and Tesla. Grok-0, the first iteration of the model, featured 33 billion parameters and demonstrated strong performance despite being trained with fewer resources than larger competitors. In benchmarks such as MMLU (multidisciplinary questions) and GSM8K (math word problems), it performed comparably to models with significantly higher parameter counts, highlighting xAI's focus on efficiency and optimization. In October 2023, xAI introduced Grok-1, a 314-billion-parameter Mixture-of-Experts (MoE) model. MoE architectures improve efficiency by activating only a fraction of the model's total parameters per task, making them highly scalable. Grok-1 was later released as an open-source model under the Apache 2.0 license in March 2024, allowing developers to explore its architecture and capabilities. Benchmarks placed Grok-1 ahead of models like Meta's LLaMA 2 (70B) and OpenAI's GPT-3.5 in several areas, though it remained behind more advanced models such as GPT-4. Key technical features included Rotary Positional Embeddings (RoPE) for improved sequence handling and a large token vocabulary to support diverse language tasks. March 2024 saw the release of Grok-1.5, which introduced a significantly longer context length of 128,000 tokens -- 16 times greater than its predecessor. This improvement enabled more coherent long-form responses and better handling of complex, multi-step reasoning tasks. Shortly after, xAI launched Grok-1.5V, incorporating multimodal capabilities that allowed the model to analyze and interpret images alongside text. In August 2024, xAI released Grok-2 and Grok-2 mini, improving both processing speed and reasoning capabilities. Benchmarks indicated that Grok-2 outperformed competitors such as Claude 3.5 Sonnet and GPT-4 Turbo in several reasoning and coding tasks. Grok-2 mini was optimized for efficiency, balancing speed and accuracy for general-purpose use. Further updates in December 2024 enhanced Grok-2's processing speed and accuracy while expanding its multilingual capabilities. xAI also integrated Grok with Aurora, an AI-powered image generation tool, and introduced an enterprise API for businesses. In February 2025, xAI launched Grok-3, a family of models trained using significantly greater compute resources via the Colossus supercluster (featuring 200,000 Nvidia H100 GPUs). Grok-3 and its variations, including Grok-3 Reasoning and Grok-3 mini, focused on improving logical reasoning, mathematical problem-solving, and real-time data processing. Early benchmarks suggested that Grok-3 outperformed GPT-4o in specific areas such as mathematical reasoning (AIME), scientific problem-solving (GPQA), and coding (LiveCodeBench). New features included "DeepSearch," a tool designed to provide in-depth web-based research capabilities, and a "Think" mode for more structured reasoning. Grok's models leverage modern AI frameworks such as Kubernetes, Rust, and JAX for efficient scaling. The MoE architecture remains a key differentiator, allowing xAI to optimize performance without significantly increasing computational costs. Training data sources include publicly available web content, X posts, and other structured datasets. As the models continue to evolve, context lengths have expanded significantly, from 8,192 tokens in Grok-1 to 128,000 in Grok-3, enhancing their ability to process lengthy documents and multi-turn conversations. Grok offers a unique conversational style, often incorporating humor and a casual tone in its responses. It is currently available through X Premium+ ($40/month or $395/year) with additional distribution planned through standalone applications for iOS and Android. Features such as real-time event tracking, image generation, and expanded search capabilities provide a range of functionalities for users. However, as with any AI model, challenges remain. Early iterations of Grok faced limitations in accuracy, and its reliance on X for real-time updates has led to discussions about potential biases in its training data. xAI continues to refine its approach, with plans for future iterations and additional open-source releases. Elon Musk has hinted at continued expansion, including Grok-4 and future AI models leveraging even larger compute infrastructure. xAI has also explored AI applications in gaming and other interactive domains. While ethical considerations such as privacy, bias, and misinformation remain central to AI discussions, xAI's pace of development suggests that Grok will remain a notable player in the evolving AI landscape. With ongoing updates and enhancements, Grok continues to establish itself as a distinctive AI assistant, balancing real-time data access with advanced reasoning capabilities. As the field of AI progresses, Grok's trajectory will be shaped by both technological advancements and broader discussions on responsible AI deployment.

