Elon Musk is not one to take a seat out a tech development. In November 2023, he launched Grok, an AI chatbot created by his synthetic intelligence startup, xAI. Musk, who co-founded OpenAI earlier than departing in 2018, created Grok to compete with ChatGPT, Claude, Perplexity and Google’s Gemini. But he gave it a sarcastic twist and direct entry to X (previously Twitter), the social platform he bought in October 2022.Initially, Grok was accessible completely to paying X Premium Plus subscribers. By December of 2024, xAI rolled out a restricted free tier, permitting you to make 10 queries per day. Now, it’s also accessible through internet browser, as standalone cellular iOS, Android and desktop apps, that means you possibly can entry Grok with out an X subscription. It must also quickly be accessible to Telegram customers, per Telegram CEO Pavel Durov’s announcement on X.On July 10, TremendousGrok entered the scene. xAI launched Grok 4 and Grok 4 Heavy in a livestream that felt extra like a late-night show-and-tell than a typical Silicon Valley polished product launch — and it got here filled with some daring claims.So let’s dive deeper into Grok’s benefits, controversies and what units it other than different generative AI instruments.The background storyGrok was developed by xAI, an AI startup based by Musk in July 2023. The firm’s said mission is “to understand the universe,” and it maintains shut ties with Musk’s different ventures like X, Tesla and Neuralink.Unlike ChatGPT, which began with static coaching knowledge and later added internet searching, Grok was constructed to be stay from the beginning. It was impressed by The Hitchhiker’s Guide to the Galaxy, promising customers a chatbot that solutions with wit, humor and perspective. On its web site, xAI emphasizes that Grok goals to supply correct data, although its edgy, less-moderated strategy with a “rebellious streak” generally blurs the road between helpful and controversial. This unfiltered “free speech” type has led to backlash, with Grok being recognized to sometimes generate biased or offensive content material — like final week’s “MechaHitler” controversy (extra on that under). And, like all the opposite AI chatbots, it warns you that it will probably generally hallucinate.Its defining function is real-time integration with X, which implies it will probably pull from trending posts, person conversations and breaking information quicker than different bots. The firm says it might use your X knowledge and interactions with Grok for mannequin coaching, fine-tuning and to personalize your Grok expertise. But you possibly can decide out utilizing your X settings. What can Grok do?Like most AI chatbots, Grok can reply questions, summarize articles, write emails, clarify advanced matters and assist with coding. If you are on X and see a submit you need extra information about, you possibly can ask Grok instantly within the feedback. You also can DM it such as you’d message a buddy. It pulls in real-time context from X and internet sources (although it would not at all times cite them clearly), regularly adopting an informal, humorous or sarcastic tone paying homage to Musk’s posts. Meta AI is built-in throughout its socials equally, however Grok excels at this. None of the opposite main chatbots provide that form of native integration with a social media community.You also can ask it to generate photos utilizing a built-in AI picture instrument. Perplexity and ChatGPT depend on OpenAI’s fashions for that half, however since Grok 3, the corporate makes use of its personal mannequin, Aurora. Previous Grok fashions used Black Forest Labs’ Flux mannequin till December 2024.Grok additionally helps math explanations, brainstorming concepts and textual content translation. For builders, there’s Grok 4 Code, a specialised mannequin designed to write down, debug and clarify code extra effectively, much like instruments like GitHub Copilot or GPT-4 Code Interpreter.”You can cut & paste your entire source code file into the query entry box on http://grok.com and Grok 4 will fix it for you!” Musk tweeted earlier in July.Another functionality with Grok 4 is enhanced cultural fluency. A current replace makes Grok adept at decoding memes, web slang and humor, doubtlessly making it probably the most “online” AI assistant but.Grok 4 and Grok 4 Heavy fashions and pricingDuring the livestream, Musk claimed Grok 4’s capabilities exceed these of people in academia. “Grok 4 is postgrad-level in everything,” Musk mentioned. “At least with respect to academic questions, Grok 4 is better than a PhD level in every subject. No exceptions.” John Licato, assistant professor of laptop science and engineering on the University of South Florida, says these claims are completely deceptive. “‘PhD-level’ is being thrown around as a marketing term with no real rigorous benchmarks behind it. It also completely misunderstands what PhDs actually do,” Licato advised CNET.xAI did not instantly reply to a request for clarification or remark.Grok provides three tiers. The Basic plan is free and consists of restricted entry to Grok 3, fundamental options like tasks and duties and restricted tokens (constructing blocks of textual content that AI fashions use to course of language). The TremendousGrok plan prices $30 per thirty days and