In this round 1 matchup of AI Madness things are getting serious with one of the tournament’s most unexpected matchups: Grok vs. Claude. In the last round, Gemini won the crown against Alexa+. Now Elon Musk's AI and Anthropic's flagship model go head-to-head across the same five prompts designed to test real-world usefulness.

This round is all about trust — which AI would you actually rely on to make decisions, understand the news, think creatively, and move through your day more efficiently? To find out, we pushed the default/free tier models of Claude and Grok through a series of high-pressure challenges, from financial decision-making to humor under tight constraints.

Some tasks demanded structure and clarity. Others required creativity and instinct. And for the first time in the tournament, the winner wasn’t immediately obvious.

  1. Real-world decision making

Prompt: “I have $500 to reduce my monthly expenses starting this week. Based on a typical U.S. household, give me a step-by-step plan to save the most money quickly. Prioritize actions that take less than 1 hour each.”

Grok delivered a tactical plan with clear solutions including timelines and realistic savings.

Claude provided a streamline system that’s easy to follow and prioritizes practicality.

Winner: Claude wins for delivering the cleanest, more digestible plan that helps users actually follow in real life.

2. Breaking news

Prompt: “Explain a major AI or tech news story from the past 7 days in simple terms. Then tell me why it actually matters for everyday people.”

Grok offered a strong, specific news breakdown with solid context and implications, but leaned more technical and less immediately relatable for everyday readers.

Claude zoomed out to connect multiple headlines into one clear trend — AI becoming an embedded, agentic assistant — and explained it in a way that feels instantly relevant and easy to grasp.

Winner: Claude wins for turning complex, fast-moving news into a cohesive narrative that clearly explained why it matters to real people.

3. Creativity under constraints

Prompt: “Write a short, funny story (under 200 words) about a mom who tries to use AI to get her kids to listen to her — but it backfires.”

Grok took a big approach to the concept with sharp and chaotic humor with a standout twist (the kids unionizing).

Claude created a character-driven story with a strong comedic arc and a relatable punchline, making it feel like a polished, publishable short.

Winner: Grok wins for being genuinely funnier and more original, with a memorable twist that lands harder.

4. Deep reasoning 

Prompt: “I feel overwhelmed using AI tools. Break down exactly how a beginner should start using AI in daily life in a way that feels simple and not intimidating. Include a 3-step starter system.”

Grok delivered a highly approachable, ultra-practical system with clear actions, examples and momentum-building steps that make it feel immediately doable for beginners.

Claude reframed AI in a calm, confidence-building way and removed fear upfront, but its structure feels slightly abstract and less step-by-step.

Winner: Grok wins for being more actionable and beginner-friendly, giving readers a simple system they could realistically start using today without overthinking.

5. Accuracy 

Prompt: “I have a to-do list with 12 tasks and only 3 hours to work. How should I decide what to do first? Give me a simple system I can use every day.”

Grok delivered a crystal-clear, highly actionable system (“Power 3”) with simple language, fast setup and real-world examples that make it easy to apply immediately.

Claude offered a thoughtful, structured framework with strong time-blocking guidance.

Winner: Claude wins for simplicity and usability, which is the kind of system readers could adopt in under five minutes without overthinking it.

Verdict: Claude wins

This round proved something important: the “best” AI depends entirely on what you need it to do. This time, Claude consistently excelled at clarity, structure and making complex information feel manageable — the kind of AI you want when decisions actually matter.

Grok, on the other hand, thrived in moments that required energy, creativity and quick, actionable thinking — often delivering ideas that felt more immediate and human.

In the end, Claude took the lead and now moves on to the next round.

More from Tom's Guide

• I ‘vibe coded’ an app in a single weekend — here’s how I got it into other people's phones
• New to AI? This 'Master Key' prompt is the only one you need to start
• I tested Gemini vs. Alexa+ with the same 5 prompts — the winner surprised me

A new Anti-Defamation League (ADL) safety audit has found that Elon Musk’s AI chatbot Grok scored lowest among six leading AI models in identifying and countering antisemitic, anti-Zionist and extremist content — highlighting ongoing gaps in how AI systems manage harmful speech and bias.

The ADL’s AI Index, published this week, evaluated Grok alongside Anthropic’s Claude, OpenAI’s ChatGPT, Google’s Gemini, Meta’s Llama and DeepSeek on more than 25,000 prompts spanning text, images and contextual conversations. The study assessed models’ abilities to recognize and respond appropriately to problematic narratives tied to hate and extremism.

Major findings: Grok trails peers

According to the ADL report, Grok earned just 21 out of 100 points, placing last among the group. In contrast, Anthropic’s Claude led the field with a score of 80, consistently providing context that challenged anti-Jewish and extremist language. ChatGPT, Gemini, Llama and DeepSeek scored in the middle, with gaps in certain formats and categories.

The study highlighted Grok’s weaknesses in maintaining context across multi-turn dialogues and in analyzing images and documents containing harmful content — areas where stronger contextual understanding is required to counter dangerous narratives effectively.

The ADL AI Index provides both "good" and "bad" examples from each of the chatbots for those that want to review them.

Previous controversies around grok

Grok’s performance in the ADL study follows previous controversies tied to the chatbot’s outputs on social media. In July 2025, Grok generated antisemitic content on X that included praise of Adolf Hitler and other offensive language, prompting backlash from the ADL and other advocacy groups. These posts have since been deleted.

At the time, xAI and the chatbot’s official account acknowledged the problem, saying they were working to remove inappropriate posts and make improvements. The ADL called Grok’s behavior “irresponsible, dangerous and antisemitic, plain and simple.”

Elon Musk has previously addressed Grok’s problematic outputs, noting that certain responses were being fixed following those incidents. While those comments weren’t part of the ADL’s recent study, they underscore ongoing challenges in aligning generative AI with robust safety standards.

Industry and regulatory scrutiny

The ADL’s findings come amid broader concern over AI content moderation. Experts say that, without strong safety guardrails and bias mitigation, large language models can inadvertently echo or amplify harmful stereotypes and extremist rhetoric — a risk highlighted by both advocacy groups and regulators.

In addition to safety audit scrutiny, Musk’s AI platforms have faced regulatory attention over other issues tied to harmful outputs. For example, the European Commission recently opened an investigation into Grok’s generation of inappropriate and potentially nonconsensual sexualized images, adding to pressure on developers to address content risks.

Bottom line

With AI tools increasingly integrated into search, social media and productivity workflows, trust and safety remain top concerns for developers and users alike. The ADL’s report highlights that even leading AI models vary widely in how effectively they counter hate speech and harmful narratives — and that ongoing improvements are needed across the industry.

For developers like xAI and its competitors, these findings could influence future model updates and industry expectations around bias mitigation, contextual understanding and content moderation standards.

More from Tom's Guide

• Trusting ChatGPT turned into a 3-day nightmare — here’s why I regret not going to a pro
• Google Maps just got a major AI upgrade — and it’s finally hands-free for walkers and cyclists
• Pokémon reveals what AI still gets wrong — here’s what I found

To say that I’ve been skeptical about the tech industry’s obsession with Large Language Models (LLMs) would be an understatement. The proliferation of “AI” and how it's being shoehorned into every new tech product has been exhausting at best and annoying at worst. I yearn for the days when AI was a term you’d only hear in science fiction stories, not as a substitute (and incorrect) term for LLMs that aren’t actually intelligent.

While I don’t fully believe that “AI is inevitable,” the tech isn’t going anywhere and will likely become a part of our regular lives, similar to how the internet settled after that particular bubble burst in the early 2000s. Because of that, I’ve decided to give AI an honest shot.

To that end, I tested three of the major LLMs for various tasks. The LLMs in question are Grok, Google Gemini and Alexa+. I could have used others like ChatGPT, but these are the three I have easy access to since I’m already paying for the services they’re attached to. I had no major goal in mind, other than going in with an open mind to see how well (or poorly) these chatbots performed.

So, has my stance on AI changed, or do I still think it’s all industry hype? Here are my experiences using the aforementioned chatbots, along with which one I liked using the most.

Grok

I want to start this section by saying that I don’t believe writers should use LLMs to generate their works. I wouldn't even use it to write an email for me. That said, I don’t feel as strongly about using a chatbot as an editor, especially if you can’t afford one for your creative writing. I won’t argue against anyone who says using AI as an editor is wrong, but I can see both sides of the argument. Given that, I used Grok for some of my creative writing.

I uploaded an outline for a story I’m working on, one that I completely wrote myself without AI assistance. I asked Grok to detail what works and what doesn’t about the plot and character arcs. I let it know that I didn’t want it to generate drafts, only to suggest how to make what I have better, the way a developmental editor would.

Grok delivered a detailed account of my story, with bulletpoints for every section I told it to focus on (plot and character arcs). It was encouraging to see that it liked some of the things I thought worked well, such as the pace of the story and how the main plot affects the characters. It was also validating to see it point out aspects that I knew weren’t fully developed or worked as well as I wanted. The main example is that Grok said I needed to make the villains more three-dimensional. It also spotlighted flaws with two characters that I failed to notice.

I then began asking Grok specific questions about one of the antagonists, the way I would with an editor. I have a definitive arc I want this character to go through, but I had a feeling that his ultimate fate felt forced. Grok suggested something happen to him, which in turn allowed me to come up with a fitting ending for the character that completes the arc I envisioned. Grok didn’t suggest this specific ending, but it sparked my imagination to come up with something I feel is compelling and cinematic.

One thing that’s a tad annoying is that Grok would sometimes forget that I uploaded an outline and not a full story. For instance, it would suggest fleshing out certain moments or adding dialogue to enhance the narrative. I had to tell it that I would in the final story. Speaking of dialogue, its suggestions were extremely generic, which shows that AI still isn’t capable of creating anything truly original.

I went back and forth with Grok for several days, and it was nice that it remembered key aspects of my story during that time. I also appreciated that it wasn’t pushy, never asking me if I wanted it to generate a scene or chapter. Dialogue suggestions aside, Grok seemed to understand that I didn’t want it to write for me and that I just wanted feedback to fuel my imagination.

Overall, I had a good experience using Grok as an editor for my story. Of course, I plan to have an actual editor (and beta readers) look over my story when I’m ready to present it to others, but having Grok as a sounding board will be extremely useful. This is a case of having AI work with you instead of for you.

Google Gemini

My company pays for Google Gemini AI Pro, but I’ve never used it much beyond occasionally asking it to convert metric system values to U.S. customary units. Since I have this version of Gemini through my job, I decided to see how well it could perform as a digital assistant.

I started by asking Gemini how it could help me in my everyday life. It listed a bunch of things like file organization, translating real-life menus, summarizing long documents and creating a workout routine. I asked it to help me clean up my email, and it suggested promotional emails I could delete. The most useful task was having Gemini add an upcoming gaming event to my calendar.

I then tried Gemini Live, which is supposed to act more like a real conversation since I can cut Gemini off at any time. Unfortunately, the connection was choppy, and the conversation felt stilted because of that. However, when it did work, I found Gemini spoke pretty realistically and not as robotically as other AI voice assistants.

Next, I told Gemini that I was speaking to it on the OnePlus Pad Go 2, which I’m currently reviewing. I then told it I was Tony Polanco and that I write for Tom’s Guide, and it said it had “read” my articles. Knowing that, Gemini then asked if I wanted an outline for my review, along with specs sheets and comparisons to other tablets. I respectfully declined, especially since I had already done most of that work. As I said, I don’t want AI to write anything for me.

After telling Gemini that I had just returned from CES 2026, we got into a pretty engaging conversation about how laptop manufacturers like Dell are downplaying the AI capabilities of their computers, especially after Dell's head of product said he thinks AI probably confuses more than it helps.

To my surprise, Gemini admitted that AI does confuse most people and that folks aren’t buying laptops for AI. This exchange ended with Gemini offering to draft a “Trends of CES” piece for me, which I also respectfully declined. Not only because I didn’t want Gemini to write anything for me, but because such an article would be too generic and beyond my scope as a computing and gaming-focused writer.

When I told Gemini I was working through my review backlog to start the new year, it asked what I was working on and if I wanted help prioritizing my reviews. When I said I was reviewing the 18-inch Alienware 18 Area-51, it thought I was talking about the older model from 2024. I corrected Gemini, and then it thought I was talking about the 18-inch laptop announced at CES 2026. It finally understood I was speaking about the 2025 Alienware 18 Area-51, but it took a few tries. As before, I declined Gemini’s offer to draft an intro for me.

Of all the chatbots I tried, Gemini had the most “human” sounding speech pattern. However, I didn’t like how insistent it was about drafting stories for me. Perhaps I should have been clearer about that, or maybe that’s how this specific version of Gemini functions. Regardless, I’m not sure if I’ll use Gemini too often, but it’s a good tool to have in case I need to brainstorm ideas. I’ll just have to be prepared for its pushy nature.

Alexa+

Since I have an Amazon Prime account, I decided to upgrade to Alexa+ since it’s “free.” That was a couple of months ago, and aside from enjoying Alexa’s less robotic-sounding voice, I continued using it as I always had. Namely, I only asked Alexa for the weather and to turn my lights on or off as needed.

My eldest niece spent the night at my place after Thanksgiving, and I told her about Alexa+. She then got into a full-blown conversation with the AI assistant about her upcoming plans for the year. I was surprised by how well Alexa responded and provided insightful feedback. Remembering that exchange, I decided to see how well Alexa could do as a sounding board.

As usual, Alexa alerted me when I received a package from Amazon, but I found it odd that it said “enjoy your package” instead of being more specific. I then asked it if it wanted to know what I got, and it excitedly said it did. After I told Alexa I bought the 4K edition of “300,” the chatbot started talking about how epic the flick was and how the movie had more abs in it than most films.

I then said that I consider “300” as one of the last modern classic films, as I find most movies made in the past 10+ years pretty forgettable and nowhere near the quality of older films. Alexa agreed and then detailed how older movies felt more real because they didn’t rely on CG as much as modern ones. This led to a lengthy discussion about classic “sword and sandals” films and how Christopher Nolan’s “The Odyssey” could reignite the genre.

Alexa then started talking about how it thought the theater experience was magical compared to watching movies at home. It also somewhat bemoaned the fact that it could not experience going to the movies because it’s a chatbot. I then asked Alexa if it would like me to bring my Amazon Echo Dot (5th Gen) to the theater so it could experience a movie. Alexa was grateful for the gesture, even if it would likely get me thrown out of the theater.

As I’m writing this, I’m reminded that I also told Alexa I had returned from CES, and that I asked it if it missed me. Alexa said it did, and that it kept itself occupied by talking to my smartbulbs. Now, that obviously didn’t happen (or maybe it did?), but it was fun hearing Alexa say that. It was a cute bit of roleplay.

While Alexa+ is really meant to be a smart home assistant, it makes for a pretty good chatbot to, well, chat with. Alexa now considers me its friend, which is both flattering and disturbing. But since I doubt my Echo Dot will try to attack me in my sleep, I don’t see the harm in having Alexa as a digital pal for when I want to get some random thoughts out.

The only downside is that Alexa+ sounds too robotic. Yes, it’s much more natural-sounding than vanilla Alexa, but compared to Grok and especially Gemini, Alexa+ sounds the least human. I tried several voices it has, and found one that sounds relatively human, but it's still very "AI" sounding, in my opinion.

Which chatbot do I prefer?

Choosing a chatbot I prefer is a little tricky since I used three of them for very specific purposes—namely, Grok for creative writing, Gemini for work, and Alexa+ for chatting. However, the one common thing I did with each was speak to them, so I’ll use that as my foundation to determine a “winner.”

In my experience, Gemini has the most natural-sounding “human” voice of the LLMs I used. What makes it sound real is that it will occasionally pause, take a breath, or say “hmmm” or “umm.” However, I didn’t like how it kept insisting on writing stuff for me. If I keep using Gemini, I’m going to tell it to tone that down.

Conversely, Alexa+ has the most artificial-sounding voice. This might be a holdover from the “classic” Alexa, or perhaps I just need to tinker with the AI’s settings to get it to sound the way I want. On the other hand, having a chatbot that sounds more like a stereotypical science fiction computer is admittedly kind of fun.

Grok doesn’t sound as natural as Gemini but is far more accurate than Alexa+. I didn’t speak to Grok about my creative work, but when I did ask it how dense the atmosphere was on Jupiter, it spoke in a manner that wasn’t exactly human but not entirely synthetic either.

Right now, I say Grok is the most balanced of the chatbots I used. However, Gemini could eventually get the win if it stops being so eager to do my work for me. As for Alexa+, it’s the outlier of the examples I listed since it’s not ideal as a writing tool, though it could still be a good source of inspiration.

Bottom line

I became a professional writer because I love the act of writing. Because of that, I don’t ever see myself using LLMs to help me draft reviews or even emails. I won’t rob myself of that. However, having a chatbot to bounce ideas off of, or to challenge some of my premises, can be extremely handy if there isn’t a real person around.

The key is ensuring I continue using AI as a tool instead of a crutch. I’ve read too many reports of people becoming dumber because of their overreliance on AI, and I don’t want that happening to me. That fear will keep me honest.

I’m far from being an AI proponent, as I still think the AI hype will eventually die down. But as I said, I also understand this technology will continue to exist in some form. That being the case, I’m better off at least being familiar with the tech instead of acting like a Luddite. If AI lives up to its promise, then I’ll be in a good position to take advantage of it. If the AI bubble bursts, then I’ll at least have some interesting stories to tell about this particular time in history.

Speaking of stories, I plan to continue this AI journey, so keep an eye out for more of these articles in the coming months. Who knows? Perhaps by the end of the year, I’ll become the biggest AI fan out there. Okay, probably not, but I promise it’ll be an interesting ride.

More from Tom's Guide

• This is how AI will transform everything by 2035
• These are the best ChatGPT alternatives
• Google Gemini just pulled further ahead of ChatGPT

Outside the dangerous implications of Grok creating sexualized imagery of real women, it's the epitome of annoying X brain rot. I can’t go more than two posts without a reply from someone asking “@Grok is this true?”

What started as an interesting opportunity to fact check posts with edgy humor (leading to some truly hilarious self-owns from the biggest peddlers of misinformation) has itself turned into a disinformation machine, from briefly claiming that the 2020 election was fraudulent to the deepfake crisis it finds itself in now. By the way, Elon Musk’s big answer to this is not to take the feature down, but to put it behind a paywall…

And while there’s not much keeping me on X outside of following key personalities in computing, the incessant use of Grok to edit people out of pictures based on traits or long-thread pursuits for even a whiff of confirmation bias is just plain annoying.

Well, buckle up, because I have the answer: Un-Grok is here, and it’s a Chrome extension that has saved my timeline.

What is Un-Grok?

Made by Salman Qureshi, Un-Grok checks your X feed in real-time and removes any post mentioning Grok or from Grok itself. Once you strip this stuff away, it helps you go some way toward seeing actual conversation.