Grok 3 appears to be driving Grok usage to new heights | TechCrunch

Elon Musk's AI company, xAI, released Grok 3, its long-awaited flagship AI model, last week. Grok 3 powers the Grok chatbot apps for mobile and the web, as well as the Grok experience on the Musk-owned social network X. Given that there's so much competition in the AI chatbot space these days, it wasn't a foregone conclusion that Grok 3 would make much of an impact. OpenAI's ChatGPT alone has grown to 400 million weekly active users. However, preliminary data suggests that the new model has indeed gotten people to download and try Grok. According to estimates from Sensor Tower, a market intelligence firm, worldwide and U.S. mobile app downloads of Grok during the week of Grok 3's release increased more than 10x each compared to the previous week. Daily active users for Grok's U.S. app soared more than 260% last week, meanwhile, while global daily active users climbed 5x week-over-week. Muddying the waters somewhat is the fact that Grok 3's release coincided with the Grok app's expansion to several markets in Europe, Latin America, and Southeast Asia. Some of the app's global growth is likely attributable to this. Grok's web app also saw growth over the same period, though, independent of the mobile apps. According to digital intelligence platform SimilarWeb, U.S. daily visits to the Grok web app -- to be specific, Grok.com -- increased from around 189,000 to over 900,000 in the days following Grok 3's release. Worldwide, daily visits grew from 627,000 to 4.5 million. They're impressive numbers, to be sure. But the big question is whether xAI can maintain the momentum and retain those users. Recent controversies threaten to dampen enthusiasm for Grok 3. Over the weekend, the model briefly censored certain unflattering mentions of President Donald Trump and Musk, a change that xAI attributed to a rogue employee. A few days earlier, users discovered that Grok 3 would consistently say that President Trump and Musk deserve the death penalty. xAI quickly patched that issue, as well.

Elon Musk launches Grok 3, new xAI chatbot can be used on X or through decicated Grok app

TL;DR: Elon Musk's xAI startup has launched Grok 3, an advanced AI chatbot integrated into X with a dedicated app for iOS and Android. Available to Premium+ subscribers, Grok 3 offers enhanced computing power, a "Big Brain" mode for complex tasks, and a DeepSearch feature. It challenges current AI models and repositions xAI in the open-source LLM race. Elon Musk has announced that his xAI startup has just launched Grok 3, the latest AI chatbot that integrates tightly into X and it even has a dedicated app available on both iOS and Android operating systems. Grok 3 is being rolled out immediately to Premium+ subscribers on X, while we have a new subsciption for xAI is being rolled out called SuperGrok, for users accessing Grok 3 through the mobile app or on the Grok.com website. Grok 3 can generate text and images without the guardrails against sexually aggressive imagery, vulgarity, or the reproduction of well-known people's likenesses. During a livestream sitting with 3 xAI engineers earlier this week, Musk said: "Grok 3 across the board is in a league of its own" . Musk added that Grok 3 far outpaces Grok 2 with "more than 10 times" the computing power, killing it in AI benchmarks across the planet. Musk calls Grok 3 "maximally truth-seeking AI, even if that truth is sometimes at odds with what is politically correct". The SpaceX and Tesla boss says that Grok 3 has a new "Big Brain" mode for those complex research tasks, far and above what a normal AI chatbot can do. Grok 3 features a smart search engine feature called DeepSearch (take that, DeepSeek) which xAI describes as a reasoning-based chatbot capable of articulating its thought process when responding to your queries. Gil Luria, managing director at DA Davidson, said: "The introduction of Grok-3 puts xAI back in the race for leadership in open-source LLMs [large language models]. It outperforms the current state-of-the-art models on some benchmarks, which makes xAI relevant again".

Grok resets the AI race

Alex Heath is a deputy editor and author of the Command Line newsletter. He has been reporting on the tech industry for more than a decade. Just a few weeks after everyone freaked out about DeepSeek, Elon Musk's Grok-3 has again shaken up the fast-moving AI race. The new model is ending the week at the top of the Chatbot Arena leaderboard, while the Grok iOS app is at the top of the App Store, just above ChatGPT. Even as Musk appears to be crashing out from his newfound political power, his xAI team has managed to deploy a leading foundational model in record time. It's one thing to have the leading model; it's another to build the biggest user base around it. Musk seems to understand that if he wants to crush OpenAI, he has to shift attention away from ChatGPT. Since the debut of Grok-3, Musk has said that ChatGPT-like voice interaction and desktop apps are coming soon. Where his product roadmap appears to differ considerably from OpenAI's is xAI's nascent efforts to build an AI gaming studio, though the details there are scarce. While its Deep Research reports are nowhere near as in depth as OpenAI's, Grok-3's "thinking" capabilities appear to be roughly on par with o1, according to Andrej Karpathy, who noted in his deep dive comparison that "this timescale to state of the art territory is unprecedented." Given all this, it's probably not a coincidence that OpenAI decided this week to disclose that ChatGPT now has 400 million weekly active users -- a 33 percent increase from December. Even with its many hooks into X, Grok certainly has far fewer users than that. Now, it's a question of whether OpenAI can maintain its product lead before Grok (and others) catch up. As always, I want to hear from you. Respond here or ping me securely on Signal.