unlocks elevated entry to each Grok 3 and Grok 4. It consists of 128,000 tokens, context reminiscence and extra options like voice with imaginative and prescient. For instance, when you present Grok a photograph or level a digicam at one thing and ask about it out loud, it can analyze the picture and generate a response.Then there’s TremendousGrok Heavy, priced at $300 per thirty days, with entry to Grok 4 Heavy (presently in unique preview), early entry to new options and devoted assist. Grok 4 Heavy concurrently deploys a number of AI brokers — which course of duties independently, work in parallel and cross-evaluate their outputs — to collaborate on a question, evaluating outcomes to ship probably the most correct response.If you are on X, you possibly can freely entry Grok 3 or mannequin 4 you probably have an X Premium Plus subscription for $40 per thirty days.Where Grok does and would not excelBecause of all of the claims Musk and his workforce made about Supergrok and its purported superior scientist-grade reasoning, I made a decision to do some testing. I requested ChatGPT’s o3 superior reasoning mannequin and Grok 4 the identical “Koch snowflake” query (picture under). Both chatbots reached the identical mathematical outcomes, however Grok 4 supplied a extra detailed, step-by-step clarification extra appropriate for studying, whereas ChatGPT supplied a extra concise, streamlined model, higher for fast reference. ChatGPT was additionally 5 seconds quicker. ChatGPT 3o on the left, Grok 4 on the proper. ChatGPT / Grok / Screenshot by CNETThen I examined the picture technology capabilities. ChatGPT gave a extra reasonable picture, whereas Grok’s resembled these typical Midjourney photos which are extra uncanny and form of shiny. However, Grok produced two photos and extra prompts on the underside so I may enrich or change the picture — like watercolor type, and wet alley — and it generated the photographs in 21 seconds, whereas it took over a minute and a half for ChatGPT to make one picture. ChatGPT’s pet on the left, and Grok’s on the proper. ChatGPT / Grok / Screenshot by CNETTone is one other differentiator. While ChatGPT, Perplexity and Claude are typically impartial or overly well mannered, Grok is deliberately casual. Sometimes it is useful, generally it is snarky, and generally it simply goes too far. xAI pitches this as a bonus: a bot that is not afraid to “tell you the truth,” even when it is controversial. However, that strategy bought Grok in scorching water simply over per week in the past with a collection of now-deleted antisemitic posts on X, after which Grok was quickly shut down.”We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts. Since being made aware of the content, xAI has taken action to ban hate speech before Grok posts on X. xAI is training only truth-seeking and thanks to the millions of users on X, we are able to quickly identify and update the model where training could be improved,” Grok tweeted on July 8.Grok’s real-time entry to X is each a energy and a danger, because the platform is thought for internet hosting unmoderated or deceptive content material. Because of its reliance on X, Grok could amplify misinformation or replicate the platform’s person biases greater than fashions skilled on broader knowledge. When I requested Licato if he discovered that regarding, he agreed and added that he is not satisfied Grok or the management behind it’s doing something to make sure accountable use of its know-how. “Grok is marketed as more of a truth-teller than its competitors, and this is leading to a lot of people believing that it is somehow more reliable,” Licato advised CNET. “But I do not see any good evidence that this is actually true, and I worry that, since so much of the public does not know how easy it is to get LLMs to spout false information, they may just be aiding in the spreading of mis- and disinformation.”Michael Berthold, CEO and co-founder of AI analytics firm KNIME, advised me he is involved in regards to the broader implications of AI chatbots rewriting or summarizing information. “The worry I have, and I see it with my own kids, is how the next generation will be learning things,” Berthold advised CNET. “There will be so much content that’s not going to be consumed in the original anymore, but in some sort of AI-rewritten, summarized form. Something can be lost, either accidentally or intentionally. We already see AI systems from different countries reflecting slightly different preferences.”Grok’s tone has additionally raised eyebrows. Some responses are laced with sarcasm or off-color jokes, which could enchantment to Musk’s followers however may flip off individuals anticipating a impartial assistant.What’s subsequent for Grok?Grok has been considerably of an ignored AI chatbot, because it’s confined to X. But as xAI continues to develop it, and Musk has additionally hinted that Grok is coming to Tesla autos as early as subsequent week, at this level, its future seems bold. It’s protected to imagine Grok’s integration into future variations of Optimus, Tesla’s humanoid robotic, as nicely.Grok is carving out a singular house within the AI arms race, particularly with its provocative digital character. With speedy developments within the AI sphere, skepticism over this know-how is rising, too.
More