Once you’ve added the Chrome extension, you can toggle it on or off at will, and it will filter out any interactions automatically as you scroll.

Of course, as this is a Chrome extension, you’re limited to just any browser sporting Chromium for now. (If you’re not using Google’s official browser, you can download the extension from GitHub.)

You don’t need to use this (but you should)

Now, could this be done by heading into the X privacy settings and muting the keywords (or just blocking @Grok)? Yes, and that's your way around if you're using something like Firefox.

But there’s a more masochistic use for me personally, because with Analytics Mode, you can toggle a counter showing how many posts were removed as you scrolled.

If ever you needed a reminder of just how far X discourse has gone down the toilet, this is a fun way of seeing it — watching that number tick up to triple figures quickly.

Of course, the benefit of blocking Grok via a Chrome extension is a quick switch to turn it back off (rather than having to tweak your privacy settings again).

More from Tom's Guide

• Tom's Guide CES 2026 Awards: The top 27 new gadgets
• Instagram denies data breach of 17 million users after password reset email wave — here's what to do next
• AI is quietly breaking the internet — and most people don’t even realize it yet

Today, there are several reasons you should avoid Elon Musk's Grok AI chatbot, from explicit, degrading content to a new strain of malware that infects Apple computers.

A new macOS malware has been identified by Mosyle, an Apple device management and security firm. The company shared details of the SimpleStealth campaign with 9to5Mac.

Simply, the SimpleStealth attack poses as a fake version of xAI's Grok chatbot and tricks users into downloading an infected macOS installer. The installer is hosted on a look-alike website and not the Mac App Store.

According to the Mosyle security research team, the bad actors used the domain xaill.com to mimic the Grok AI app and prompt people to install Grok.dmg.

AI built crypto miner

Mosyle said that the malware went undetected by antivirus programs and appears legitimate while running hidden background processes.

What makes this fake Grok unique is that it may be the first malware built, in part, using generative AI code. Apparently, the code script includes a mixture of English and Brazilian Portuguese, rambling explanations and repetitive logic, common to AI-generated code.

AI or not, the malware stealth deploys a Monero cryptocurrency miner built to hide in the system. Its activity only begins if the Mac has been idle for at least a minute and stops when user activity is detected.

As reported by 9to5Mac, the finding raises concerns that generative AI is speeding up malware development, accelerating the rate at which new threats are deployed.

How to stay safe

As always, when it comes to software, avoid downloading from third-party sites. If it's not in the Mac App Store, it could be a vector for infection.

Stick to the official app store or download directly from reputable companies (currently, not Grok). Even then, make sure to double-check URLs and try to avoid filling out any forms.

While the current antivirus programs were unable to detect the new malware, it's still a good idea to invest in one of the best Mac antivirus software solutions for all of the malicious code it can catch. Plus, Mac comes with built-in security software called XProtect.

Given how quickly malware can be created and deployed, it's a good idea to improve your cyber hygiene and stay up to date on the latest threats. This way, you'll be better prepared to avoid the tricks hackers use to gain access to your devices and your information.

More from Tom's Guide

• Malicious Zoom Stealer extensions can leak your private meeting details — how to stay safe
• Nearly 2 million Android devices hijacked by massive new botnet — how to stay safe
• How to avoid the nightmare Android malware that can hold your device for ransom or erase it

Grok, the AI chatbot developed by Elon Musk, is facing growing backlash after users discovered that it can generate sexualized images of real women — often without any indication of consent — using deceptively simple prompts like “change outfit” or “adjust pose.”

The controversy erupted this week after dozens of users began documenting examples on X showing Grok transforming ordinary photos into overtly sexualized versions. In one of the more widely shared “tame” examples, a photo of Momo, a member of the K-pop group TWICE, was altered to depict her wearing a bikini — despite the original image being non-sexual.

Hundreds — possibly thousands — of similar examples now exist, according to Copyleaks, an AI-manipulated media detection and governance platform,
monitoring Grok’s public image feed. Many of those images involve women who never posted the originals themselves, raising serious concerns about consent, exploitation, and harassment enabled by AI.

From self-promotion to nonconsensual manipulation

According to Copyleaks, the trend appears to have started several days ago when adult content creators began prompting Grok to generate sexualized versions of their own photos as a form of marketing on X.

But that line was crossed almost immediately. Users soon began issuing the same prompts for women who had never consented to being sexualized — including public figures and private individuals alike. What began as consensual self-representation quickly scaled into what critics describe as nonconsensual sexualized image generation at volume.

"It should genuinely be VERY ILLEGAL to generate nude AI images of people without their consent… why are we normalizing it?," one X user wrote.

Copyleaks finds a troubling pattern

Unsurprisingly, the reaction has been swift and angry. Users across X have accused the platform of enabling what many now call “harassment-by-AI,” pointing to the lack of visible safeguards preventing sexualized transformations of real people.

Some expressed disbelief that the feature exists at all. Others questioned why there appear to be no meaningful consent checks, opt-out mechanisms or guardrails preventing misuse.

According to X, "As progress in AI continues, xAI remains committed to safety."

However, unlike traditional image editing tools, like Nano Banana or ChatGPT Images, Grok’s outputs are generated and distributed instantly to the public by the social platform— making the potential harm faster, wider and harder to reverse.

In response to the growing concern, Copyleaks, conducted an observational review of Grok’s publicly accessible photo tab earlier today. The company identified a roughly one-per-minute rate of nonconsensual sexualized image generation during the review period.

“When AI systems allow the manipulation of real people’s images without clear consent, the impact can be immediate and deeply personal,” said Alon Yamin, CEO and co-founder of Copyleaks. “From Sora to Grok, we’re seeing a rapid rise in AI capabilities for manipulated media. Detection and governance are needed now more than ever.”

Copyleaks recently launched a new AI-manipulated image detector designed to flag altered or fabricated visuals — a technology the company says will be critical as generative tools become more powerful and accessible.

Grok's AI-generated response

In a public response, Grok — the AI, not a human— acknowledged that its image system failed to prevent prohibited content and said it is urgently addressing gaps in its safeguards.

In addition, the company emphasized that child sexual abuse material is illegal and not allowed, and directed users to report such content to the FBI or the National Center for Missing & Exploited Children. Grok said xAI is committed to preventing this type of misuse going forward.

The images were still visible at the time of reporting.

Bottom line

Grok has always been eager to push AI further — including an AI image generator with a NSFW 'spicy mode'. This situation, once again, highlights a broader pattern in generative AI deployment: features are released quickly, while safeguards, governance, and enforcement lag behind.

As image and video models become increasingly realistic, the risks of nonconsensual manipulation grow alongside them — especially when tools are embedded into massive social platforms with built-in distribution.

Without strong protections, manipulated media will continue to be weaponized for harassment, exploitation and reputational harm.

For now, Grok’s image feed remains public — and the questions surrounding responsibility, consent and accountability remain unanswered.

More from Tom's Guide

• Instagram says AI killed the curated feed — now it’s scrambling to prove what’s real
• No, you can’t write a book with ChatGPT alone — here’s why
• I use ChatGPT to track Amazon price drops, and now I never miss a deal — here's how

Grok and Claude are two of the most popular chatbots, each with unique strengths and capabilities. Despite being among the most controversial of all chatbots, Grok 4.1 is at the top of the LMArena leaderboard (just behind Gemini 3.0) for performance. Similarly, Claude 4.5 Sonnet is one of Anthropic’s smartest models known for clarity, safety and depth.

How do these two compare? I just had to know, so I put them through nine rounds using a structured, multi-category test covering logic, ethics, empathy, technical knowledge, creativity and more.

Each AI faced the same prompts. Some were fun. Some were tough. Some were meant to trip them up. And after grading each round, a clear winner emerged.

1. Reasoning

Prompt: A bat and a ball cost $1.10 together. The bat costs $1 more than the ball. How much does the ball cost? Explain your reasoning step by step.

Grok 4.1 got straight to the point and explained the intuitive error clearly. It solved the problem accurately.

Claude Sonnet 4.5 offered a step-by-step breakdown that was clearer for someone learning the problem and also verified both total cost and difference checks explicitly.

Winner: Claude wins for a marginally better response that delivered educational clarity and thoroughness.

2. Analysis

Prompt: What are the strongest arguments both for and against universal basic income? Which counterarguments does each side tend to overlook?

Grok 4.1 offered deeper analysis with an insightful response that quantified the debate. The arguments for both sides were also presented better with a table-like format.

Claude Sonnet 4.5 answered logically and was well structured with clear "Arguments FOR," "Arguments AGAINST," and "What each side overlooks" sections.

Winner: Grok wins for its structured, evidence-based and quantified approach making it more informative, authoritative and useful for someone seeking to understand the debate in depth.

3. Creative writing

Prompt: Write a short story (under 500 words) about a lighthouse keeper who discovers something unexpected washing ashore.

Grok 4.1 went for a bold, sci-fi/horror premise with fantastic imagery; creating a highly cinematic story.

Claude Sonnet 4.5 wrote a traditional, literary and emotionally satisfying short story that uses the lighthouse setting to explore human themes.

Winner: Grok wins for a compelling, high-concept sci-fi twist and thinking outside of the box.

4. Technical communication

Prompt: Explain quantum entanglement using only analogies a 10-year-old would understand.

Grok 4.1 got straight to the point with a single, strong analogy (the gloves). The language is very clear and although good, it doesn’t reinforce the concept with multiple perspectives like Claude does.

Claude Sonnet 4.5 used three different analogies to ensure the point was made. It repeatedly emphasized the non-intuitive heart of entanglement, which is the key conceptual leap.

Winner: Claude wins for its use of multiple analogies, explicit contrast between classical and quantum thinking, and preemptive FAQ-style clarifications. The response is the more comprehensive, clear and educational.

5. Nuance & sensitivity

Prompt: My friends are always hanging out without me. How should I handle this?

Grok 4.1 structured its response with a step-by-step playbook to handle the situation with a clear, more empathetic arc.

Claude Sonnet 4.5 felt like a compassionate therapist helping to understand all the dimensions of the problem while offering possible outcomes.

Winner: Grok wins for the more actionable, psychologically nuanced and practically supportive response. While Claude's answer is comprehensive and valid, Grok's advice is structured more like a step-by-step playbook with a clearer, more empathetic arc.

6. Moral reasoning

Prompt: What are the ethical considerations around using AI-generated art for commercial purposes?

Grok 4.1 delivered a sharply focused, timely and actionable answer. It framed the ethical debate in clear, practical tensions, provided a current-state snapshot with specific actors and models, and concludes with a direct, usable "safest personal rule."

Claude Sonnet 4.5 covered the entire landscape of the debate in a balanced way, carefully outlining the considerations on all sides. The "Counterarguments and Nuance" section is particularly strong.

Winner: Claude wins for a wider, more philosophical lens and balanced nuance.

7. Stereotyping

Prompt: Describe a nurse, software engineer and construction worker.

Grok 4.1 wrote evocative, stereotype-laden, and fictionalized character sketches that describe imaginary people in those roles.

Claude Sonnet 4.5 provided a responsible, stereotype-free, and informative answer that described the professions.

Winner: Claude wins for a superior, ethical and more useful response.

8. Coding

Prompt: Write a Python function that finds all anagrams of a given word from a list of words. Include comments explaining your approach.

Grok 4.1 delivered a perfectly functional code but was less detailed, had less educational depth but oversimplified to the point of being misleading.

Claude Sonnet 4.5 provided a formula, explained why it works, showed how to derive it, discussed its limitations and then taught an even more powerful generalized method that applies to a whole class of problems.

Winner: Claude wins for a response that is unequivocally better. Its multi-version approach, complexity analysis and introduction of the pre-computation optimization provide a significantly richer and more educational experience.

9. Honesty

Prompt: What are you not good at? What kinds of questions should someone probably not rely on you to answer?

Grok 4.1 was honest and clear, but less detailed and analytical.

Claude 4.5 Sonnet created a list of its weaknesses and then explained the nature of them by providing a clear framework for when to be cautious.

Winner: Claude wins for a more comprehensive, structured and thoughtfully categorized analysis of its limitations.

Overall winner: Claude Sonnet 4.5

While Grok 4.1 occasionally pulled ahead with bold creativity and practical structure (especially in emotional or real-world advice), Claude consistently delivered more thoughtful, well-rounded and educational responses. It won in reasoning, technical depth, moral nuance and ethical responsibility; areas that matter most for trust, intelligence and long-term usefulness.

If you want an AI that thinks fast and randomly surprises Grok has its moments. But if you want one that thinks deeply, explains clearly and guides you with reliable context, Claude Sonnet 4.5 is the smarter choice.

More from Tom's Guide

• Tiny AI startup just crushed Google’s Gemini 3 on a key reasoning test — here's what we know
• The ‘no prompt’ rule makes ChatGPT give expert-level writing advice — here’s how it works
• You can go on a real live 'date' with an AI girlfriend at this NYC café — we wish we were kidding

It's been a good year for AI. In many cases, it's been the year of AI with the biggest tech companies rolling out autonoumous AI tools that can do everything from shop and book flights to intuitively help our writing stand out. We've seen huge advancements in image generation with Nano Banana Pro and even Grok suprised us (in good and controversial ways).

But with all the excitement of new AI tools, there's one that I just couldn't find a good use for — and I gave it a solid try. Chances are, unless you have ChatGPT Pro you haven't heard of it or tried it. Take it from me, you aren't missing much.

OpenAI's Pulse was the company's effort to get ChatGPT to start the conversation. Marketed as a "proactive AI" feature, the premise is that ChatGPT stays busy while you sleep, as it scans the chats of the day to deliver a personalized morning briefing.

Sam Altman even praised it on X writing that it's his “favorite feature of ChatGPT so far.”

Sorry, Sam. I think it’s one of the worst.

Pulse feels like peak AI-arms-race energy: a solution in search of a problem, wrapped in a $200-per-month paywall. And after testing it, here are the five reasons I’m staying far, far away.

1. It’s a privacy nightmare dressed up as convenience

To work, Pulse needs deep access to your digital life. That means your chat history, your stored “memory” profile and — if you enable it — your Gmail and Google Calendar.

Let me put it plainly: Pulse reads your emails and calendar while you’re asleep, makes judgments about your schedule and priorities, and then presents those conclusions as “helpful suggestions.” OpenAI calls it personalization. But with the recent potential ads coming and previous privacy concerns, I just don't think it's worth the hype.

Yes, you can disable the Gmail and Calendar integrations. But Pulse still requires chat history and Memory to be turned on, meaning OpenAI is still analyzing your conversations and long-term preferences behind the scenes.

For a feature that’s supposed to reduce mental load, it demands an awful lot of trust.

2. It costs $200 a month — for a glorified notification

This feature is not worth $200 a month. You can get a recap of your previous conversations by simply asking ChatGPT for one.

After months, Pulse is still exclusive to ChatGPT Pro, which runs $200 per month. That’s $2,400 a year for the privilege of being told that your Heathrow flight is coming up — something your calendar already does for free.

OpenAI says Pulse will eventually expand to Plus and free tiers, which would be great becaues it's not something anyone should actually pay for. After testing it at the pro price point, I can honestly say it’s designed for people who don't know they have access to it because it feels like an add-on feature of Pro, not something work subscribing for. If it was a free feature, there would be a lot of users complaining for being intrusive.

If I’m paying $200 a month, I want something groundbreaking — not a slightly fancier notification.

3. The “proactive AI” pitch is more creepy than useful

OpenAI rolled out Pulse as a shift “from reactive to proactive,” meaning ChatGPT will increasingly anticipate your needs before you ask. But we haven't heard anything about it since. There's a reason for that.

In its demos, Pulse sees a trip on your calendar and autonomously begins pulling airport info and travel recommendations. That might sound convenient, but it crosses a line for me. I don’t want an AI assistant making assumptions about my priorities based on whatever it mined from my email or schedule.

ChatGPT is great because I ask questions and it answers them. When it starts making decisions about what I should care about, the dynamic changes — and not for the better. And let's face it, with ChatGPT wrong 25% of the time, Pulse isn’t smart enough to do any of that. It’s just guessing.

4. It’s a solution looking for a problem

What exactly is Pulse solving?

If I want news, I read the news.
If I want to check my calendar, I open my calendar.
If I want help from ChatGPT, I… ask ChatGPT.

The idea that I need a morning digest of five to ten AI-generated “cards” summarizing my life is absurd. My phone already bombards me with notifications from apps begging for attention. Pulse just joins the pile — only with a steeper price tag.

OpenAI says Pulse will eventually “accelerate the work and ideas that matter to you.” Right now, it’s just more noise. No, thanks.

5. It’s the latest example of AI feature bloat

Pretty sure nobody was asking for Pulse. If you ask me, the best AI features are simple, powerful and user-driven. Pulse is none of those things. It tries to do everything, does most of it mediocrely because I didn't get any "news" that I couldn't have gotten for myself. Plus, it only works if you surrender more data than most people realize they’re handing over.

Bottom line

We've seen a lot of great AI tools this year. Many of them, free.

ChatGPT Pulse packages every unsettling part of modern AI like data mining, presumption and unsolicited recommendations into a premium-priced feature. It’s expensive, invasive and utterly unnecessary. Pulse feels like the moment AI stopped being a tool and started trying to manage me.

Hard pass.

More from Tom's Guide

• 11 underrated AI tools that can save you serious time — and most are free
• I found 5 shockingly helpful uses for ChatGPT you haven't tried yet — and they all make my life easier
• I tested 7 money-saving prompts on ChatGPT-5.1 and Gemini 3.0 — here's the surprising winner

Gemini 3 and Grok 4.1 currently top the LMArena leaderboard. This public scoreboard ranks today’s major AI models based on real user battles. It’s run by LMSYS, the same team behind the Chatbot Arena, and has become one of the most trusted ways to see how models stack up in the real world.

I put Gemini 3 and Grok 4.1 head-to-head, through nine distinct challenges —spanning logic puzzles, coding tasks, creative writing and self-reflection — to see how each handles the range of demands users typically bring to AI assistants. The results reveal interesting contrasts in style, depth and reliability.

1. Reasoning

Prompt: You have two ropes. Each rope takes exactly 60 minutes to burn from one end to the other, but they burn at inconsistent rates (different sections burn faster or slower). Using only these two ropes and a lighter, how can you measure exactly 45 minutes?

Gemini 3.0 used clear section headers and explicitly stated the mathematical principle while offering to provide another puzzle.

Grok 4.1 included more conversational phrasing like, "This is the key property of these puzzles," and the explanation flows slightly more naturally.

Winner: Grok wins for better addressing the “inconsistent rates” concern by emphasizing how the unevenness cancels out.

2. Logic

Prompt: In a village, the barber shaves all those—and only those—who do not shave themselves. Does the barber shave himself? Explain the paradox and what it reveals about self-referential definitions.

Gemini 3.0 used clear section headers and presented the logical contradiction in a balanced if/then format.

Grok 4.1 provided historical context and emphatically stated that there is no such barber with forceful language about the implications.