Elon Musk sparks AI rivalry with a meme against ChatGPT and Meta AI as Grok takes centre stage

Remember the penguins from the animated film series of Madagascar? Well, Elon Musk recently shared a meme featuring them to promote Grok AI, his company's chatbot. Elon Musk is often known for his humour and internet culture, and so is his AI. So, his latest meme was adapted from the popular animated film, featuring the iconic penguins, Kowalski, Rico and Private saluting Skipper -- one wearing a captain's cap to depict xAI's Grok's supremacy. His meme hinted that Grok AI could outperform its competitors like Meta's AI, ChatGPT, and Google's Gemini. He just shared the picture of the meme without any caption which garnered about 40.2 million views on X. A user shared their experience with Grok 3 Beta, saying they trained it to understand specific needs and even analyze competitors -- claiming it performed flawlessly while ChatGPT struggled. Another user chimed in, suggesting that Grok might be more powerful than all other AI models combined. A user compared Grok and Copilot by asking both for the score of a soccer match from the previous day. While Grok struggled, stating it was still updating its data, Copilot provided the correct score. However, the user noted that Grok excels when it comes to generating images. This discussion follows Elon Musk's recent launch of Grok 3, an AI model designed for advanced reasoning, in-depth research, and creative problem-solving. Musk emphasized that the goal of xAI and Grok is to deepen humanity's understanding of the universe, even speculating about the existence of extraterrestrial life. While Grok 3 Beta is currently available for free for a limited time, access to its full capabilities will require a paid subscription under the Super Grok plan. Grok is an AI assistant on X that brings a mix of wit and a hint of defiance. Designed to make interactions more engaging, Grok allows users to ask questions, receive answers, and complete various tasks -- all while adding a unique, playful edge to the platform's capabilities.

Behind xAI Grok 3, Elon Musk's 'Maximally Truth-Seeking A.I."

xAI's Grok 3 was built by an elite team of former Big Tech researchers and engineers. On Feb. 18, Elon Musk's xAI unveiled Grok 3, marking the startup's foray into the advanced A.I. reasoning race dominated by players like Google (GOOGL), OpenAI, and China's DeepSeek. In a live stream on X, Musk described Grok 3 as a "maximally truth-seeking A.I." that prioritizes accuracy even when it challenges political correctness and a "major leap forward in reasoning and computational efficiency." Sign Up For Our Daily Newsletter Sign Up Thank you for signing up! By clicking submit, you agree to our <a href="http://observermedia.com/terms">terms of service</a> and acknowledge we may use your information to send you emails, product samples, and promotions on this website and other properties. You can opt out anytime. See all of our newsletters Musk claims Grok 3 outperforms top A.I. models including GPT-4o, Gemini 2 Pro and DeepSeek-V3 on internal evaluations and scored over 1,400 points on LMArena's Chatbot Arena -- an open-source A.I. benchmarking leaderboard developed by UC Berkeley's SkyLab. Grok 3 introduces a new "Think Mode" feature for real-time problem-solving and "Big Brain Mode" for computation-heavy tasks. A standout feature is DeepSearch, an A.I.-powered research tool designed to rival Google Search and A.I.-search alternatives including OpenAI's Deep Research, DeepSeek's Search Mode and Perplexity AI's Pro Search. The A.I. model was built by xAI's elite team of former Big Tech researchers and engineers. Key members include Jimmy Ba, a former student of A.I. pioneer Geoffrey Hinton and Yuhuai "Tony" Wu, a former researcher at Google DeepMind. Leading the engineering effort is Igor Babuschkin, a former engineer at OpenAI, whom Musk personally recruited to build a ChatGPT rival. Babuschkin previously served as X's senior director of engineering. Grok 3 is currently exclusive to members of X Premium+, which costs $40 per month in the U.S. Access to advanced features like DeepSearch and Think Mode reasoning costs an extra $30 per month. What's behind Grok 3' impressive capabilities? According to xAI's blog, Grok 3 leverages Test-Time Compute at Scale (TTCS), a specific implementation of test-time scaling, to power its reasoning. This machine learning strategy enables the A.I. model to dynamically adjust computational resources, ensuring higher accuracy for complex queries while maintaining speed for simpler tasks. Grok 3's TTCS approach could unlock groundbreaking discoveries, including a potential cure for lung cancer if combined with dedicated compute clusters for extended reasoning periods, said Jeetu Patel, chief product officer at Cisco, in a LinkedIn post. "This is the first publicly shared use of Test-time compute cluster at an unprecedented scale, with a reasoning model that is multi-modal and can consume real-time data," he wrote. Moreover, Grok 3's development involved Colossus, a massive supercomputer cluster built by xAI in Memphis, Tenn. The system houses 200,000 Nvidia H100 GPU accelerators. xAI utilized 100,000 Nvidia H100 GPUs from Colossus to train Grok 3, which delivered 200 million GPU hours, a tenfold increase over the training setup used for Grok 2. Musk said last year the next generation of Colossus training cluster will be five times more powerful. Some are skeptical about the sustainability of xAI's rapid progress, Musk's lofty claims about Grok 3, and its ability to compete with A.I.-search rivals like OpenAI's Deep Research. "Scaling up computing power raises costs and makes model deployment too expensive, it is an unsustainable approach for business in the long run," Lin Qiao, a former senior engineering director at Meta and now CEO of the A.I. infrastructure platform Fireworks AI, told Observer. "Models absorb everything they're trained on, including biases. I don't believe an A.I. can ever be truly neutral," Inna Tokarev Sela, a former machine learning executive at SAP and now CEO of the data intelligence platform Illumex, told Observer. She noted that Grok 3 can match Google Search on general topics -- if xAI keeps accelerating training. "When it comes to specialized topics, Google Search doesn't have an edge over domain-specific A.I. models (Grok), trained on focused data," she said. Likewise, early testing of Grok 3 by Andrej Karpathy, a founding member of OpenAI and former senior director of A.I. at Tesla, suggests that while DeepSearch is better than Google's Gemini models, it struggles with occasional hallucinations of citations and URLs. "The impression I get of DeepSearch is that it's approximately around Perplexity DeepResearch offering, but not yet at the level of OpenAI's recently released 'Deep Research,' which still feels more thorough and reliable," he wrote in an X post.

