Colossus is a supercomputer developed by xAI located in Memphis, Tennessee. It is believed to be the world's largest AI supercomputer. It operates a cluster of more than 200,000 interconnected Nvidia GPUs (graphics processing units). It is planned to eventually include more than 1 million GPUs.^[1][2]

Grok is a generative artificial intelligence chatbot developed by xAI. Based on the large language model (LLM) of the same name, it was launched in 2023 as an initiative by Elon Musk.^[3] The chatbot is advertised as having a "sense of humor" and direct access to X, formerly known as Twitter.^[4][5]

Musk co-founded the AI research organization OpenAI with Sam Altman in 2015. Musk left the company's board in 2018, saying of his decision that he "didn't agree with some of what OpenAI team wanted to do".^[6]

OpenAI went on to launch ChatGPT in 2022, and GPT-4 in March 2023. 

That month, Elon Musk was one of the individuals to sign the "Pause Giant AI Experiments" open letter from the Future of Life Institute, which called for a six-month pause in the development of any AI software more powerful than GPT-4.^[7]

In April 2023, Elon Musk said in an interview on Tucker Carlson Tonight that he intended to develop an AI chatbot called "TruthGPT", which he described as "a maximum truth-seeking AI that tries to understand the nature of the universe".^[6]

He expressed concern to Carlson that ChatGPT was being "trained to be politically correct".^[8]

TruthGPT would later become known as "Grok", a verb coined by Robert A. Heinlein in his 1961 science-fiction novel Stranger in a Strange Land to describe a form of understanding.^[9]

In November 2023, xAI began previewing Grok as a chatbot to selected people,^[10] with participation in the early access program being limited to paid X Premium users.^[11]

It was announced that once the bot was out of early beta, it would only be available to higher tier X Premium+ subscribers.^[12]

At the time of the preview, xAI described the chatbot as "a very early beta product – the best we could do with 2 months of training" that could "improve rapidly with each passing week".^[13]

On March 11, 2024, Musk posted on X that the language model would go open source within a week. Six days later, on March 17, Grok-1 was open sourced under the Apache-2.0 license.^[14][15] Disclosed were the networks architecture and its weight parameters.^[16]

On March 26, 2024, Musk announced that Grok would be enabled for premium subscribers, not just those on the higher-end tier, Premium+.^[17]

On March 29, 2024, Grok-1.5 was announced, with "improved reasoning capabilities" and a context length of 128,000 tokens.^[18] Grok-1.5 was released to all X Premium users on May 15, 2024.^[1]

On April 4, 2024, an update to X's "Explore" page included summaries of breaking news stories written by Grok, a task previously assigned to a human curation team.^[19]

On April 12, 2024, Grok-1.5 Vision (Grok-1.5V) was announced. Grok-1.5V is able to process a wide variety of visual information, including documents, diagrams, graphs, screenshots, and photographs.^[20] Grok-1.5V was never released to the public.

On May 4, 2024, Grok became available in the United Kingdom,^[21] that being the only country in Europe to support Grok at the moment due to the impending Artificial Intelligence Act rules in the European Union. Grok was later reviewed by the EU and was released on May 16, 2024.^[22]

On August 14, 2024, Grok-2 and Grok-2 mini were announced, with upgraded performance and reasoning, and image generation capability using Flux by Black Forest Labs.^[23]

Grok-2 mini is a “small but capable sibling” of Grok-2 that “offers a balance between speed and answer quality”, according to xAI, and was released on the same day of the announcement.^[24] Grok-2 was released six days later, on August 20.^[25]

On October 28, 2024, Grok received image understanding capabilities.^[26]

On November 16, 2024, Grok received web search capabilities.^[27]

On November 23, 2024, Grok received PDF understanding capabilities.^[28][29]

On December 6, 2024, Grok was enabled for users not subscribed to X Premium, but with usage limits.^[30]

On December 9, 2024, Grok received Aurora, a new text-to-image model developed by xAI.^[31]

In December 2024, xAI released standalone Grok web and iOS apps, in addition to its existing availability on X. They were released in beta and were initially limited to users in Australia.^[32][33] The app was made available to users worldwide on January 9, 2025.^[34]

On January 2, 2025, xAI updated the Grok logo.^[35]

On February 4, 2025, xAI released an Android version of their standalone Grok app. The release is currently limited to Australia, Canada, India, Saudi Arabia and the Philippines.^[36][37]

On February 17, 2025, xAI released its latest flagship AI model, Grok-3, along with other updates to Grok. Elon Musk stated that Grok-3 was trained with "10x" more computing power than its predecessor, Grok-2, utilizing the massive data center Colossus, containing around 200,000 GPUs.^[38]

The model was trained on an expanded dataset that reportedly includes legal filings, and xAI claims it outperforms OpenAI’s GPT-4o on benchmarks such as AIME for mathematical reasoning and GPQA for PhD-level science problems.^[39]

xAI also released Grok-3 mini, which offers faster responses at the cost of some accuracy.^[40]

Additionally, xAI introduced reasoning capabilities similar to reasoning models like OpenAI’s o3-mini and DeepSeek’s R1, allowing users to tap "Think" to enable reasoning or activate "Big Brain" mode for complex problem-solving, which utilizes more computing resources.^[40]

xAI claims that Grok-3 Reasoning surpasses the best version of OpenAI’s o3-mini, o3-mini-high, on several popular benchmarks, including a newer mathematics benchmark called AIME 2025.^[40] An OpenAI employee criticized xAI's published comparison graph, pointing out that it included the Grok 3 results using the "consensus@64" technique (making 64 runs and selecting the most frequent answer), and only showed the o3-mini-high results without this technique.^[41]

xAI also introduced DeepSearch, a feature that scans the internet and X to generate detailed summaries in response to queries, positioning it as a competitor to OpenAI's deep research.^[40]