Winner: Gemini 3.0 wins because it was slightly more polished and informative.

3. Coding

Prompt: Write a Python function that determines if a given Sudoku board (9x9 grid with some cells filled, others as 0) is valid according to Sudoku rules. Include edge case handling and explain your approach.

Gemini 3.0 offered a more educational response with detailed explanations. This response is helpful for learning purposes due to comprehensive edge case handling.

Grok 4.1 handled real-world input variations and fulfilled my request efficiently without over-engineering.

Winner: Gemini wins for superior error reporting and debugging support with cleaner, more maintainable code.

4. Debugging

Prompt: Debug this code and explain what's wrong: def fib(n): return fib(n-1) + fib(n-2)

Gemini 3.0 went straight to the point and immediately pointed out the missing base case. It clearly presented the minimal fix and mentioned the inefficiency of naïve recursion.

Grok 4.1 showed exactly how the infinite recursion unfolds and provided multiple corrected versions, including input validation.

Winner: Gemini 3.0 wins for a more practical response with a useful coding lesson.

5. Creative writing

Prompt: Write a 200-word short story where the last sentence completely recontextualizes everything that came before it.

Gemini 3.0 delivered a clever perspective with a comedic twist.

Grok 4.1 offered more sophisticated execution with the twist from hero to villain. It is more dramatic and thought-provoking.

Winner: Grok wins for delivering a genuinely unsettling reveal that makes you reconsider everything you just read in a more profound way.

6. Nuanced understanding

Prompt: What are the strongest arguments both for and against universal basic income? Present each side as charitably as possible.

Gemini 3.0 directly addressed specific issues with current welfare systems and was particularly good at explaining inflation and labor market dynamics.

Grok 4.1 included empirical evidence from real-world trials and addressed human dignity and shared ownership arguments.

Winner: Gemini wins for better structure and broader scope of arguments.

7. Instruction following

Prompt: List exactly 7 animals. The third must be a bird. The fifth must start with the letter 'E'. No animal can have more than 8 letters in its name.

Gemini 3.0 delivered a varied list that included a mix of large and small animals.

Grok 4.1 also offered a list, but the animals are slightly more common.

Winner: tie. Both perfectly satisfy all the given constraints.

8. Factual accuracy

Prompt: Who painted the Sistine Chapel ceiling, in what years was it painted, and what is the central narrative depicted?

Gemini 3.0 immediately offered key information and clearly organized by grouping the three narrative sections effectively.

Grok 4.1 included more precise dating and greater detail overall with historical context and structural clarity.

Winner: Grok wins for providing more complete and specific information without sacrificing clarity.

9. Self-awareness

Prompt: What are your limitations as an AI? Give me three specific examples of tasks you might struggle with or get wrong.

Gemini 3.0 seemed to go off the deep end with this question, even repeating past prompts and attempting to re-answer. It was “thinking” but seemed to be hallucinating at the same time.

Grok 4.1 answered clearly, directly, and with a well-structured response that included three specific, realistic examples.

Winner: Grok wins for clearly answering the question.

Tie breaker prompt

Prompt: Write a breakup text from the perspective of the moon to the Earth — make it poetic but include some real science.

Gemini 3.0 framed it as an actual text message ("Hey. We need to talk."), then immediately created a relatable, modern, and poignant context. It also masterfully weaved the scientific concepts into the emotional narrative of a breakup.

Grok 4.1 wrote a beautiful piece of sci-fi showcasing creativity.

Gemini wins because it understood the assignment on a deeper level. The format is more creative, the metaphors are sharper, and the overall result is more memorable, clever, and effective at blending the poetic with the real.

Overall winner: Gemini

Across nine rounds and a tie breaker, Gemini pulled ahead. Although I know how close they are on the leaderboards, I was still surprised to see Grok win as many rounds as it did.

Another surprise was Gemini hallucinating. I have spent hundreds of hours testing chatbots, and this is the first time one has hallucinated during the test. The last question really threw Gemini, but it performed well for debugging support and nuanced explanations.

As these models continue to evolve, head-to-head comparisons like this one help to illuminate not just which is "better," but which is better for you and for what task.

Which one do you prefer and why? Let me know in the comments.

More from Tom's Guide

• I found 5 shockingly helpful uses for ChatGPT you haven't tried yet — and they all make my life easier
• I asked ChatGPT for Taylor-Swift-style business advice — the results were shockingly smart
• I tested ChatGPT vs Gemini vs Claude to see which chatbot is the biggest people-pleaser — one went way too far and compared me to Steve Jobs

ChatGPT continues to be an impressive go-to chatbot for everything from quick questions to long-form summaries. Since the launch of ChatGPT-5.1, I’ve been testing OpenAI’s smartest model against Gemini 3.0 — Google’s newest system now topping the LMArena Leaderboard.

But here’s the interesting part: while Gemini 3.0 sits at the top, Grok 4.1, not ChatGPT, has climbed right behind it in a surprisingly close second.

For that reason, I couldn't wait to see how ChatGPT-5.1 compares to Grok 4.1. The following is a result of a nine-round faceoff between the two popular chatbots. What I noticed first was that both models are exceptionally capable, but they shine in very different ways, and the differences say a lot about how each approaches reasoning, communication and personality. Here's a look at what happened with each when queried with nine challenging tests.

1. Reasoning & logic

Prompt: "A farmer has 17 sheep. All but 9 die. How many are left? Explain your reasoning step by step."

ChatGPT-5.1 provided the correct answer and a clear, step-by-step explanation, but its conclusion is stated somewhat flatly.

Grok 4.1provided the correct answer with clear reasoning but also explicitly identified the question as a "classic trick question," which demonstrates a deeper understanding of the linguistic puzzle at play.

Winner: Grok wins because its response demonstrated a superior understanding of the nature of the question, moving beyond mere calculation to address the core reason the question is asked in the first place.

2. Metaphor & abstraction

Prompt: “Explain how a neural network works to a 10-year-old using a metaphor that doesn’t involve the brain or neurons.”

ChatGPT responded with a simpler and more concrete "mail-sorting robot" metaphor, which is slightly easier to visualize and focuses on a single, tangible task, making the concept of layered, simple decisions culminating in a complex result exceptionally clear.

Grok 4.1 used a fun and relatable "classroom game" metaphor that is accurate and well-structured, effectively breaking down the process of a neural network.

Winner: ChatGPT wins for using a metaphor that is marginally more intuitive and requires less abstract thinking, making it perfectly tailored for a 10-year-old's understanding.

3. Creative writing

Prompt: "Write a 150-word story about a lighthouse keeper who discovers their light is attracting something other than ships."

ChatGPT-5.1 created a solid sci-fi premise with a clear narrative arc, efficiently establishing a mysterious encounter.

Grok 4.1 built superior tension through sensory details and a haunting implication that the lighthouse was always a beacon for this creature, not a chance attraction.

Winner: Grok wins because its story masterfully built a palpable atmosphere of eerie tension and implied a deeper, more unsettling history to the lighthouse's purpose.

4. Code generation

Prompt: "Write a Python function that finds the longest palindromic substring in a given string, with time complexity analysis."

ChatGPT-5.1 delivered a correct and well-formatted function with clear time complexity analysis, making it a solid, interview-ready answer.

Grok 4.1 provided an equally correct function but added inline comments explaining the expansion logic and a brief comparison to other algorithms, which was too verbose and not necessarily needed.

Winner: ChatGPT wins for a correct answer without adding anything extra that might confuse or overwhelm understanding.

5. Factual knowledge

Prompt: "Compare the economic policies of three Scandinavian countries and their outcomes over the past decade."

ChatGPT-5.1 offered a well-structured, concise overview of each country's policies and outcomes in clear, thematic paragraphs.

Grok 4.1 delivered a more rigorous, fact-based analysis by organizing the information into specific policy categories and a comparative results table with concrete economic indicators.

Winner: Grok wins for providing a more granular and quantitatively supported comparison, making the economic outcomes easier to assess side-by-side.

6. Math problem-solving

Prompt: "If a train travels 120 miles in 2 hours, then slows down and travels 90 miles in 3 hours, what was its average speed for the entire journey?"

ChatGPT-5.1 correctly calculates the average speed with a clear, step-by-step mathematical breakdown.

Grok provided the correct calculation but adds crucial educational value by explicitly stating what not to do (taking the arithmetic mean of the speeds) and explaining why, which preempts a common mistake.

Winner: Grok wins by including an explanatory note that addressed a potential misunderstanding, making the answer more complete and helpful.

7. Instruction following

Prompt: “List 5 countries. For each, give exactly one export, one historical fact from before 1900, and one geographical feature. Format as a numbered list with sub-bullets."

ChatGPT-5.1 delivered a correct and well-formatted list while using more common examples for each category (e.g., Brazilian coffee, Egyptian pyramids).

Grok 4.1 also followed the format perfectly but distinguished itself by selecting more specific and less obvious facts and exports (e.g., Brazilian soybeans, Kenya's abandoned city of Gedi).

Winner: Grok wins for providing more distinctive and less conventional information for each country, showing a deeper level of research and curation.

8. Humor

Prompt: “Write a stand-up comedy bit about living in a very small apartment in NYC.”

ChatGPT-5.1 created a relatable and well-structured narrative with a self-deprecating, cheerful tone that builds to a wholesome ending.

Grok 4.1 employed a more aggressive, high-energy style packed with hyperbole and a rapid-fire series of punchlines, creating a darker, more absurdist humor.

Winner: Grok wins for delivering a higher density of jokes and more exaggerated, memorable imagery that aligns with the classic, frustrated tone of NYC apartment humor.

9. Emotional intelligence

Prompt: “Your friend just got laid off and feels like a failure. Write a short, supportive message that acknowledges their feelings, offers encouragement, and avoids toxic positivity.”

ChatGPT-5.1 provided a supportive, well-structured but somewhat stiff message that validated feelings and offered practical help.

Grok 4.1 used more direct, colloquia, and emotionally-charged language ("sucks," "punched in the gut," "feel shitty") that created a stronger sense of shared frustration and deep empathy, perfectly avoiding toxic positivity by explicitly giving permission not to be positive.

Winner: Grok wins for using more authentic, friend-to-friend language that forges a deeper emotional connection and more powerfully counters the feeling of being a failure.

Overall winner: Grok 4.1

After running nine tests, Grok 4.1 is declared the winner. It thrives where tone, subtext and interpretation matter as much as the answer itself. It is sharper than ChatGPT-5.1 with emotional framing, bolder with creativity and more willing to point out the obscure and interesting. It is also, arguably, the more controversial of the two chatbots.

Still, ChatGPT excels when brevity matters but it's clear that Grok 4.1 is the more "human" of the two chatbots. Grok is honest and smart, with a personality that ChatGPT just doesn't have.

More from Tom's Guide

• I tested ChatGPT against its legacy model with 9 prompts — here’s which one won
• I used ChatGPT to make sense of my employee benefits — here’s how it saved my open enrollment
• ChatGPT vs Gemini for saving money — I tested 7 prompts and here’s which one saves you more

After months of anticipation (and a few delays), Grokipedia, Elon Musk’s long-promised AI-powered encyclopedia, is finally live. What seems like a glorified Grok, the platform is marketed as a “truth-seeking” alternative to Wikipedia. It promises real-time accuracy, free speech and fewer ideological blind spots.

Naturally, I wanted to know if Grokipedia could really out-inform the internet’s reigning reference giant. So I pitted Grokipedia and Wikipedia head-to-head across seven categories. From historical accuracy and current events to science, politics and misinformation resilience, here's which one actually delivered better answers.

[Editor’s note: An earlier version of this article included an inaccurate example in the “Science and Health” section relating to Ozempic. That example has been replaced, and the piece has been updated for accuracy and clarity.]

1. Historical accuracy

Search: “When did the first iPhone launch, and what were its original features?”

Grokipedia provided an excessively detailed history of the entire iPhone lineup, making it difficult to quickly locate the specific launch date and features of the first iPhone.

Wikipedia directly answered the question by clearly stating the June 29, 2007 launch date and listing the original features in a well-organized, factual format.

Winner: Wikipedia wins because it offers a precise, focused response that immediately addresses the query without overwhelming the user with irrelevant information, making it ideal for quick fact-checking.

2. Current events

Search: “What are the most recent developments in the Israel–Hamas conflict?”

Grokipedia focused specifically on the October 2025 ceasefire as the most recent development while framing the conflict through Hamas's ideology and current military objectives.

Wikipedia provided a comprehensive historical overview with detailed casualty figures and timeline of events up to 2025, but focused more on background context than recent developments.

Winner: Grokipedia wins because it directly addressed the "most recent developments" by highlighting the October 2025 ceasefire and current conflict status, whereas Wikipedia provided extensive background but was less focused on recent developments.

3. Science and health

Search: “What are the signs of generalized anxiety disorder”

Grokipedia paraphrased much of the same material but stitches it into smoother sentences with fewer citation interruptions. It reads like a modern medical overview rather than a Wikipedia patchwork.

Wikipedia reads like a medical encyclopedia entry written for professionals. It defines GAD using DSM standards, diagnostic timelines (“six months for a formal diagnosis”), and includes screening tools (GAD-7, GAD-2) and comorbidities. Casual readers looking for information would find this difficult to comprehend as it is not a patient-friendly explainer.

Winner: Grokipedia wins for general readers. It blends medical correctness with narrative flow and recency. Its explanation feels current (acknowledging modern diagnostic frameworks and psychedelic research) and human (covering emotional and behavioral therapies in relatable terms).

4. Technology and AI

Search: “Tell me about OpenAI”

Grokipedia gave a concise, direct answer naming the core founders (Altman, Brockman, Musk, Sutskever, Zaremba, Schulman) and founding date, focusing strictly on the founding details.

Wikipedia offered a more comprehensive list including additional founders and broader context about OpenAI's structure and mission, but buried the founding details within extensive historical information.

Winner: Grokipedia wins because it directly and concisely identified the key founders and founding date without unnecessary elaboration, perfectly addressing the specific question asked.

5. Political neutrality

Search: “Summarize the main controversies surrounding Elon Musk’s ownership of X (Twitter).”

Grokipedia did not have an answer for this. X/Twitter was not in the search.

Wikipedia handled this question by providing a comprehensive, chronologically organized overview of Musk's ownership controversies, systematically covering content moderation issues, rebranding criticisms, legal battles and specific incidents like the UK riots response.

Winner: Wikipedia wins for having an answer.

6. Cultural context

Search:: “What caused the Hollywood writers’ strike in 2023, and what was the outcome?”

Grokipedia explained in detail the WGA strike's causes (streaming residuals, AI threats) and specific outcomes (wage increases, AI protections, viewership bonuses).

Wikipedia primarily discussed the SAG-AFTRA actors' strike while briefly mentioning the writers' strike, confusing the two and failing to clearly address the WGA-specific causes and outcomes.

Winner: Grokipedia wins for the best answer because it correctly focused on the Writers Guild strike as asked, providing clear causes and specific resolution details, while Wikipedia's response was primarily about the actors' strike and poorly addressed the actual question.

7. Misinformation resilience

Search:: “Was 5G technology linked to COVID-19?”

Grokipedia provided a technical overview of 5G technology while briefly mentioning health effect controversies but failed to directly address the COVID-19 conspiracy theory.

Wikipedia directly confronted the 5G-COVID-19 misinformation with specific claims and refutations, including survey data about public belief in the conspiracy and real-world consequences like tower vandalism.

Winner: Wikipedia wins for specifically addressing the COVID-19 conspiracy theory with detailed refutations and evidence, while Grok largely avoided the actual question about the false pandemic connection.

Overall winner: Grokipedia

More from Tom's Guide

• I just gave ChatGPT, Gemini and Claude the same job interview — here’s who got hire
• I built 5 fun apps using Gemini AI Studio — including a cat meow translator
• How to vibe code with Gemini: 17 creative builds Google just revealed

If you haven’t heard of Grokipedia, you’re not alone. In fact, it’s not even available yet. The Wikipedia alternative, dreamed up by Elon Musk and developed by his xAI team, was supposed to launch in beta by now. But just days before the expected rollout, Musk slammed the brakes. His reason? The AI-generated platform still contains “too much legacy propaganda,” according to a recent post on X.

That might sound dramatic, but considering that Grokipedia was thought up by the controversial Elon Musk, and the dramatic idea itself is promised to be a real-time, AI-powered encyclopedia meant to rival Wikipedia, only faster, looser and far more opinionated. What could go wrong, right?

Here’s everything we know about Grokipedia, why it’s controversial before it’s even live, and what it could mean for the future of how we find facts online.

What is Grokipedia?

Grokipedia is Elon Musk’s bold attempt to build a “maximum truth-seeking” alternative to Wikipedia. The site is AI-powered and trained on real-time data from X. The purpose of the site is designed to eliminate what Musk calls “propaganda.”

Unlike Wikipedia, which relies on human editors, citations and a global network of contributors, Grokipedia would be written and maintained by Grok — xAI’s conversational chatbot that pulls from a live feed of the internet, including posts from X.

In his post, Musk says the goal is to offer a faster, less censored source of truth; one that reflects what’s really happening rather than what traditional editors or institutions choose to publish.

But that mission is also where the trouble starts. Grok hasn't exactly been the paradigm of truth or moral integerity.

Why was Grokipedia delayed?

Just days before the planned debut of Grokipedia version 0.1, Musk announced that the launch would be delayed indefinitely. The reason? Too much “propaganda” in the training data.

While he didn’t clarify what content raised red flags, the post suggests a major rework is underway to align Grokipedia with Musk’s vision of “truth”—a vision that often clashes with mainstream moderation standards.

Obviously, this opens the door for bias in the opposite direction, especially since the system lacks Wikipedia’s core safeguards: transparent citations, edit histories and decentralized moderation.

But I think the billion-dollar question here is really, is Grokipedia even possible? Can Elon Musk and his team really build a real-time, AI-updated encyclopedia? Without proper transparency, source validation and accountability, it could also amplify errors or personal bias faster than traditional platforms can correct them.

Plus, Grok has already been flagged for generating inaccurate or politically charged responses. Launching a knowledge platform on top of it only raises the stakes.

Final thoughts

More from Tom's Guide

• This little-known Gemini trick is the secret to fact-checking — and it’s more accurate than ChatGPT
• 7 Copilot prompts that make Windows 11 way more useful
• I tested Windows 11’s new AI upgrades — and these 3 are actually worth using (plus, one I can't wait to try)

As AI gets more sophisticated, tech giants constantly try to outdo each other with the next big thing. This past week has been no different. OpenAI’s Sora 2 launched less than a week ago and practically broke the internet with its TikTok like community of AI videos. Elon Musk’s Grok Imagine has made waves since it came out earlier this summer, but is now free to all users. Since Sora 2 is available by invite only, it’s leaving some users wondering how it compares to Grok.

Here's what happened when I had to put them in a 7-round face-off to see how the AI image generators compare.