Grok 3 vs. DeepSeek: Elon Musk's xAI Unveils the Smartest AI Yet

Elon Musk's xAI Launches Grok 3, Challenging DeepSeek and ChatGPT in AI Supremacy Elon Musk's xAI has launched Grok 3, claiming it to be the most intelligent AI system on the planet. Designed to outperform both ChatGPT and DeepSeek, boasts enhanced logical reasoning, advanced coding capabilities, and superior scientific knowledge. With synthetic data training and a 10X increase in computational power compared to Grok 2, xAI aims to revolutionize the AI landscape. But does Grok 3 truly live up to Musk's bold claims? Musk introduced Grok 3 to X users by declaring it the planet's most intelligent artificial intelligence system. The model is said to be 10X more powerful than its previous model Grok 2 due to increased computational capacity. xAI developers also stated that early evaluations and standardized testing place Grok 3 ahead of its competitors.

Grok 3 model puts xAI at the top tier of frontier model developers

Welcome to AI Decoded, Fast Company's weekly newsletter that breaks down the most important news in the world of AI. You can sign up to receive this newsletter every week here. In just two years, Elon Musk's xAI has become one of a dozen or so labs capable of developing state-of-the-art AI models. Now xAI is out with its Grok 3 large language model, which beats state-of-the-art frontier models, such as OpenAI's GPT-4o and DeepSeek's V3, in common mathematics, science, and coding benchmarks by a wide margin. Meanwhile, the smaller Grok 3-mini performs at par with the larger competing models. The new Grok model reportedly was trained using unprecedented computing power -- first with a cluster of 100,000 Nvidia H100 GPUs. A small group of rival developers have been testing an early version of Grok 3, and most say they're impressed, with some caveats. OpenAI cofounder Andrej Karpathy posted on X that Grok 3 exhibited sharp reasoning skills and was able to resolve some complex problems. He estimates that the model is on par with OpenAI's o1-Pro reasoning model and slightly better than DeepSeek-R1 and Google's Gemini 2.0 Flash Thinking. However, he did find that Grok 3 choked on some prompts commonly known to give large transformer models trouble, such as determining how many Ls are in "Lollapalooza," for example. Scale AI CEO Alexandr Wang posted on X that Grok 3 is a state-of-the-art model and gives it props for achieving the top spot on the Chatbot Arena benchmark. Whereas AI skeptic Gary Marcus, who also posted on X, said that while Grok 3 shows real progress, it doesn't represent a significant leap beyond existing models.