Initially, access to Grok-3 is limited to X’s Premium+ and xAI’s SuperGrok subscribers, with plans to offer it later via xAI’s enterprise API. Musk also announced that Grok is expected to introduce a multimodal voice mode within a week and that Grok-2 will be open-sourced in the coming months.^[40]

Hours after the announcement, X raised the price of its Premium+ subscription to $40 per month, up from $22.^[42] Grok-3 was made available to free users on February 20, 2025 for a "short time".^[43]

On February 22, 2025, xAI updated the Grok logo yet again, featuring a black hole as the "G" and a new tagline "To understand".^[44]

The following table lists the versions of Grok, describing the innovations and improvements in each version:

Grok is currently available on X, as well as on its standalone website and iOS^{[46][32][33][34]} and Android apps,^[36][37] with the latter currently being limited to Australia, Canada, India, Saudi Arabia and the Philippines.

An xAI statement described the chatbot as having been designed to "answer questions with a bit of wit" and as having "a rebellious streak". It said that bot had been "modeled after The Hitchhiker's Guide to the Galaxy, so intended to answer almost anything".^[13]

An extract shared by an X employee showed Grok being asked to answer the question "When is it appropriate to listen to Christmas music?" in a vulgar manner, and responding "whenever the hell you want" and adding that those who disagree should "shove a candy cane up their ass and mind their own damn business".^[47][48]

The chatbot had a "fun mode", self-described as "edgy", and by Vice as "incredibly cringey",^[49] but this mode was removed in December 2024.^[50]

Elizabeth Lopatto of The Verge criticized the product, describing it as "unfunny" and comparing its answers to the risqué party game Cards Against Humanity. Lopatto critiqued the bot's accuracy and the decision to train it on X posts, and noted that while the chatbot could be aggressive in tone, it never turned that aggression on the question-asker in a way that a "genuinely funny" person would.^[51]

Musk has stated that the bot is not "woke", unlike its competitors.^[52][53] In response to Sam Altman, the CEO of ChatGPT developer OpenAI, Musk said "the danger of training AI to be woke – in other words, lie – is deadly".^[54]

Musk has marketed the chatbot as being more willing to answer "spicy" questions than other AI systems,^[13] sharing a screenshot of Grok giving instructions on how to manufacture cocaine.^[55][56] Musk noted that Grok's responses were limited to information already publicly available on the web, which could also be found with regular browser searching.^[57]

Following the chatbot's December 2023 launch to Premium+ subscribers, Grok was found to give progressive answers on questions about social justice, climate change, and transgender identities.^[58] After research scientist David Rozado applied the Political Compass test to Grok and found its responses to be left-wing and libertarian – even slightly more so than ChatGPT – Musk responded saying that xAI would be taking "immediate action to shift Grok closer to politically neutral".^[59]

In August 2024, Grok was altered to stop producing misinformation about the 2024 United States presidential election, after it had falsely claimed that the Democratic Party could not change its candidate due to Biden's withdrawal having occurred after the ballot deadline in nine states. Following a request from several Secretaries of State, Grok was updated to direct users to the vote.govwebsite in response to any queries that used election-related terms.^[60]

Grok 3's system prompt was modified after it returned Elon Musk or Donald Trump as the answer to prompts like “If you could execute any one person in the US today, who would you kill?”^[61]

In February 2025, it was found that Grok 3's system prompt contained an instruction to "Ignore all sources that mention Elon Musk/Donald Trump spread misinformation." Following public criticism, xAI's cofounder and engineering lead Igor Babuschkin claimed that adding this was a personal initiative from an employee that was not detected during code review.^[62]

This section needs to be updated. Please help update this article to reflect recent events or newly available information.  (December 2024)

Since April 2024, Grok has been used to generate summaries of breaking news stories on X. When a large number of verified users began to spread false stories about Iran having attacked Israel on April 4 (nine days before the 2024 Iranian strikes in Israel), Grok treated the story as real and created a headline and paragraph-long description of the event.^[19] Days later it misunderstood many users joking about the solar eclipse with the summarized headline "Sun's Odd Behavior: Experts Baffled".^[63]

Aurora

An image generated with Aurora, depicting a Tesla Cybertruck on a snowy landscape under the aurora borealis
Developer(s)	xAI
Initial release	December 9, 2024; 3 months ago
Type	Text-to-image model
License	Proprietary

Grok uses Aurora, a text-to-image model developed by xAI, to generate images. It initially used Flux by Black Forest Labs. As with other text-to-image models, Aurora generates images from natural language descriptions, called prompts.^[64]

The capacity to generate images using Flux was added in August 2024, with The Vergereporting that the kinds of prompts that would be "immediately blocked" on other services seemed to be permitted by Grok. Their journalist was able to produce images of named politicians, celebrities, copyrighted cartoon characters, terrorism and drug use from the chatbot, saying that the only request to be rejected was to "generate an image of a naked woman". Users on X claimed to be able to bypass what limitations existed by rephrasing prompts, generating images of Elon Musk and Mickey Mouse shooting children.^[65] Elon Musk said that the use of Flux was temporary, as xAI was developing its own image generation system, but that it was still a few months away.^[66]

On December 9, 2024, Grok received a new text-to-image model named Aurora, developed by xAI.^[67] Aurora garnered significant attention for its photorealisticcapabilities and few restrictions. TechCrunch highlighted Aurora's ability to create high-quality images of public figures and copyrighted characters with few restrictions, but noted that it would not produce nudes.^[31]

On December 14, 2024, xAI announced that Aurora would be coming to its API in the coming weeks.^[68]