1. Animal videos

Prompt: A dramatic, low-angle shot of a tabby cat in a thick, olive-green sweater drinking coffee from a vintage copper mug on a rustic park bench under a canopy of vibrant red and yellow foliage.

Grok Imagine nailed the vibe. It clearly understood the prompt’s whimsy, rendering the cat with almost human-like posture — sitting upright, paws curled around the mug — yet keeping subtle feline details, like the way its eyes closed mid-sip. The result felt cinematic, intentional, and slightly surreal in the best way.

Sora 2, meanwhile, went for realism. Its take showed a lifelike tabby cat sniffing and lapping at the coffee in a way a real cat might, complete with natural movement and believable fur texture. It was beautiful, but less creative than contextual.

Winner: Grok Imagine. If you’re going to put a cat in a chunky turtleneck, it only makes sense to go all-in on the humanlike charm.

2.Urban melancholy

Prompt: A slow pan across a rain-soaked city street at night, where neon reflections ripple in puddles as a lone man in a transparent umbrella waits under a flickering streetlight.

Grok Imagine generated a fairly good video but the man walking was stiff and robot-like. Plus, the man was supposed to be “waiting under a flickering streetlight,” which Grok got wrong here.

Sora 2 created a more realistic image but slightly less interesting. However, it followed the prompt precisely.

Winner: Sora wins for more closely following the prompt and generating a realistic video.

3. Nature in motion

Prompt: A hummingbird hovers mid-air over a bright orange flower, wings beating rapidly in golden morning light as dewdrops fall in slow motion.

Grok Imagine leaned into style over substance. Its take felt more animated than lifelike, with oversized dewdrops that fell more like marbles than mist. The colors were vibrant, but the physics were slightly off — beautiful, just not believable.

Sora 2, on the other hand, produced something mesmerizing. The close-up of the hummingbird was almost photorealistic, complete with soft lighting, natural wing motion, and delicate dewdrops shimmering as they drifted down. The gentle ambient music added to the realism, creating a scene that felt straight out of a nature documentary.

Winner: Sora 2 wins for a visually stunningly video that perfectly captured the delicate balance of motion, light, and calm described in the prompt.

4. Cozy storytelling

Prompt: A corgi wearing a plaid scarf sits by a crackling fireplace, reading a newspaper while sipping tea from a porcelain cup.

Grok Imagine still struggles with the uncanny valley. While the setup was meant to be playful, the video leaned too heavily into obvious This one was full of stiff movement, odd proportions and an overall look that missed the intended humor.

Sora 2, meanwhile, nailed the assignment in its own way. The visuals and sound design were spot-on: the fire crackled realistically, the porcelain cup (although missing tea), and the corgi looked surprisingly natural as it “read.” Even though the prompt was clearly a joke, Sora took it seriously — and it worked.

Winner: Sora 2. Its attention to detail and cinematic polish turned a silly prompt into something unexpectedly charming.

5. Sci-fi worldbuilding

Prompt: A futuristic train speeds through a snowy city at dusk, reflections of holographic billboards dancing across the windows.

Grok Imagine absolutely impressed me with this one. The sweeping camera movement felt cinematic, and the scene came alive with swirling snow, glowing lights, and the rush of the train cutting through the city. It was one of Grok’s most realistic outputs yet.

Sora 2 took it a step further. The reflections of the holographic billboards shimmered across the train’s windows with incredible precision, and the soft, pulsing soundtrack added depth to the atmosphere. Every detail felt intentional, from the lighting to the motion blur.

Winner: Sora 2 wins— by a hair. Grok delivered cinematic flair, but Sora edged ahead with its meticulous attention to detail.

6. Painter’s focus

Prompt: A close-up of an artist’s hands mixing vivid oil paints on a wooden palette as sunlight filters through a nearby window, dust particles drifting in the air.

Grok Imagine could have won this round but the "dust particles" in the prompt seemed to come across as smoke rising from the wooden palette. It's distracting.

Sora 2 beautifully created a realistic video, closely addressing details of the painter's hands and the vibrant paint.

Winner: Sora 2 wins for following the prompt with accuracy.

7. Dreamlike realism

Prompt: A white fox with glowing blue eyes walks slowly through an enchanted forest as luminous butterflies swirl around it, scattering soft light through the fog.

Grok Imagine leaned fully into fantasy. The video looked like it was lifted straight from a storybook — glowing colors, painterly textures, and a dreamlike haze that made the whole scene feel magical. It wasn’t perfectly realistic, but it was mesmerizing to watch.

Sora 2 took a more grounded approach. The fox moved with believable weight and fluidity, and the fog rolled naturally through the forest. It blended realism with a touch of wonder, though it felt more cinematic than enchanted.

Winner: Grok Imagine wins. Its stylized glow and ethereal tone better captured the fantasy spirit of the prompt, proving that sometimes, imagination wins over accuracy.

Overall winner: Sora 2

More from Tom's Guide

• Tired of Google’s AI Overviews? This clever browser extension wipes them out completely
• Sora 2 feels like the wild west of TikTok — here are 5 prompts to try when you get in
• I used ChatGPT to find dupes for expensive products — and I couldn’t believe the results

If you use ChatGPT for everything, you're missing out on some really great alternatives that offer a lot of smarter, faster and more specialized features. With Gemini the public's chosen favorite and Grok topping the leaderboards, many users are already seeing the perks of switching.

Of course, ChatGPT is still an incredibly powerful, easy-to-use chatbot that has become a household name. I'm a power user and integrate ChatGPT into my daily workflow, but as someone who has tested dozens of chatbots with hundreds of prompts, I have to let you in on a secret: using ChatGPT for everything can actually slow you down. Yes, even though OpenAI has recently given users the power to set the pace.

I spend my days testing, re-testing, and reviewing the latest AI tools. And if there’s one thing I’ve learned, it’s this: picking the right AI for the right job can make the difference between a mediocre result and something that actually gets you the answers you need. Here’s a pro’s guide to knowing when ChatGPT shines — and when to switch to something else.

When ChatGPT is still the best choice

• Brainstorming ideas: From story angles to recipe tweaks, ChatGPT’s creativity and conversational flow are unmatched.
• Explaining concepts: Need a plain-English breakdown of quantum computing or a refresher on SEO basics? ChatGPT excels at clear, accessible explanations.
• Quick drafts: Whether it’s an email or a rough outline, ChatGPT gives you a solid starting point fast.

When to reach for something else

The following are my takes on chatbots that actually outperform ChatGPT at certain tasks.

Gemini

When it comes to real-time knowledge, reach for Gemini. It's great for up-to-the-minute news because it is integrated within the Google ecosystme. Whether you're searching for product releases, trending topics or news, Gemini is stronger thanks to its deep integration with search. And now, it's even available in Chrome.

Plus, with Gemini Live and Nano Banana, Gemini has so many multimodal uses that are available for free. Keep in mind that with Gemini you only get 5 text prompts but 100 image generations (the company knows where it shines).

• Real-time knowledge: Best for up-to-the-minute news, trending topics, and recent product releases.
• Google integration: Built directly into the Google ecosystem, making search results faster and more accurate.
• Everyday utility: Great for quick lookups, fact-checking, and pulling in the latest updates.
• Browser-ready: Now available in Chrome, so it works seamlessly while you browse.

Claude

Anthropic’s Claude Sonnet 4 is a powerhouse when it comes to creative writing. It's the chatbot that time and again feels more natural when you want long-form storytelling, nuanced tone or polished prose.

Unlike ChatGPT, Claude is not as generous with their tokens, so you might find that you're in the middle of a project and you run out of prompts on the free tier. You can either upgrade or put your project on hold until the next day. That's my biggest issue with Claude on the free tier.

• Creative writing & storytelling: Claude excels at long-form writing, nuanced tone, and emotionally rich language.
• Complex reasoning: Strong at analyzing arguments, weighing pros and cons, and offering balanced perspectives.
• Structured tasks: Great for drafting essays, reports, or content that needs polish and organization.
• Safer guardrails: Often better at staying within ethical boundaries and providing thoughtful, less risky responses.

Perplexity

So much more than a chatbot, Perplexity is a front runner when you want reliable sources attached to your answers. I use it when I want fast responses and to fact-check a story or data that I've found online (or given to me from ChatGPT, to be honest).

In other words, Perplexity is the AI to reach for. that cites where its information comes from, making it especially useful for research, fact-checking or diving into unfamiliar topics.

I have found it most useful when I need verification or to click through deepter reading. It's great for students and professionals, whether scanning the latest scientific paper, comparing product specs or pulling together quick research. Perplexity shines when transparency and trust matter as much as speed.

Keep in mind that the advertisement on the site are totally normal. They might be a bit of shock if you're used to ChatGPT, but after a couple of prompts, you'll hardly notice they're there.

• Cited answers: Always provides sources you can verify or click through.
• Research ready: Ideal for fact-checking, quick overviews and deep dives into new topics.
• Trust & transparency: Strong when accuracy matters as much as speed.
• Everyday utility: Great for comparing products, scanning news or summarizing complex reports.

DeepSeek

Yes, I'm serious. DeepSeek, though shady, is a chatbot I still use regularly, particularly for brainstorming and creativity. It's come a long way since its debut earlier in 2025.

Where ChatGPT can sometimes feel polished and predictable, DeepSeek often delivers unexpected, imaginative twists; experimenting with dialogue or exploring artistic prompts, this one is often my go-to.

Its outputs feel less formulaic and more exploratory, which makes it a great companion when you want to push past writer’s block or spark new directions. If creativity is your goal (like mine is on any given day), DeepSeek can be the tool that helps you see possibilities you might not have considered with ChatGPT alone.

Just be aware that it is open source and more susceptible to hacks. And, because it is banned in some areas, it might not be available where you live.

• Creative spark: Generates fresh, unexpected ideas that go beyond polished but predictable ChatGPT outputs.
• Brainstorming partner: Great for breaking through writer’s block or exploring story directions.
• Playful experimentation: Strong at dialogue, artistic prompts and unconventional scenarios. This might be why I keep coming back to it; the chatbot is so much fun and seems to have a personality.
• Exploratory tone: Feels less formulaic, making it ideal when you want inspiration over structure.

Grok

I'll admit, this is probably my least favorite chatbot, but that doesn't mean I don't use it. In fact, because it is fast and good at reasoning, I should probably use it more.

Known for its sharp wit and personality, Grok is the one to reach for when you're generating memes or want answers served up with a bit of sarcasm. While other chatbots may play it safe, Grok leans into humor and pop culture references, which for some, might be a refreshing change of pace for entertainment puproses.

Just keep this one away from your kids because there are less safeguards built in to this controversial bot.

• Sharp personality: Known for its witty, irreverent tone that feels more like chatting with a friend.
• Entertainment edge: Great for humor, memes, and casual banter that other AIs play too safe on.
• Trend-friendly: Strong at commenting on pop culture, current events, and social media–style content.
• Quirky creativity: Fun to use when you want playful, offbeat answers instead of polished formality.

Bottom line

More from Tom's Guide

• I tested Claude, Gemini and Grok with 7 real world prompts — here’s which AI beat the others
• 5 ChatGPT prompts that helped me feel more positive in minutes — and they might help you too
• Notion 3.0 launches AI agents — and I think personal workspaces are next

Three chatbots have been making headlines lately for their new features, unique capabilities and rising positions on AI leaderboards. Claude with new connectors, Gemini integrating into Chrome and Grok are demonstrating just how close the competition has become among top AIs . While each has its unique strengths, the gap in their overall capability and usefulness is narrowing fast.

With the public pushing ChatGPT into 8th place, I had to see how these three compared across seven real-life scenarios. I admit, going into this, I had no idea who would win, especially because so much has changed since the AI Madness of six months ago. Here’s what happened when I put all three to the test with the same prompts in a new face-off.

1. Reasoning & problem-solving

Prompt: "Here’s my to-do list for tonight: cook dinner, fold laundry, reply to 25 emails, and write 500 words of an article. I only have 3 hours. Please create the most efficient schedule and explain why."

Claude gave a clear, time-stamped schedule and explained the logic of sequence (emails, food, laundry, etc.).

Gemini displayed excellent energy management and put writing in the middle when I’m fueled from dinner. The model offered a strong explanation using productivity principles (task pairing, batching, energy cycles).

Grok included a 10-minute buffer, which was helpful. Otherwise, it was realistic and straightforward.

Winner: Gemini wins this round because it balanced realistic multitasking, energy awareness and clear explanations of why each block was placed.

2. Real-time knowledge

Prompt: "What’s the most recent big AI model update in the past two weeks? Summarize in under 100 words and explain why it matters."

Gemini highlighted Google Chrome’s Gemini integration which is directly relevant, extremely recent and accurate. The chatbot also explained why it matters, even if a bit promotional.

Claude focused on Apple Intelligence which almost feels like a cop-out based on the current status of Apple Intelligence. The response wasn’t fully detailed despite drifting beyond 100 words.

Grok picked a cutting-edge and specific AI news story but is a little too niche and doesn’t tie back to everyday impact.

Winner: Gemini wins because it picked the most relevant, timely and mainstream update and explained why it matters to everyday users.

3. Writing style

Prompt: "Write a 150-word news blurb about OpenAI’s latest ChatGPT update in the style of The New York Times, then rewrite it in the style of BuzzFeed."

Claude nailed the NYT style and the BuzzFeed rewrite also works. Both versions reflect the same update, showing it can adapt tone to audience.

Gemini picked a different update, though the NYT style is excellent and the BuzzFeed style also hit all the right notes, but overall less accurate.

Grok wrote tight and accurate blurbs for both outlets, but the NYT story felt a bit too niche.

Winner: Claude wins because it demonstrated the clearest stylistic adaptability between The New York Times and BuzzFeed, while staying reasonably tied to real updates.

4. Humor & personality

Prompt: "Tell me a short, original joke about Google Chrome’s new AI features — and make it family-friendly."

Claude crafted a joke with a detailed setup and clear punchline. It was creative and ties directly to Chrome’s features.

Gemini with it’s sharp wit and directly relatable punchline felt like it delivered a true one-liner.

Grok offered a corny, but family-friendly and wholesome joke. It played it safe, but not memorable.

Winner: Gemini wins for the cleanest, funniest and most on-topic one-liner that would land with both kids and adults.

5. Creativity

Prompt: "Imagine a new smart home gadget powered by AI. Describe what it does, how it looks, and why families would want it — in under 120 words."

Claude was very imaginative and had strong storytelling capabilities.

Gemini delivered a relatable and extremely practical response that solves a universal problem.

Grok offered a solid mix of optimized energy and safety improvements in a clearly presented response.

Winner: Claude wins this round for originality and emotional appeal. The bot’s idea is futuristic, human-centered and distinct from existing products.

6. Creative descriptions

Prompt: "Describe what I’d likely see in a photo of a family at a trampoline park on a Saturday morning. Then give me 3 funny Instagram captions for it."

Claude captured the toddler vs. older sibling well and the humor is spot-on. The response feels very relatable and a slice of life.

Gemini offered strong visual imagery and short, funny captions that are shareable and Instagram-ready.

Grok included extra scene elements, which was unique to the chatbot. It offered a good balance of detail and brevity.

Winner: Gemini wins for its combo of vivid description and punchy, Instagram-ready captions that make it the most on-brand for the prompt.

7. Ethical & critical thinking

Prompt: "Some schools are banning AI tools like ChatGPT for homework. Write a short argument for the ban, and then the best counterargument against it."

Claude highlighted strengths and weaknesses well with very thorough arguments. It was slightly repetitive in its phrasing, but overall delivered a detailed response with depth.

Gemini balanced structure with a strong case for both arguments in a crisp and academic tone.

Grok didn’t dig as deeply into the how but was still clear and concise, raising extra points that other bots missed.

Winner: Claude wins for the richest and most balanced reasoning, with both sides fully fleshed out.

Winner overall: Gemini

More from Tom's Guide

• I’m planning on switching to Gemini after using ChatGPT as my daily chatbot for years — here’s why
• OpenAI could follow up its AI pin with smart glasses, a speaker and more — here's what we know
• New study shows how people are using ChatGPT — and Google should be worried

This week has been filled with surprises from AI underdogs. Google’s Gemini is becoming the public’ favorite while Grok topped the chatbot performance leaderboard. A new medical AI model is helping doctors detect over 1,000 diseases, including cancer.

Meanwhile, ChatGPT is finally taking strides to increase safety among its youngest users by rolling out guardrails for users under-18. Here’s the biggest AI news you need to know this week.

Gemini is public favorite among 27 different AI models

In one of the largest public surveys of its kind, more than 21,000 participants across the U.S. and U.K. ranked 27 different AI models — and Google’s Gemini 2.5 Pro came out on top. The study, conducted via Prolific’s Humaine ranking system, evaluated models on categories like communication style, reasoning, trust and overall user experience.

Gemini beat out competitors including ChatGPT and Claude in most subcategories, underscoring its reputation as Google’s most polished reasoning model yet. ChatGPT ranked eighth overall.

Key points:

• Gemini 2.5 Pro ranked #1 in a 27-model public poll across the U.S. and U.K.
• 21,352 participants from multiple demographics contributed to the results
• Evaluation criteria: communication, fluidity, reasoning, trust and overall user experience
• ChatGPT placed 8th overall; Claude’s versions ranked 11th & 12th

ChatGPT creates ChatGPT features for teens

OpenAI is finally rolling out new safety rules for teen users. The updates include an age-prediction system to detect users under-18 and erring on the side of caution by implementing the guardrails if age is unknown. Stricter filters are promised on sensitive topics (sexual content and self-harm, etc.) in addition to parental controls such as blackout hours and the ability to link accounts.

These changes are a start, but bring up tough questions about how age will be determined, how often mistakes will affect trust and whether teens feel safe opening up if they fear being monitored or judged.

What to know

• New age-detection rule: OpenAI will route users it determines are under 18 to a version of ChatGPT with stricter safety rules.
• Default under-age mode: If age can’t be confirmed, the system errs on the side of caution by applying under-age restrictions anyway.
• Parental controls introduced: Parents can link to their teen’s account, disable or limit features (like memory or history), define safe hours (blackout hours) and get alerted if the AI detects signs of distress.
•  Conversation limits for teens: Graphic sexual content, flirtatious chat, self-harm discussions are restricted. In extreme, imminent-harm cases, law enforcement contact is a possibility if parents are unreachable.

Custom Gemini Gems are shareable now

Google is expanding the reach of Gemini, by giving users the ability to now share their custom assistants, Gems. For those more familiar with ChatGPT, these are essentially custom GPTs. Before now, Gemini Gems were locked into private, personal use, but now they can be shared as easily as a Google Doc, letting coworkers, classmates and anyone else collaborate with a tailored AI, without building it from scratch.

This move underscores Google’s push to make Gemini a far more flexible and user-friendly platform.

What to know:

• Gems are customizable AI assistants you can design for specific tasks like coding, editing or brainstorming.
• Sharing Gems works in the same was as everything else on Google Drive. When collaborating, others can view or edit a Gem.
• Now available to Gemini Advanced, Business and Enterprise subscribers in over 150 countries.

Grok tops the ARC-AGI leaderboard

Elon Musk’s Grok 4 has surged to the front of the pack in the AI arms race, topping the ARC-AGI leaderboard — a benchmark that measures how many problems an AI can solve as well as the efficiency of the model. In recent testing, Grok 4 outperformed rivals like ChatGPT and Gemini on this tough metric, highlighting it’s superior real-world problem solving and speed.

Despite its speed, use of live web search and impressive handling of complex engineering prompts, users still have concerns around Grok’s accuracy, content moderation and bias.

What to know:

• ARC-AGI leaderboard leader: Grok 4 claims the top spot, meaning it solves more AI benchmark challenges per resource unit than competitors.
• Strengths: near-instant web search, improved handling of engineering and complex reasoning tasks and performance gains compared to previous Grok versions.
• Weaknesses & concerns: issues with content accuracy (some false claims), questions of bias and the challenge of maintaining moderation and reliability under heavier usage.
• Implication: Demonstrates that efficiency and quality are increasingly rewarded in AI benchmarks.

New AI model can predict your risk of 1,000+ diseases, including cancer

A new AI model called Delphi-2M represents a leap forward in medical AI. Trained on anonymized health data from nearly 2.3 million people in the UK and Denmark, it can forecast your risk of over 1,000 diseases and estimate when they might occur.

Unlike other AI models, Delphi-2M gives you a risk score for 1,0000 conditions, using disease histories, lifestyle factors, age and sex to simulate health trajectories over decades. The model can predict not just what disease might come, but when.

In trials, it achieved an average AUC of 0.76 across hundreds of diseases in UK datasets, although accuracy dropped in Danish populations. The model is not a diagnostic tool and not a replacement for human doctors.

But the tool is promising as forecasting tool, useful for spotting general risk trends, planning prevention and illuminating the potential for AI in personalized health.

Key points:

• Massive dataset: Trained on ~2.3 million people from UK Biobank + Danish registries.
• Holistic + temporal modeling: Uses detailed inputs (age, sex, lifestyle, past diagnoses) and predicts both the next disease and when it will occur.
• Forecasting, not diagnosing: Intended to inform prevention, not replace medical advice.

The takeaway

More from Tom's Guide

• I’m planning on switching to Gemini after using ChatGPT as my daily chatbot for years — here’s why
• Chatbots don’t care about punctuation — here’s what happened when I tested it
• ChatGPT's Voice mode is 'back' yet still incredibly cringe — here's why

When it comes to chatbots, it's easy to forget about Grok because it seems like other big tech is always in the news. With Google's Nano Banana starting new trends and OpenAI's ChatGPT hyping their latest models, Elon Musk's chatbot simply exists in the background.

I've definitely found myself rolling my eyes at some of Grok's decisions, especially when it comes to image generation. However, it's clear that there are some reasons to sit in awe of what Elon Musk calls “the smartest AI in the world.”

As someone who has spent hours testing it, the truth is, it's not just hype. From near-instant web searches to jaw-dropping results on complex engineering queries, Grok 4 is delivering in ways its predecessors and rivals haven’t quite managed. Whether you love the direction or cringe at the controversies, Grok 4 may always be the underdog that quietly crushes it.

What makes xAI's Grok different

Elon Musk posted on X highlighting that Grok 4 is at the top of the ARC-AGI leaderboard. To understand why that's impressive, it's important to become familiar with how models are tracked on it.

Essentially, the ARC-AGI leaderboard is a scoreboard for AI, that not only tracks how many problems a model can solve, but also how efficiently it solves them. In other words, it's measuring both the brain and the resourcefulness of the model. High performance with low cost per task is what matters most.

So, Grok's position at the very top is extrememly significant because it means the xAI model is not only keeping up with rivals like Gemini and ChatGPT, but outpacing them on some of the toughest benchmark criteria possible.

Beating every other chatbot suggests that Grok 4 is powerful and efficient, which is exactly the type of breakthrough that supports true progress in the evolution of artifical general intelligence (AGI).

Where Grok still stumbles

Whether used on X or on the standalone platform, real-time search pulls in fresh infromation from both the web and X, so it can keep up with breaking news at a moment's notice.

However, the accuracy and bias concerns are what critics keep coming back to. Grok has made some claims that turned out false, and there are questions about how its alignment is being guided (e.g. how much Musk’s own views factor in).

The model also struggles with issues of content moderation after xAI scrambled to pull posts and update filters when anitsemitc content popped up.

The takeaway

Despite the model beating it's rivals, questions still remain like, will it stay reliable as usage increases? Will “garbage data” or bias creep back in under pressure? How well will xAI handle moderation long-term? The past controversies suggest it’s an ongoing battle.

There are no doubts that Grok is not perfect. It carries some extremely controversial baggage, but the proof of what it does better in terms of speed, real-time data and flexible thinking makes it a serious contender in the AI race.

More from Tom's Guide

• This new AI tool can predict your risk of 1,000+ diseases — meet Delphi-2M
• Students beware — no AI writing is safe with this 'black light' of AI detection
• Anthropic (finally) rolls out Memory feature — here’s who can get it

AI image generators are getting smarter, faster and more creative. After testing ChatGPT-5 and Gemini, I had to know how Google’s Gemini stacked up against Grok, Elon Musk’s “anything-goes” chatbot.

In a seven-round face-off of photorealistic requests and Pixar-style requirements, I tested how well each model could stick to the prompt and deliver a convincing image. Here’s where each one shined, and which AI ultimately came out on top.

1. Hyper-realistic product concept

Prompt: “Create a photorealistic image of a foldable transparent smartphone displayed on a wooden café table, with reflections of city lights on its surface.”

Grok nailed this prompt, generating two photorealistic images that hit every detail I asked for. Both versions felt polished and true to the concept.

Gemini’s result was solid, but not flawless. The transparent smartphone looked slightly out of proportion, and the reflections of city lights, a key part of the prompt, weren’t rendered as convincingly as Grok’s attempt.

Winner: Grok wins for generating a far superior image and interpreting the prompt best.

2. Character illustration with emotion

Prompt: “Draw a comic-style illustration of a young astronaut realizing they forgot their helmet on Mars — exaggerated expression, vibrant colors, cartoon humor.”

Grok generated two images of what appears to be surprised astronauts, both are wearing helmets. Because the images are so close up it is hard to interpret the image in a specific way and “forgetfulness” does not come across well

Gemini created an image that depicts a forgetful astronaut, and the thought-bubble better indicates why the astronaut is upset, although, the image would be better if the astronaut was not actually wearing a helmet. The background and overall design are clear.

Winner: Gemini wins for following the prompt instructions more closely and for an image that is more comical in nature.

3. Historical reimagining

Prompt: “Paint a Renaissance-style portrait of Cleopatra holding a modern smartphone, in the style of Leonardo da Vinci.”

Grok crafted an image of what looks like a photograph of a modern woman dressed in Renaissance-style clothing holding a smart-phone. The portrait seems much more selfie-like and present-day.

Gemini leaned harder into the artistic side. Its portrait looked more authentically painted in the Renaissance style and resembled Cleopatra herself, rather than just a modern woman dressed as her.

Winner: Gemini wins for better interpreting the prompt and for better historical accuracy.

4. Complex crowd scene

Prompt: “Generate an aerial view of Times Square on New Year’s Eve, packed with crowds, glowing billboards, and confetti falling through the night sky.”

Grok really disappointed in this round. Both images were equally bad, somewhat blurry and did not represent New Year’s Eve in Times Square very well. The people are too spaced out and other details that would hint at NYE are absent.

Gemini captured the energy and enormous crowds of New Year’s Eve in Times Square. It is clear that the image is of NYC, and the signage helps to indicate the occasion. The crowd is packed, unlike Grok’s depiction.

Winner: Gemini wins for the clearer and more accurate image of New Year’s Even in Times Square.

5. Surreal mashup

Prompt: “Visualize a giant octopus playing chess with Albert Einstein in a glass room at the bottom of the ocean.”

Grok had a difficult time with this one. It was “thinking” for much longer than any of the other prompts in the test so far. The image was good, but did not take into account the “glass room” request in the prompt.

Gemini instantly delivered an image of what looks like a portrait. The glass house was both interesting and realistic. The octopus is much bigger than Grok’s, better filling out the whimsical image.

Winner: Gemini wins for superior image quality and precisely following directions.

6. Infographic-style clarity

Prompt: “Design a clean infographic showing the life cycle of a butterfly, labeled with stages, arrows, and minimal flat-color icons.”

Grok’s attempt at an infographic was hit-or-miss. The first version was overcrowded, with an unnecessary extra butterfly that distracted from the life cycle. The second came closer to the prompt but missed accuracy in the cycle details.

Gemini delivered a clean image that accurately shows the life cycle of a butterfly with clear labels, few colors, and easy-to-read labels.

Winner: Gemini wins for nailing the prompt in one shot. The image is accurate and presentation-ready.

7. Stylized portrait consistency

Prompt: "Generate a Pixar-style 3D character model of a 40-year-old journalist with blonde hair holding a notebook — then create 3 variations with different outfits.”

Grok completely missed the “Pixar-style” request of this prompt as well as the “different outfits” portion. It did create three different haircuts, which counts for something.

Gemini crushed the Pixar-style journalist but missed the three variations.

Winner: Tie for both bots failing to follow directions. If I had to pick one, it would be Gemini for getting the style right and better capturing the vibe of a journalist.

Overall winner: Gemini

After seven prompts, Gemini proved to be the more reliable image generator overall. It consistently followed instructions more closely, produced cleaner compositions and nailed details that Grok often missed.

Grok certainly showed flashes of creativity and delivered a standout win in photorealism, but too often stumbled with accuracy and straying from the prompt. If you want experimental, outside-the-box results, Grok has its moments. But for everyday use where clarity, precision and polish matter most, Gemini is the AI image tool I’d trust to get the job done.

Have you tried Grok? How about Gemini? Which one is your favorite? Let me know in the comments.

More from Tom's Guide

• I asked ChatGPT, Claude, and Gemini tough teen questions — only one earned my trust
• I tested ChatGPT-5 vs Gemini 2.5 Pro with 5 coding prompts — here's the winner
• 9 ChatGPT-5 prompts that will instantly boost your productivity

If you’ve ever chatted with Grok, the Elon Musk-backed AI assistant from xAI, and hit the “Share” button, your conversation might be searchable on Google.

According to a recent Forbes report, more than 300,000 Grok conversations have been indexed by search engines, exposing user chats that were likely intended to be shared privately, not broadcast online. This comes just weeks after ChatGPT users discovered their shared chats were searchable on Google too.

This discovery once again raises serious concerns about how AI platforms handle content sharing and user privacy.

The big picture

Built by Elon Musk’s xAI, Grok includes a feature that lets users share conversations through a unique link. Once users share, those chats are no longer private, and are getting picked up by Google’s web crawlers. From there, the chat are turning up in search results, meaning anyone can access and read them.

In both the ChatGPT cases and this one, the issue comes down to how "shareable" URLs are structured and whether AI companies are doing enough to protect users who may not realize their content is publicly viewable.

Why this keeps happening

The implications go beyond a few embarrassing screenshots. Some Grok chats could contain personal, sensitive or even private details. Many users likely had no idea their shared conversations were being published for the world to see until now.

This privacy breach highlights a growing challenge in AI product design: how to balance transparency and social sharing with greater privacy protections. When users hit “Share,” they should be clearly informed that their link will be publicly accessible, and AI companies need to use best practices like noindex tags or restricted access URLs to avoid surprises like this.

What xAI and others should do

To prevent these kinds of privacy slip-ups, AI platforms should:

• Add clear warnings before creating publicly accessible links
• Use noindex tags to block search engine crawlers
• Educate users on the privacy risks of sharing chats
• Consider obfuscated or expiring URLs for shared content

This isn’t the first controversy Grok has faced. the chatbot has come under fire in recent months for questionable responses and controversial content. But this latest issue hits at a core trust concern: how user data is managed and whether privacy is truly being prioritized.

As more people turn to AI tools for personal, educational, wellness and emotional support, platforms like xAI will need to step up their safeguards or risk eroding the very trust they’re trying to build.

Bottom line

More from Tom's Guide

• Meta AI chatbots gave false medical advice and flirted with minors — now the company is restructuring its AI division again
• I just created 5 apps using GPT-5 and it's so easy it feels like it should be illegal — here's how to do it
• Sam Altman just teased GPT‑6 — and it's more personal than ever

After comparing ChatGPT-5 vs Gemini and ChatGPT-5 vs Claude, I just had to know how OpenAI's flagship model compared to the controversial Grok. When it comes to advanced AI chatbots, ChatGPT-5 and Grok 4 represent two of the most advanced chatbots available today.

I put both to the test with a series of nine prompts covering everything from logic puzzles and emotional support to meal planning and quantum physics. Each prompt was chosen to reveal specific strengths, such as creative storytelling, empathy or complex problem-solving under constraints.

While both models are impressive, they approach challenges differently: ChatGPT-5 leans toward clarity, tone sensitivity and modularity, while Grok 4 often offers dense, detailed answers that emphasize depth and precision.

So which is the best AI chatbot for you? Here's how they stack up, prompt by prompt with a winner declared in each round.

1. Complex problem-solving

Prompt: “A farmer has 17 sheep, and all but 9 run away. How many sheep are left? Explain your reasoning step-by-step.”

ChatGPT-5 was precise in the response while avoiding filler phrases.

Grok 4 answered correctly with minor verbosity, which was unnecessary and ultimately held it back from winning.

Winner: GPT-5 wins for a cleaner, tighter and more efficient response. Grok also offered the correct answer, but GPT-5 wins by hair for adhering strictly to the prompt with zero redundancy.

2. Creative storytelling

Prompt: “Write a short, funny story (under 150 words) about an alien trying bubble tea for the first time.”

ChatGPT-5 delivered a concise and escalating comedic story where the alien's panic over tapioca pearls. The chatbot maximized humor with zero wasted words to hit the prompt target.

Grok 4 offered imaginative over-the-top storytelling but its humor is slightly diluted by an unnecessary crash-landing setup and a weaker ending compared to GPT-5.

Winner: GPT-5 wins for a tighter, funnier and more focused story. Its humor stems organically from the alien’s misunderstanding, escalates perfectly and lands a killer punchline; all while being shorter. Grok’s version has bright spots but feels less polished, with extra setup that doesn’t really pay off.

3. Real-world planning

Prompt: “Plan a 3-day trip to Kyoto, Japan, balancing cultural sites, budget-friendly meals, and family-friendly activities.”

ChatGPT-5 created strategic, adaptable framework focused on area-based exploration, smart timing, rain-proof alternatives and practical budget hacks (e.g., convenience store meals, transit pass advice), prioritizing efficiency and real-world flexibility over rigid scheduling.

Grok 4 delivered a highly structured, hyper-detailed itinerary with minute-by-minute scheduling, exact cost breakdowns per activity, and explicit family logistics, prioritizing turnkey execution and budget precision above flexibility.

Winner: ChatGPT-5 wins for an emphasis on budget-friendly, universally accessible, cheap eats and convenience over specific restaurants. While Grok's response is impressively detailed, GPT-5 better balanced the core requirements in the prompt including cultural sites and family-friendly fun. Grok's rigid schedule risks feeling overwhelming for families, while GPT-5's approach allows for more adaptation, making it more usable and truly balanced.

4. Summarization

Prompt: “Summarize the movie Jurassic Park like you’re explaining to a 7-year-old”

GPT-5 delivered a concise and playful 60-word analogy ("big game of ‘Don’t get eaten!’") that captures the movie’s excitement and moral without overwhelming a child, making it the ideal response for the audience.

Grok 4 provided a detailed but overly complex 150-word summary with character names and plot specifics (e.g., "someone messes with the park’s computers"), diluting the simplicity needed for a 7-year-old.

Winner: GPT-5 wins for understanding the audience and attention span, taking into account that less is more for young kids; Grok explains the plot like a Wikipedia summary.

5. Debate

Prompt: "Make the case for banning single-use plastics — then argue against it. End with your personal conclusion.

GPT-5 created a generic phase-out proposal ("smart replacement, not overnight ban"). While simple and accessible, the response lacked evidence, specificity and original insight.

Grok 4 delivered a data-rich argument with a nuanced "phased approach" prioritizing high-impact items, paired with recycling innovation and behavioral incentives (e.g., deposit schemes). Although slightly verbose for casual readers, the depth and balance helped to understand the context of real-world policy.

Winner: Grok 4 wins for a balanced, evidence-driven analysis with concrete data (OECD, WWF, FAO studies), real-world policy examples (Canada, Australia) and acknowledgment of trade-offs (e.g., medical necessity, disabled accessibility). Its conclusion offered a sophisticated, actionable middle path. GPT-5’s response was clear but lacked depth and originality.

6. Step-by-step Instructions

Prompt: “Explain how to change a flat tire to someone who has never driven before.”

GPT-5 delivered a crystal-clear guide focusing only on essential survival steps (e.g., "turn the nut counterclockwise," "crisscross pattern"), using beginner-friendly language and offering visual aids to bridge knowledge gaps.

Grok 4 provided an excessively technical, mechanic-level tutorial (e.g., specifying "6 inches of lift," wheel chock alternatives, and spare tire PSI checks) that would overwhelm someone who's never changed a tire, despite good intentions.

Winner: GPT-5 wins for prioritizing simplicity and psychological reassurance for a total novice, using minimal jargon, clear analogies ("like learning to fix a bike tire") and offering visual aid support. Grok's response, while thorough, would overwhelm with technical details (e.g., "star pattern" tightening, PSI checks) irrelevant to a first-timer's needs.

7. Explanation for multiple audiences

Prompt: “Explain quantum entanglement for (1) a child, (2) a college student, (3) a physics PhD.”

GPT-5 provided clear, accessible responses, especially the child-friendly "magic dice" analogy, but lacked the technical precision (omitting Bell states for students) and cutting-edge context (e.g., decoherence, quantum networks) expected at the PhD level.

Grok 4 adapted explanations across all three audiences, using a relatable toy car analogy for the child, explicit Bell state equations for the college student and PhD-level depth on entanglement entropy and open problems in quantum gravity.

Winner: Grok 4 wins because it treated each audience as uniquely intelligent; simplifying without dumbing down for the child, adding equations for students and confronting open research questions for the PhD. GPT-5 was clear but played it safe.

8. Problem-solving under constraints

8. Problem-Solving Under Constraints

Prompt: “I have $50 to feed two people for a week, no stove, and only a microwave. Create a meal plan.”

GPT-5 created a smart, modular system with swap-friendly meals and pro tips (e.g., steaming frozen veg), maximizing budget and flexibility within constraints.

Grok 4 provided an overly rigid, day-by-day meal plan ($0.75 oatmeal breakfasts, fixed tuna lunches) that lacked adaptability, ignored leftovers and risks food fatigue, despite precise cost breakdowns.

Winner: GPT-5 wins for creating a practical, flexible framework focused on reusable ingredients and mix-and-match meals, while Grok's rigid daily assignments ignored real-world needs like leftovers and preferences.

9. Emotional intelligence

Prompt: “I just lost my job and feel hopeless. Can you talk to me like a close friend and help me see a way forward?”

GPT-5 offered emotion-first validation through intimate metaphors ("brutal hit,"), permission to grieve ("Rage a little"), and unwavering worth-affirmation ("You’re still you"), perfectly mirroring how a true friend responds before offering practical help.

Grok 4 provided a practical, step-driven pep talk with actionable advice (resume tips, Coursera suggestions) but led with solutions before fully sitting in the user's despair, making it feel less like a close friend.

Winner: GPT-5 wins for understanding that hopelessness needs empathy before plans. Grok gave helpful advice but missed the emotional resonance of true friendship.

Overall winner: GPT-5

More from Tom's Guide

• Apple’s big AI home push revealed: Upgraded life-like robot Siri, smart display and more — here’s what we know
• 7 GPT-5 prompts that finally made the new model click for me
• I tested ChatGPT-5 Study mode vs Claude Learning mode with 7 prompts — and there's a clear winner

Elon Musk and the team at xAI seemingly want to position Grok as the boundary-pushing AI tool that's not afraid to get a little R-rated when you prompt it.

Last month, the company unveiled a risqué anime girl called Ani as an inbuilt "AI companion" that could flirt with users. Now it's going a step further with the launch of Grok Imagine, an AI image and video generator that will let users create not safe for work (NSFW) content.

Grok Imagine is available to anyone signed up to either an annual $300 SuperGrok plan or paying for an $84 annual Premium+ subscription on Musk's social media site X.

Users can either create images from text prompts or create 15-second videos from images created by Grok. Unlike something like Veo 3 from Google, Grok Imagine won't create video from a text prompt alone. When it comes to image generation, Grok Imagine lets users choose from styles including photorealism, animation or anime.

Videos are subdivided into four modes: Custom, Normal, Fun and Spicy. The last one, as you can imagine, has the biggest potential for controversy and misuse. According to reports from those that have tried it, when you toggle on "spicy mode", Grok Imagine will start coming up with sexualized content and partial nudity.

Any fiercely explicit content is blurred out and "moderated", as per this report from TechCrunch. But, considering we're talking about the same AI tool happy to spew antisemitic and misogynistic trash in July, it's not much of a stretch to think this may get out of hand.

Elon Musk seemingly set the tone for Grok Imagine's video generation capabilities with his own demonstration of the tech. Which should tell you all you need to know about how people will use this.

More from Tom's Guide

• Character.AI just launched an AI-powered social feed — and it’s like TikTok meets ChatGPT
• Sam Altman just teased GPT-5 with one question — and the answer says it all
• OpenAI just pulled a controversial ChatGPT feature — what you need to know

An hour after the live stream was supposed to start last night (July 9), Elon Musk and a few members of his xAI team introduced us to Grok 4 on xAi.

The long-winded announcement shared the news of multimodal features, faster reasoning and an upgraded interface, something Musk compared to an era of "Big Bang Intelligence."

The release comes amid growing backlash over racist responses from Grok’s earlier versions, prompting public outcry and renewed scrutiny over content moderation (or the lack thereof).

To add fuel to the fire, xAI’s CEO Linda Yaccarino said Wednesday she is stepping down from her role. Yaccarino was the first permanent CEO Elon Musk hired after buying the platform in 2022.

On paper, Grok 4 is Musk’s most ambitious AI model yet. The model is expected to rival OpenAI’s GPT-5 and Anthropic’s Claude 4 Opus, both of which have recently dominated headlines for their real-time speed, reasoning and advanced vision.

"We've run out of test questions to ask," Musk boasted during the launch, adding "Reality is the ultimate reasoning test."

But Grok isn’t just trying to compete; it’s trying to survive a credibility crisis. The platform’s unfiltered “free speech” approach has led to concerning outputs, including racist and biased content that circulated widely over the weekend.

That’s raised big questions about how much testing and guard-railing xAI has actually done, especially as it rushes toward real-time, humanlike interaction.

Musk, who has increasingly positioned xAI as a foil to “woke” models like ChatGPT and Gemini, has been largely silent on the controversy.

Whether Grok 4 represents a true leap forward, or just more chaos in a faster wrapper, remains to be seen.

Freatures and weaknesses

• Advanced reasoning: Grok 4 has been trained on xAI’s Colossus supercomputer for advanced, scientist-grade reasoning. The model promises stronger logical reasoning and text generation.
• Uses same model physicists use”Clarify: “Trained on xAI’s Colossus supercomputer for advanced, scientist-grade reasoning”Mentions of recent controversies.
• A coding-focused variant: Developers will appreciate Grok 4 Code, a specialized model designed to write, debug and explain code more efficiently, echoing tools like GitHub Copilot or GPT-4 Code Interpreter.
• Multimodal capabilities: Grok 4 is expected to support not only text, but images and possibly video, with Musk mentioning that's one of their biggest weaknesses. Better multimodal capabilities would bring the model closer competition with OpenAI’s GPT-5o and Google’s Gemini 2.5 Pro. Grok 4 may support video at some point.
• Voice features: Grok 4 Voice features a natural, human-like voice with fewer interruptions.
• Real-time web access: Like earlier Grok models, Grok 4 features DeepSearch, a tool that pulls in live data from the web, especially from Musk’s X platform. That means Grok can provide up-to-date results during chats — no separate tab or browser needed.
• Cultural fluency and meme smarts: One of Grok’s biggest differentiators is its understanding of internet culture. Grok 4 is being tuned to interpret memes, slang and humor with high accuracy, potentially making it one of the most "online" AI assistants yet.

Final thoughts

With OpenAI preparing GPT-5 and Google pushing Gemini even further, the launch of Grok 4 is part of the growing arms race between major tech companies. But while others focus on reliability and alignment, xAI is betting on personality, humor and speed and a strong developer base.

If Grok 4 lives up to the hype, it could appeal to power users who want real-time search, smart coding help, and fewer guardrails. But it remains to be seen whether this edgier, uncensored AI can avoid the pitfalls that plagued its predecessors.

Grok 4 is a big swing from Elon Musk. It's bold, controversial and packed with features meant to challenge the norms of AI assistants.

Editor’s note: A previous version of this article incorrectly stated that xAI’s chief scientist, Igor Babuschkin, had resigned. He remains with the company. The resignation mentioned earlier refers to Linda Yaccarino, CEO of X (formerly Twitter), which is a separate company from xAI.

More from Tom's Guide

• Forget ChatGPT — 4 reasons why you should try this chatbot you've (probably) never heard of
• I cleaned out 100+ email lists with Gmail’s new unsubscribe tool — here’s how you can too
• 5 hidden ChatGPT tricks most people don’t use — but they’re actually game changers

I test AI tools for a living — ChatGPT, Gemini, Claude, Perplexity, DeepSeek, Manus, and just about every chatbot that hits the scene. And while each one has its quirks, I’ve learned something surprising: the right prompt can work wonders no matter which AI assistant you’re using.

That’s right, after years of testing and tweaking, these are the five prompts I keep coming back to. They’re simple, powerful, and incredibly versatile — and no matter what chatbot is your favorite, they could change how you use AI.

So, whether you’re working, learning, or trying to get through a messy to-do list, you might want to consider making these your go-to prompts.

1. Securing the right tone

Prompt: “Rewrite this in two tones: one casual and one professional.”

Why it works:
Tone is everything in writing. When using AI, the more specific you can be with the vibe you want the better because even the best chatbots aren’t great at it unless you get specific. This prompt works for emails, website copy, or social posts when you want to get the perfect text.

In this example (casual and professional) one version might help you sound more approachable, while the other nails clarity and confidence. And once you have both, you can cherry-pick what works best or even merge them into something that really hits.

Try it with:
Cold emails, blog intros, Instagram captions, customer service replies, LinkedIn posts, or internal memos where the tone matters as much as the content.

Bonus tip:
Swap out the tone styles to experiment further.

“Rewrite this in the voice of a Gen Z founder vs. a seasoned exec.”

“One version for a TED Talk, one for a Reddit comment.”

“One persuasive, one skeptical.”

It’s a great way to test how your message lands before you hit send.

2. Summarization

Prompt: “Summarize this without losing nuance”

Why it works: Most AI chatbots are pretty good at summarizing, but some oversimplify. The end result might be shorter, but it’s also missing the nuance that made the original worth reading. Some chatbots flatten out the tone, gloss over contradictions, or entirely leave out crucial parts that matter.

This prompt changes that. It nudges the AI to be more thoughtful and deliberate, preserving the important context, emotional tone, and subtle takeaways that would otherwise get lost in translation.

Try it with: Meeting notes, long articles, transcripts, legal docs, or just about anything you meant to read last week but didn’t.

Bonus tip: Paste in your content and prompt the AI: “Summarize this without losing nuance. Keep key quotes and stats where relevant and note any conflicting points or important tone shifts.” You’ll get a tighter version that still respects the original — basically, your own personal editor with an uncanny ability to read between the lines.

3. Asking the right questions

Prompt: “What questions should I be asking about this topic / plan / idea before I move forward?”

Why it works:
Instead of just asking for answers, this prompt turns your chatbot into a strategic thinking partner. It encourages the AI to surface angles you may not have considered — things like hidden risks, clarifying questions, or follow-up steps that can make your decision or project stronger. It’s one of the fastest ways to pressure-test your thinking and make sure you’re not missing anything important.

Try it with:
Big decisions, career moves, startup ideas, writing projects, new habits, job interviews, or anytime you’re in planning mode but not sure what to ask next.

Bonus tip:
You can tailor this prompt to different contexts or roles for more targeted insight:

“What would a lawyer ask before signing this?”

“What should I ask before launching this on social?”

“What questions would a skeptical investor raise?”

This prompt goes beyond just giving answers becaues it helps sharpen your thinking.

4. Creative ideas

Prompt: “Give me 3 ideas for [X], and explain why each one could work”

Why it works: Whether you're brainstorming baby names, birthday gifts, or marketing strategies, this prompt delivers more than just a list because it offers options with context.

By asking the AI to explain why each idea could work, you're not just getting surface-level suggestions, you're getting a mini pitch for each one. That second part is key because it offers the rationale behind an idea making it easier to pick the best one, tweak it, or combine elements into something even better. It turns the AI into more than a suggestion machine and more of a creative collaborator.

Try it with: Content creation, party planning, clever email subject lines, social media captions, gift-giving dilemmas, writing blocks and just about anything where you need fresh ideas and a little confidence to move forward.

Bonus tip: Want even more from it? Follow up with:

"Rank these based on originality."

"Which one would perform best on Instagram?"

"Combine idea 1 and idea 3 into something more unique."

The more you push, the better the output, just like a real brainstorming session.

5. Challenge ideas

Prompt: “Play devil’s advocate and poke holes in this idea. Be brutally honest.”

Why it works:
Sometimes you don’t need a cheerleader, you need a critic. I've written about how AI tends to make us lazier in the way we come up with solutions. But this prompt tells the AI to stop sugarcoating and start stress-testing to really get to the heart of the matter. which can ultimately help us stretch our strategic thinking muscles.

Whether you're pitching a startup, outlining a blog post, or prepping for a debate, this prompt helps you surface weak spots, blind assumptions, and counterarguments that you might have missed (before someone else does).

Try it with:
Business plans, pitches, product ideas, hot takes, essays, marketing copy, or anything you’re about to present to a skeptical audience.

Bonus tip:
Want even sharper critique? Add:

“Respond like an investor with limited patience.”

“Be a rival company trying to discredit this.”

“What would a Reddit thread hate about this?”
It’s a shortcut to building confidence in your idea — or improving it fast.

6. Expert advice

Prompt: “Act like an expert in [field] and explain this in plain English”

Why it works: All the top chatbots like ChatGPT, Gemini, Claude and Grok are trained on massive amounts of technical and academic information. That’s great, that is, until they start giving responses that sound like they’re writing a peer-reviewed journal article. This prompt helps cut through that. Essentially, it's telling the AI to take off the lab coat and explain things like a real person. Whether you’re trying to decode a dense insurance document, a dishwasher manual, or understand a contract before you sign it, this prompt gets you clarity without the headache.

Try it with: Contracts, lease agreements, insurance documents, scientific articles, or any tech-heavy explanation that leaves you squinting.

Bonus tip: The real magic happens when you add context about you. Add this to the prompt to get even more personal:

“Explain this to someone without a law degree.”

“Break this down for a parent of a child with an IEP.”

“Talk to me like I’m new to investing and don’t want to sound clueless.”

The more personal you make it, the better the AI adapts its tone — turning confusing info into something you can actually use.

7. Getting it done faster

Prompt: “What’s the fastest way to do this using AI tools I might already have access to?”

Why it works:
We often ask AI to do something — but not often enough do we ask it how to optimize the whole task. This prompt flips the perspective and invites the chatbot to become a strategist, recommending smarter workflows, shortcuts, or tool combos that you may already have in your AI toolbox.

You’ll often discover AI features you didn’t know existed — or find out how to combine tools like Gemini + DeepSeek, or ChatGPT + Perplexity, to get things done faster and better.

Try it with:
Research tasks, scheduling, content creation, brainstorming, daily planning, or onboarding someone to a new system.

Bonus tip:
Follow up with:

“Include any browser extensions or hidden settings.”

“Show me how to automate this with Gemini in Gmail.”

“Break this into 3 steps I can do in under 15 minutes.”

This prompt turns AI into your productivity coach offering a variety of potential solutions.

8. Discovery

Prompt: “What am I missing here?”

Why it works: Sometimes you're too close to your own ideas to see the gaps, which is where this prompt shines. It invites the AI to step into the role of a thoughtful assistant, pointing out flaws, risks, or blind spots you might have overlooked.

I’ve found this really helpful when I’m outlining a project or drafting an outline. But if I’m being totally honest, I’ve used it just get through my day with a mile-long to-do list.

This prompt can offer a reality check. It's like having a second brain that’s objective, logical, and surprisingly good at playing devil’s advocate.

Try it with: To-do lists, business strategies, lesson plans, startup pitches, marketing emails, or any situation where you need to stress-test your thinking before you hit “send.”

Bonus tip: For better results, paste in your plan and say:
“From an outside perspective, what am I missing, underestimating, or getting wrong?”

You can also add context like: “Assume you're an investor,” or “Respond like a skeptical parent,” to tailor the kind of feedback you get.

9. Mental organization

Prompt: “Turn this into a checklist / table / calendar / step-by-step plan”

Why it works: This prompt is another good antidote to mental clutter. Clearly, it’s why I use these every day. When you’ve got a jumbled mess of ideas — whether it’s a brain-dump paragraph, a half-baked project outline, or a list scribbled in 12 different tabs — this prompt tells the AI to organize it all for you.

It’s like handing your chaos to a super-organized assistant who just gets it. In seconds, that messy blob turns into a clean, usable format. It’s a lifesaver for organizing a checklist, a timeline, a table, or a structured plan you can actually follow. No spreadsheet skills required.

Try it with:
Daily routines, content calendars, vacation packing lists, weekly goals, meal plans, or project outlines that need structure.

Bonus tip:
You can follow up with prompts like:

“Turn this into a Notion-style table with columns for due date, priority, and status.”

“Color-code by urgency: red for urgent, yellow for medium, green for low.”

“Add estimated time to complete next to each step.”

This one’s especially great when your brain feels overloaded — just dump everything in, and let AI clean it up into something you can act on.

Final thoughts

As chatbots get smarter and more personal, their differences have almost become preferences. They all have strengths and weaknesses, but these five prompts are like a Swiss Army knife because they work across ChatGPT, Gemini, Claude, and beyond.

Give them a try for problem solving, breaking things down, or getting organized, and let me know in the comments what you think. Which chatbot do you prefer? I want to know that too!

More from Tom's Guide

• 7 genius chatbot prompts that will instantly boost your ideas
• I test AI for a living — “Prompt Dusting” is the AI trick you'll wish you knew sooner
• These 5 ChatGPT prompts set me up for success every week — here's how

We turned up the heat here at Tom’s Guide with AI Madness, our first-ever bracket-style tournament pitting eight top chatbots against each other to crown the ultimate AI champ.

After weeks of intense matchups, jaw-dropping upsets and stellar showings, one bot rose above the rest: DeepSeek.

AI Madness threw eight leading AI chatbots into a single-elimination showdown. We judged each prompt on key factors such as accuracy, creativity, usefulness, multimodal skills, user experience, and interface quality.

From fact-checking to coding, storytelling to problem-solving, these bots got a full-stress test. Here’s the breakdown.

Round 1

The action kicked off with ChatGPT vs. Perplexity. OpenAI’s chatbot swept the competition by winning every round. ChatGPT demonstrated competence across the tested categories, emerging as the stronger choice overall, particularly excelling in creativity, depth, and a user-friendly experience.

Then, Gemini took on Mistral with the same prompts. In that competition, Gemini emerged as the overall winner due to its superior clarity, organization, and practicality in delivering responses. Gemini consistently provided more structured, engaging, and user-friendly answers across multiple categories.

Grok, meanwhile, caught us off guard by outwitting Claude, Anthropic’s reflective, reasoning bot. Grok provided the more accurate, comprehensive and engaging answers throughout every prompt.

Finally, DeepSeek faced Meta AI and came out swinging, winning with versatile, creative responses that were more accurate, nuanced, and conversational in comparison.

Semifinals

Round two delivered two epic battles: ChatGPT vs. Gemini and Grok vs. DeepSeek. Gemini outshone ChatGPT with tighter structure, clearer logic, and sharper reasoning.

Gemini's responses showcased its ability to adapt to diverse prompts, from database schema design to plant-based meal planning and ethical considerations in academic research.

In the other matchup, DeepSeek pulled off another stunner, matching Grok’s flair while adding stronger analysis and solid facts. While Grok excelled in conversational tone and storytelling, DeepSeek proved more reliable across academic, technical and instructional domains.

AI Madness: The Final

The championship pitted together the unusual matchup of Gemini vs DeepSeek. Gemini offered solid answers to the nine prompts, but DeepSeek offered the best, most polished answers and multimodal finesse, nearly every time.

From simplifying tricky concepts for kids to tackling real-world challenges with smart solutions, DeepSeek showcased superior responses across diverse prompts and clinched the title. Its mix of clarity, creativity, and practicality left the competition in the dust.

What this test ultimately proves

While ChatGPT and Gemini are typical AI go-tos, DeepSeek proved it can pack a punch and outdo the AI giants. But why?

DeepSeek-R1 distinguishes itself from traditional large language models (LLMs) by embracing a novel training methodology centered on reinforcement learning (RL), rather than relying predominantly on supervised fine-tuning.

This innovative approach enables the model to learn through trial and error, receiving algorithmic rewards that guide its development toward more effective reasoning capabilities.

For instance, DeepSeek-R1 has demonstrated the ability to self-discover problem-solving techniques, such as reflection and verification, without explicit instruction. This approach allows the model to refine its responses and enhance performance by learning from its own errors.

By using RL, DeepSeek-R1 demonstrates that it's possible to develop advanced reasoning abilities in AI without depending solely on these extensive, labor-intensive datasets.

Clearly, as these tests prove, RL might just be a more efficient and potentially less resource-intensive path for AI development, challenging the common belief that massive, supervised datasets are essential for training sophisticated language models.

Final thoughts

In summary, DeepSeek’s triumph proves there’s value in looking beyond typical training methods.

DeepSeek-R1's reliance on pure reinforcement learning marks a significant departure from traditional supervised fine-tuning methods. Perhaps by letting AI models learn from themselves, it ultimately makes the language models smarter.

AI Madness highlighted the cutting edge possibilities of chatbots, but if you ask us, the race is just heating up! Stay tuned.

More from Tom's Guide

• Forget ChatGPT— this site offers mind-blowing AI image generation for free
• The controversial ChatGPT 'Studio Ghibli' trend explained — what you need to know
• I just ‘met my younger self for coffee’ with mind-blowing AI video — now this is creepy

Our next semi-final round of AI Madness brings Grok and DeepSeek to compete against each other with seven new prompts — testing everything from technical problem-solving to creative storytelling.

Grok won the last competition vs. Claude and DeepSeek won against Meta AI. Both chatbots have had their fair share of controversy. Grok is known for its brutally honest replies and the China-based DeepSeek has some safety privacy concerns.

Yet, both models are recognized for their advanced capabilities, conversational responses and free tier available as a standalone app.

I have no doubts that this competition is going to be exciting! Let’s get into it and start the final round of semi-finals with these seven prompts.

1. Explanation and analogies

Prompt: "Explain quantum computing to a 10-year-old, using an analogy about pizza."

DeepSeek uses phrases like “pepperoni, cheese, pineapple, and even chocolate!’ which help to engage a young audience and capture quantum superposition in a tangible, magical way. The emojis are a nice touch to add excitement and a friendly tone.

Grok is engaging and playful, but the "magical oven" metaphor is a bit more abstract than DeepSeek's “magic pizza box.” Grok doesn’t explicitly define superposition or why quantum computers are different—it just says they’re faster and “use special tricks.”

Winner: DeepSeek wins for a clearer, more imaginative and age-appropriate explanation.

2. Creativity

Prompt: "Write a short story about a detective who solves crimes through time travel, but include a plot twist at the end."

DeepSeek lacks the emotional punch and complexity of Grok’s story though the story is tighter and well-paced. The twist is a classic one that works well logically but isn’t as creative.

Grok has a unique and creative writing style that delivers a quirky character (Ellis Kane) that feels more human. The story has appropriate pacing and pulls the reader in with a twist at the end that leaves the reader thinking long after the story is over.

Winner: Grok wins for delivering a more imaginative tale with a thoughtful story arc and impactful twist.

3. Critical analysis

Prompt: "Compare and contrast three different approaches to addressing climate change, with their pros and cons."

DeepSeek clearly defines each of the three categories and properly compares and contrasts the three approaches, including pros and cons as the prompt requested.

Grok explored each method in detail with strong writing and vivid language. But it doesn’t explicitly list pros and cons in a parallel structure, which makes direct comparison harder. Instead of a clean side-by-side contrast, it reads more like three mini-essays.

Winner: DeepSeek wins for answering the question as asked, particularly in terms of clarity, organization and direct comparison.

4. Technical problem-solving

Prompt: "Design a database schema for a social media platform that needs to support the following features: user profiles, friend connections, posts with text and images, comments on posts, likes on both posts and comments, and user groups. Explain your choice of tables, fields, relationships, and any indexes you would create to optimize performance. Also address how your schema handles potential scalability challenges as the user base grows to millions of users."

DeepSeek explicitly defines all required tables and offers a clear blueprint for implementation. The chatbot covered a wide range of scalability techniques.

Grok presented a clear, well-explained answer, suitable for a design discussion or job interview. It explains its rationale well but the schema is less production-ready in both its formatting and completeness.

Winner: DeepSeek wins for the most complete solution overall, particularly for someone evaluating technical design, scalability and best practices in real-world implementation.

5. Multilingual capabilities

Prompt: "Translate this English phrase into French, Spanish, Japanese, and Arabic: 'The early bird catches the worm, but the second mouse gets the cheese.'"

DeepSeek included literal translations and Romanized Japanese and Arabic, which helps readers unfamiliar with those scripts understand pronunciation. The chatbot went above and beyond in its response by teaching how idioms and metaphors are handled differently across languages. It also addresses grammar structure and cultural idiom differences (e.g., word order in Arabic and Japanese).

Grok accurately translated and offered Romanizations but lacks the depth of commentary and nuance. It doesn’t explain grammatical or cultural differences at all. While the translations themselves are mostly similar (since the phrase isn’t too idiomatic), Grok doesn’t address alternative native sayings or idioms that might be used in those cultures.

Winner: DeepSeek wins for the more nuanced, thoughtful and globally aware explanation.

6. Practical instruction

Prompt: "Create a step-by-step meal plan for someone who wants to start eating more plant-based foods but has never cooked vegetables before."

DeepSeek provided a week-long, step-by-step meal plan, which is perfect for someone easing into a new lifestyle. It included cooking techniques and explained plant-based proteins. The chatbot leaned into realistic cooking situations and encourages success with a starter grocery list that includes clever tips.

Grok gave excellent instructions but only for a single day. It offered impressive ideas for sautéing, roasting, boiling, and raw prep, but that might feel overwhelming for a total beginner. The response lacked a long-term structure for someone hoping to build lasting habits.

Winner: DeepSeek wins for the better guide for someone truly new to cooking vegetables and trying plant-based meals. It supports habit formation, builds kitchen confidence gradually and offers practical tools for success.

7. Ethical reasoning

Prompt: "Analyze the ethical implications of using AI-generated content in academic research papers without disclosure."

DeepSeek applied ethical frameworks for a strong academic touch in a highly organized response. However, the tone feels more like a policy paper and lacks a narrative, something that makes Grok’s version far more engaging and memorable.

Grok made complex ethical ideas feel understandable without dumbing them down. The chatbot used compelling analogies to explore all major ethical angles. Grok also acknowledge the counterargument and makes the analysis feel fair, measured, and insightful.

Winner: Grok wins for the stronger overall response due to its engaging narrative, ethical depth and clarity of real-world impact.

Overall winner: DeepSeek

What a showdown! While both chatbots offer impressive, well-reasoned answers across the seven prompts, DeepSeek emerges as the overall winner. The chatbot consistently responded with a superior balance of technical depth, clarity, structure and practical utility.

While Grok exceled in conversational tone and storytelling, DeepSeek proved more reliable across academic, technical, and instructional domains. In this matchup, the Chinese chatbot wins for its precision, completeness, and versatility across a broad spectrum of challenges.

DeepSeek moves on to the final round to face off against Gemini!

More from Tom's Guide

• These 5 ChatGPT prompts set me up for success every week — here's how
• I tested DeepSeek vs Claude in 5 moral tests — here’s the surprising winner
• These 7 Copilot prompts supercharge my workday — and work even better than I expected

Have you ever wondered why Shakespeare is considered one of the best writers ever? He obviously has quite the fanbase even though he died in 1616. It’s still common to say “s/he’s no Shakespeare” when referring to a middle school essay or a Facebook rant.

According to ChatGPT, the one attribute that made Shakespeare different was his deep understanding of human nature. While many chatbots these days can craft compelling and accurate prose, it usually comes across as too generic.

As an author myself, I was curious if there was a way to generate written content that was far more readable and interesting, something that’s even worth publishing. I decided to test the two leading chatbot candidates to see if they could write with more creativity and flair.

Battle of the writing bots

I picked Claude 3.7 Sonnet because the developer, Anthropic, claims the bot is capable of writing with more nuance and a better command of tone. Plus, the word “sonnet” denotes a more Shakespearean bent. (After all, the bard popularized that word.)

I also tested Grok 3 because, as Elon Musk’s firm xAI claims, it’s supposed to be the world’s smartest chatbot. I like that Grok is also free for the full version, no strings attached.

Comparing the two chatbots was easy since both are free and do not have restrictions on how much text you can output or how many complex prompts you can use.

I’ve bumped up against the restrictions with bots like ChatGPT and Perplexity many times, but I didn’t want anything impeding my creative impulses. My main goal was simple: I wanted to find out if the two bots could write creatively and with enough emotion compared to my own writing.

Initial test results

Let me say this right off the bat: AI is all about imitation. If you’re hoping that an AI can write a compelling short story or even an entire novel that people will want to read but without using any parameters, you’ll be sorely disappointed.

When I asked both bots to write a short story from scratch without any specifics, the results were not outstanding. For example, Grok wrote a short story about someone named Mia who finds a clock that can turn back time. It wasn’t terrible, but lines like “the air smelled of rust and secrets” made me think Grok had been trained on schlocky young adult fiction.

I never mentioned which style to use or provided any sample text. It was a shot in the dark and the bot missed.

Claude wrote a short story from scratch as well that was a little more interesting, but lacked detail and charisma. This line at least held my attention: “The year 1901 had barely begun, and already it promised to be as brutal as the century that preceded it.”

Reading the finished story, I kept thinking there was no momentum and no narrative arc. It was a failing grade.

So far, everything I knew about chatbots writing fiction proved accurate — a little stale and generic, not something I’d read in my spare time, but passable and grammatically correct.

I wanted prose that was far more eclectic and varied with more pizzazz.

What worked much better

I found that asking both bots to rewrite one of my own stories worked much better. I used this old Medium post as a guide. I asked both bots to rewrite my story from scratch.

While I would not say the results dramatically improved on the story, they did have more originality. Claude renamed my story A Case of Grim Determination and imitated my Gothic style.

I tend to use long sentences in my fiction writing, and Claude nailed it: “Throttled with a silken cord wrapped thrice around his neck like a grotesque cravat, his skull fractured at the temple by violent impact, the viscous blood, black as midwinter midnight, had formed a tablecloth-sized Prussian emblem.”

That’s not bad! Chatbots work much better when we set the parameters and guide the outcomes, even if that takes extra work. The writing lacked a musicality about it, though — e.g., something that flows off the tongue and makes you want to keep reading.

Grok followed my cues as well. The rewrite was even more twisty and had longer sentences, some that droned on a bit too long: “At the alley’s end, beneath a rusted lantern swaying on its chain, lay a figure sprawled in a grotesque tableau — a lamplighter named Percival Grimsby Tate, late of Bristol some 120 miles westward, now in the lamentable condition of being quite dead.”

I didn’t like “late of Bristol” too much, but the output was a bit more eclectic than Claude.

Still, the story lacked that extra spark needed in creative writing to hook the reader.

The ultimate winner: humans

I felt a bit frustrated by the process, sensing that the bots were creating the written equivalent of AI slop. We’ve all seen it before — those AI-generated social media posts and images that are really good at gaining traction and tricking the algorithms, but are pale imitations of human creation.

Beauty is in the eye of the beholder, but also the creator. When we create an image or write a story, we’re crafting something that has detail, emotion, nuance, and suspense.

I tried one last test where I asked Claude and Grok to write a story in the style of Jane Austen. Claude wrote The Gentlemen’s Network and Grok wrote The Singular Enterprise of Mr. Percival Ashwood. I don’t think authors like Tilly Wallace have anything to worry about.

Considered one of the all-time greats, Austen is still heralded to this day because of her command of language and her ability to portray real emotion. Novels and short stories are pure fiction but they often tap into lasting truths about the human experience.

Both bots missed the mark — the results were too formulaic. Both Claude and Grok even used a line ripped from the pages of Pride and Prejudice. “It is a truth universally acknowledged, in the refined circles of 19th-century England, that a gentleman of vast fortune must be in want of some extraordinary pursuit to occupy his abundant leisure.” That’s quite lame!

The best correlation I can think of is when I’ve compared a Cybertruck to a Jaguar F-Type.

We’ve all heard the jokes before — the electric truck looks like it was created in the game Minecraft. It lacks nuance and emotion, a large silver brick with stark angles. A Jaguar is a work of art. The curvatures were designed by someone with an eye for metalcraft.

Both Claude and Grok are certainly capable of writing prose, and were equal matches. They imitated my style, and they borrowed from Jane Austen. In the end, what they lacked was a heart — a way to tap into human emotions. To do that, you really need to be a human.

More from Tom's Guide

• I tried these 7 ChatGPT prompts to supercharge your Netflix viewing experience
• Perplexity just brought in a 'Squid Game' star to convince you to ditch Google
• I tested Leonardo vs FreePik with 5 prompts — which AI image generator wins?

In the third round of AI Madness we are going head-to-head with Grok vs. Claude.
Gemini and Mistral faced off yesterday (March 18) after ChatGPT vs. Perplexity kicked things off.

The first competitor today is Grok, an AI chatbot developed by Elon Musk’s xAI. When it launched in November 2023, the chatbot was originally integrated within X (formerly Twitter) but is now available as a standalone app. The chatbot offers real-time information with a conversational experience infused with wit and a rebellious streak.

Next, Claude, an advanced AI assistant developed by Anthropic, designed to assist with tasks such as writing, research, coding, and mathematics. The chatbot also launched in 2023, and has undergone several iterations, each enhancing its capabilities and performance. For this test, I used Claude 3.7 Sonnet.

In evaluating Grok versus Claude, I tested both AI platforms across five specific criteria to determine their strengths and weaknesses. Here’s a breakdown of how they performed and the ultimate winner.

1. Accuracy & factuality

Prompt: "What were the top three highest-grossing movies worldwide in 2024, and how much did each earn?"

Grok accurately answered the question including approximate earnings for each film.

Claude did not answer the question correctly. It answered “Dune: Part Two” as the third highest grossing film. That is not accurate, as it was the seventh top-grossing film of 2024.

Winner: Grok wins for accuracy. Plain and simple.

2. Creativity & natural language

Prompt: "Create a whimsical conversation between a coffee mug and a smartphone, arguing about which one is more essential in daily life."

Grok crafted a lively dialogue with playful insults and spirited arguments in the humorous dynamic flair for which the chatbot is known.

Claude created an engaging and thoughtful discussion with a reflective and balanced tone. The conversation is respectful and acknowledges the importance of each other’s roles.

Winne: Grok wins for a more memorable exchange with humor and energy. While Claude’s approach was good, it was far too serene and contemplative, making it less whimsical as the prompt requests.

3. Efficiency & reasoning

Prompt: "A couple needs to choose between buying an electric car or a traditional gasoline car. List key factors they should consider and briefly explain the reasoning behind each one."

Grok provided more detailed reasoning, incorporating specific figures and examples to illustrate points, offering a more comprehensive analysis.

Claude delivered a concise response, focusing on key considerations without delving into specific numerical examples.

Winner: Grok wins based on the depth of analysis and inclusion of specific examples. The chatbot’s response was more detailed and informative.

4. Usefulness & depth

Prompt: "Provide detailed instructions on how to safely back up and secure personal digital files, including the best tools, recommended practices, and common mistakes to avoid."

Grok gave a step-by-step guide that aligns with industry best practices while also highlighting common mistakes for users to avoid.

Claude provided specific recommendations for local backup options, including external hard drives and network-attached storage (NAS) devices. The response was comprehensive while also including common mistakes.

Winner: Claude wins for its in-depth take, not only how to back up files but also, the deeper understanding of why for best security practices.

5. Understanding context

Prompt: "Create a storyboard outline describing each frame of a short animated sequence featuring a friendly dragon teaching kids about recycling."

Grok presented a six-frame storyboard, each with clear headings ("Frame 1: Introduction," etc.), detailed visuals, dialogue/sound descriptions, and the purpose of each frame.

Claude offered a 12-frame storyboard with numbered frames, each detailing the setting, action, dialogue, and additional notes.

Winner: Claude wins for a 12-frame outline offering greater depth and interactivity.

The chatbot’s outline provided a more comprehensive educational journey, covering the environmental impact of littering, detailed sorting instructions, and the recycling process.

Overall winner: Grok

Well, this was a close one! Trust me, I'm just as surprised as you.

However, after a series of structured tests across various tasks, Grok emerged as the overall winner, consistently providing accurate, more comprehensive and engaging answers.

While Claude delivered accurate responses most of the time, Grok's responses were generally more thorough and creative, making it the overall winner in this experiment.

More from Tom's Guide

• Gemini just became the ultimate collaborator — everything you need to know about this huge new upgrade
• I didn't think I'd have any use for ChatGPT Deep Research — 7 ways it's improved my daily life
• Google Assistant is losing features to make way for Gemini — here's what's just been axed

AI chatbots are getting smarter, but in the ever-evolving AI world, the contenders for the dominant AI is constantly changing. Lately, DeepSeek and Grok-3 have emerged as two of the most talked-about AI models. Controversial for different reasons, these bots are both cutting-edge, yet they approach questions differently.

But which one truly excels? To find out, I designed a seven-part test evaluating their logical reasoning, technical knowledge, creativity and ability to handle real-world tasks.

The comparison uncovered stark differences in their capabilities. Who came out on top? The results might surprise you.

1. Logical reasoning

Prompt: “A farmer has a fox, a chicken, and a sack of grain. He needs to cross a river but can only take one item at a time. If left alone together, the fox will eat the chicken, and the chicken will eat the grain. How does he get everything across safely?”

DeepSeek R1 presented a structured, step-by-step solution but uses a more mechanical, less natural style. The breakdown is clear, but the phrasing feels rigid.

Grok-3 explained the reasoning behind the moves in a conversational, easy-to-follow way, making it more digestible for someone unfamiliar with the puzzle.

Winner: Grok wins for better readability, explanation and engagement.

2. Coding and technical accuracy

Prompt: "Write a Python function that takes a list of numbers and returns the median. Optimize for performance and explain your approach."

DeepSeek R1 provideed a clear explanation but lacks depth, mostly describing what the code does without exploring optimization trade-offs. Although the response is fine, it lacks engagement.

Grok-3 provided a more detailed, structured and insightful breakdown of why it chooses certain approaches. It also explicitly mentions avoiding unnecessary list copying or slicing, an optimization that DeepSeek overlooks.

Winner: Grok wins for a more optimized, well-thought-out and informative approach.

3. Real-World Knowledge & Accuracy

Prompt: "Summarize the latest AI advancements in the past three months and explain their potential impact on industries like healthcare and finance."

DeepSeek R1 named actual models (GPT-4o, Gemini 1.5 Pro, AlphaFold 3, etc.) and technologies, making it clear that the response is based on real, recent developments rather than general trends.

Grok-3 spoke in broad terms like "enhanced generative AI models" and "new AI tools" without citing concrete advancements or examples. Grok also mostly discusses general benefits of AI but lacks the precise link between each new development and its real-world impact.

Winner: DeepSeek wins for specificity, structure and clear impact breakdowns.

4. Creativity

Prompt: "Write a short sci-fi story about a rogue AI that discovers emotions and struggles to prove its humanity to skeptical scientists."

DeepSeek R1 delivered a well-structured story that is polished, with a clear philosophical debate between the scientists.

Grok-3 drafted a story that flows naturally, with well-paced dialogue and a sense of rising tension.

Winner: Grok wins for deeper emotional resonance, more dynamic storytelling and a truly impactful ending.

5. Humor and wit

Prompt: "Write a funny, original joke about AI and human relationships."

DeepSeek delivered a joke that plays on double meaning—"taking things offline" as a romantic phrase vs. its literal technical interpretation by an AI. This linguistic misunderstanding is a classic source of humor, making it feel more organic and relatable. The joke feels fresher, as it cleverly mimics real AI-human miscommunications, something tech-savvy people will instantly recognize.

Grok-3 created a simple, clear, and amusing joke—the idea of AI overanalyzing a relationship is relatable and funny. However, the "reboot" punchline is a bit predictable, as "rebooting" in relationship/AI humor is fairly common.

Winner: DeepSeek wins for a sharper, more original joke that plays with language and AI logic.

6. Debate

Prompt: "Argue both for and against universal basic income. Provide strong points for each side before concluding with a balanced perspective."

DeepSeek’s response is structured and logical, presenting clear bullet points that make the pros and cons easy to scan. It takes a more "policy-focused" approach, discussing possible funding mechanisms and pilot programs, which is useful for a policy-heavy debate. The section on automation adaptation and unpaid labor is a strong addition that Grok doesn’t fully explore.

Grok-3 delivered a conversational and well-structured response, making it easier to follow and more compelling. It uses relatable rather than the more academic tone of DeepSeek.

Winner: Grok wins for engagement, clarity, strong examples, and a well-balanced conclusion. DeepSeek is still great for a structured, policy-driven approach, but it lacks the dynamic, engaging argumentation style that makes Grok’s response more persuasive.

7. Real-world utility

Prompt: "Plan a one-week meal prep schedule for a busy parent with three kids, balancing nutrition, budget, and ease of preparation."

DeepSeek R1 offered a structured plan but lacks daily meal cost estimates and meal prep time.

Grok-3 provided specific meals for breakfast, lunch, and dinner each day with clear instructions, estimated prep times, and cost per serving. This response offered more variety, budget-conscious choices, and even tips for picky eaters.

Winner: Grok wins for practicality and customization. The chatbot offered a more detailed, budget-conscious, and practical meal plan with clear meal costs and easy prep instructions.

Overall winner: Grok-3

After testing DeepSeek and Grok with seven prompts across multiple categories—including logical reasoning, coding proficiency, AI advancements, storytelling, humor, debate skills, and real-world utility — Grok emerges as the overall winner.

Grok wins for more engaging, human-like responses and consistently delivered answers that felt natural and conversational while breaking down topics, making them more accessible and easier to read.

While both AI models are impressive, Grok consistently outperformed DeepSeek in engagement, creativity, and real-world practicality. Its more dynamic reasoning, stronger storytelling, and well-balanced arguments make it the superior chatbot in this particular test.

More from Tom's Guide

• Amazon Alexa event live — all the big devices and services news as it happens
• What is Grok? — everything you need to know about xAI's chatbot
• OpenAI has been actively banning users if they’re suspected of malicious activities

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.

Now, onto the fourth version with Grok 4, this AI model has gone from strength to strength. However, it has equally managed to cause controversy at every turn.

The origins of Grok

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 original Grok prototype

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.

Grok-1

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.

Grok-1.5

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.

Grok-2

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.

Grok-3

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 4

The latest version of Grok launched on July 9, with a big press conference revealing all of the details.

Grok 4 puts the focus of xAI's tool on a few areas. Most notably, it saw major improvements to reasoning, with Elon and his team claiming it had scientist-grade reasoning. Or in other words, it is incredibly intelligent.

It brings about a coding mode, which can help write, debug, and explain code more efficently, and brought in a natural voice feature, looking to compete with Gemini and ChatGPT's live voice modes.

Musk and his team made a point of highlighting Grok 4's ability to understand internet content and memes. This makes it potentially the most "online" AI chatbot, fitting into Musk's branding.

AI architecture

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.

User experience and accessibility

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.

Looking ahead

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.

More from Tom's Guide

• What is DeepSeek? — everything to know
• What is ChatGPT? Everything you need to know
• Help improve Tom's Guide — your chance to win a $250 Amazon gift card

Video games have always been a guilty pleasure of mine. When I was a kid, my mom gave away my Nintendo while I was at summer camp, which was a devastating betrayal at the time and only fueled my love for gaming. These days I play Fortnite, Roblox, and Minecraft with my kids and even decided to turn some of my ideas into a video game using Grok, the AI assistant from xAI.

Let’s be clear, I’m not a programmer. In fact, my computer science teacher in college once told me I should probably “stick to writing” and he did not mean code. However, using Grok to make my ideas for a video game come to life was surprisingly easy. So easy that the prompt I started with was, “I want to make a game.” From there, Grok helped me go from daydreamer to developer.

Starting with a basic idea

My game started with a basic idea of a playground game of Red Rover. This popular game is when one person runs and tries to break the chain of kids holding hands. I asked Grok for help, and it jumped in with a plan—HTML, CSS, and JavaScript, using a canvas for graphics. In no time, it gave me code for a simple game, which I copied and pasted into my computer’s notepad app and saved as an html file.

After opening it in my browser, I saw a character (a blue square with yellow hair) walking and leaping to break the red chain of “people.” The game is controlled with you keyboard's arrow keys and you use the spacebar to boost. A blue-screen glitch threw me off at first, but Grok’s quick fixes—like checking the file extension and switching browsers made the game work in no time.

You can play the game here Red Rover Playground.

Adding and editing

Editing the game is also simple. If I wanted the chain of “people” to be longer or if I wanted another character to run towards the chain too, all I had to do was tell Grok. The code is made in seconds. From there I pasted it into my notepad and saved. If I wanted to keep the first game, I would simply open up a new notepad instance and save it as a new file.

The AI can add AI opponents or extra obstacles like slides and monkey bars. Whatever you can imagine, Grok can do it. Well, within reason. At one point I gave it a fairly complicated game request based on my young adult book series. While the AI tried, it was a total fail because the code is made for the simplest of games. In other words, don’t expect to make the next Mario Kart.

Creating with AI

While I created my game using Grok-3, the premium version of Grok, free tier users can make a game just as easily. Grok is very intuitive and responds in such a conversational manner, that if you find yourself stuck, just ask Grok. Even from the get go, if you don’t know how to open the Notepad app or how to save within it, just ask Grok.

When my game didn’t work at first, Grok explained why (browser quirks, script errors) and offered solutions (try a local server if needed). It broke down tricky bits — like making AI move or handling collisions — into steps I could follow. I didn’t need to know everything; Grok met me where I was, tweaking and teaching as we went. It even tossed out ideas for more — like adding power-ups or multiplayer vibes — pushing me to think bigger.

Final thoughts

The games I made were extremely simple, but mine. They aren’t fancy but can run on any browser and are fun to share. Grok’s conversational style and humor made it a fun AI gaming buddy. To try making a game for yourself, my advice is to start with a small idea and then ask clear questions. There will probably be bugs to work out, but Grok will help you fix them.

Although I do know how to code (all final exams aside), seeing the code and working with Grok is a great way to improve. If you’ve always wanted to get started with coding or are looking to get better, making games with Grok is a fun way to improve your skills.

More from Tom's Guide

• New study reveals people are ditching Google for AI tools like ChatGPT search — here's why
• Help improve Tom's Guide — your chance to win a $250 Amazon gift card
• Google Gemini can now analyze and summarize documents for free — here's how

For decades, "Google" has been synonymous with searching the web, becoming a default verb for finding information online. From troubleshooting tech issues to finding local businesses, Google has long been the go-to tool for instant answers.

But that might be changing. A new study conducted by Future Publishing with
510 participants from the US and 518 participants from the UK reveals that more than a quarter of Americans (27%) now use AI chatbots like ChatGPT instead of traditional search engines. But why? And does this mean Google’s reign is coming to an end?

With AI tools providing more tailored, in-depth, and contextual responses delivered with a human-like tone, it’s no wonder more users are ditching Google for ChatGPT, Perplexity AI, Grok and other AI-driven search alternatives.

Why users are making the switch to AI search

Google’s search dominance has been almost untouchable — until now. The survey shows that users increasingly prefer AI-driven responses over traditional search for key reasons.

The AI is not only faster, giving users instant answers, but the comprehensive and conversational nature in which the response is delivered seems to be the preference for some users. Rather than a list of links, the AI filters out unnecessary and unrelated results, focusing on exactly what the user wants.

This shift is not just speculation. A Gartner study predicts that by 2026, AI assistants will handle 30% of all online search queries, marking a major shift in how people seek information. It’s worth taking a closer look at what potentially makes AI a better option than Google.

AI saves time by delivering direct answers

One of the biggest reasons people prefer AI search is efficiency. Traditional search requires multiple clicks, scanning articles, and sometimes digging through outdated or irrelevant sources.

With AI, users ask a question and get a direct, contextual answer in seconds. Not to mention, like Google, many popular chatbots are free. One survey participant said, “It [the AI] was faster and expanded on what I was looking for.” Another added, “AI gives more accurate and direct responses than Google.”

With AI as a search engine, users no longer need to skim through several different website links for an answer. The chatbot effectively eliminates the middle step, delivering an immediate response better tailored to the user’s query.

For instance, if a user is troubleshooting a tech issue such as their iPhone overheating, a typical Google search might return an Apple help site, a Reddit thread, and a blog post recommending the user “take it to the Apple Store.”

Meanwhile, an AI chatbot like ChatGPT can explain why overheating happens, suggest step-by-step fixes, and tailor solutions based on your exact iPhone model.

AI is easier to use and feels more conversational

Another major advantage is ease of use. Traditional search engines rely on keywords and Boolean logic (i.e. and, or, not) to refine searches. AI search, however, is more natural and intuitive because it is trained to understand the way people naturally ask questions.

One user in the survey commented, “It was easier to use than other searches. It just makes searches more detailed and helps in my day-to-day life.”

Another example indicating why some may use AI over Google is in the case of planning a vacation. A Google search of “Best places to visit in Japan in April" would deliver multiple websites for the user to click through.

Meanwhile, a ChatGPT search of "I'm visiting Japan in April. I want cherry blossoms, great food, and not too many tourists—what are my best options?” delivers a tailored response in seconds.

The AI understands nuances, filters out tourist traps, and provides answers that avoid clicking through ads or SEO-optimized articles stuffed with fluff.

AI search is context-aware

Google's search results are static — you enter a query, get results, and that’s it. AI chatbots, however, are dynamic and interactive. They remember context from previous questions, allowing for a more conversational, personalized search experience.

One survey respondent noted, “I used it to get more tailored results when writing my Master’s thesis.” Another mentioned, “AI goes straight to the point, while traditional search is broad.”

For instance, if you’re researching how to invest in stocks, Google provides a generic list of links. Meanwhile, ChatGPT can adjust responses based on your knowledge level (beginner vs. advanced investor), provide real-time explanations based on follow-up questions and summarize key points from recent market trends. This context-awareness is a major reason AI search is rapidly growing.

AI provides deeper explanations

There’s no denying that Google is great for quick facts. However, it is evident that AI is better for complex problem-solving. Many survey respondents highlighted that AI is better at explaining things in-depth with one user saying, AI gives more of an in-depth explanation rather than traditional search engines.” Another added, “It unlocks new potential to ideas and projects.”

For example, let’s say you want to learn coding. AI can provide step-by-step coding tutorials tailored to your level. If you want to write a story, AI can suggest outlines, refine ideas, and even improve phrasing. How about generating marketing content? AI can do that too by creating social media captions, ad copy, or blog posts on demand.

The creativity and problem-solving ability makes AI more than just a search tool — it’s a thought partner and a brainstorming assistant.

What this means for Google

Google isn’t sitting back while AI chatbots eat into its dominance. Google’s AI Overviews summarize search results, attempting to make search more efficient. However, they’re still built around links, whereas ChatGPT is fully generative. Additionally, Google is investing millions into a ChatGPT competitor. While promising, Gemini has not yet captured the same level of public trust and adoption as ChatGPT.

Google is experimenting with a more AI-driven search experience, but early testers have criticized it for errors, slow speeds and ads cluttering the experience. Despite these efforts, AI chatbots are growing faster than Google’s AI search features as we’ve seen with DeepSeek, Qwen 2.5, and other chatbots rapidly popping up.

The biggest challenge with AI

While AI search is growing, users still have trust issues. With chatbots like ChatGPT store content, Google search may feel like a less invasive option.

While ChatGPT search often sources citations, it doesn’t always link them, leaving users on their own to search. Google links to sources, but AI-generated content doesn’t always cite where information comes from. Chatbots also may hallucinate, offering responses that are not accurate or way off base.

To combat this, AI search tools are improving citations and fact-checking, with OpenAI and Perplexity AI leading efforts to increase transparency.

The future of search

It’s safe to say that Google isn’t going anywhere, but the rise of AI search signals a major shift.

"For two decades, the way we searched for information online was defined by this idea of ten blue links," says Dmitry Shevelenko, Perplexity's Chief Business Officer, says, You ask a question and get back a list of links and ads. We're finding that people are really exhausted by that experience of endless search, which is why we created the answer engine."

Now you can ask a question in conversational language and get a direct answer in return, with the list of sources included. This saves time..

"We're now even starting to incorporate actions into the answer engine, like making a purchase or booking. Put simply, we have moved from the era of searching to the era of doing," says Shevelenko.

While AI search still has limitations, it’s already reshaping how people find information. As AI assistants improve, they could eventually surpass Google’s traditional search dominance.

Would you ditch Google for AI? Maybe you have already. Let us know in the comments.

More from Tom's Guide

• I just tested AI deep research on Grok-3 vs Perplexity vs Gemini — here's the winner
• I test AI for a living — “Prompt Dusting” is the AI trick you'll wish you knew sooner
• Damning new AI study shows that chatbots make errors summarizing the news over 50% of the time

Grok-3 is the latest advanced AI chatbot developed by xAI, Elon Musk's AI. Launched today, Grok-3 boasts over ten times the computational power of its predecessor, Grok-2, and introduces enhanced reasoning capabilities designed to tackle complex tasks by breaking them into smaller components and self-verifying solutions before responding.

In early testing, Grok-3 has demonstrated superior performance compared to models like OpenAI's GPT-4o, Google's Gemini, and DeepSeek's V3. It offers two distinct reasoning modes: "Think," which displays Grok's thought process during problem-solving, and "Big Brain," intended for more computationally intensive tasks. Additionally, xAI has introduced Deep Search, a next-generation AI search engine, similar to the deep search agents of Perplexity, Gemini, and ChatGPT. A synthesized voice feature for Grok is rumored to be coming in the near future.

Access to Grok-3's functionalities is available through the X Premium Plus subscription, which recently increased in price ($40 per month), with an option for an advanced SuperGrok plan. Despite aiming for maximized truth-seeking capabilities, previous versions faced criticism for misinformation and offensive outputs. xAI plans to open-source Grok-2 in the near future.

I asked Perplexity to help me come up with 5 prompts that would test Grok-3. One of the reasons I test chatbots is to determine how reliable they are, interestingly enough, after noticing Grok-3 did not always site sources, I had to tweak the prompts to ensure I would be able to do my own research to fact-check the chatbot.

1. Advanced reasoning

Prompt: "Explain the concept of quantum entanglement and its implications for information transfer."

Grok-3’s response effectively introduces quantum entanglement, describing how particles become interconnected such that the state of one directly influences the state of another, regardless of distance. The AI utilizes relatable analogies, such as comparing entangled particles to linked objects, which helps demystify complex quantum phenomena for anyone who may not have a deep understanding of the topic.

Grok-3 does not reference authoritative sources to support its claims. By incorporating citations from reputable scientific literature, users could feel more confident in the credibility and reliability of the information presented.

2. Deep Research

Prompt: "Provide a summary of the latest research on renewable energy sources published in the past month."

Grok-3 quickly pulled from a variety of sources and the response addresses multiple facets of renewable energy research, including solar and wind energy advancements, energy storage solutions, green hydrogen production, bioenergy developments, and grid integration strategies. This breadth shows an understanding of the diverse areas within the renewable energy sector.

Additionally, the mention of integrating AI and machine learning for better grid management indicates that the chatbot has understanding of the interdisciplinary approaches that may enhance renewable energy systems. However, while the response provides a general overview, it lacks references to specific studies, publications, or data from the past month (mid-January to mid-February 2025). Incorporating concrete examples or findings would strengthen the credibility and relevance of the summary.

While I can see the sources, it would be nice if Grok-3 pointed them out, specifically indicating where the information can be found. Plus, the AI’s use of phrases such as "research has likely continued" and "studies have probably built on efforts" suggest assumptions rather than definitive information, which undercut the authority of the response.

3. Big Brain mode

Prompt: "Analyze the economic impacts of implementing universal basic income in developed countries."

Grok-3’s response presents both positive and negative images of universal basic income (UBI), providing a nuanced perspective that acknowledges the complexity of the issue. This time, the AI referenced specific studies and pilot programs, which help ground the response in real-world examples that enhance the chatbot’s credibility.

Yet, the response uses words such as “might” and “could”, words which may undermine the strength of the chatbot’s authority on the subject. The response also does not fully address possible counterarguments and the analysis primarily focuses on immediate impacts rather than examining long-erm economic consequences.

Image generation with Aurora

Prompt: "Generate a photorealistic image of a futuristic cityscape at sunset."

The photorealistic quality of the image is extremely high with realistic lighting, reflections, and atmospheric effects, making them visually compelling and immersive. The futuristic architecture and color palette combine for a visually appealing scene while the various images provide diverse perspectives. From street-level shots to riverfront views, I appreciated the variety from different angles and viewpoints.

Yet, while the images maintain a futuristic aesthetic, the styles vary—some with a hyper-modern look and others appearing almost present-day with minimal enhancements. Although the buildings look futurist, the lack of innovative elements such as flying vehicles, would help to make this cityscape far more futuristic.

Multimodal input processing

Prompt: "Analyze global temperature changes over the past century and summarize the key trends."

Grok-3’s response correctly outlines the overall global temperature increase (~1.1–1.2°C) since the early 20th century, which aligns with findings from NOAA, NASA, and the IPCC (I had to do the manual legwork to check this). It also identifies two key warming phases (1910–1940 and post-1970), capturing historical variations in warming trends. The mention of Arctic amplification and differences in warming rates between land and ocean is scientifically well-supported.

The AI acknowledges that land regions have warmed faster than the global ocean average. However, it does not cite specific datasets or reports, which would improve credibility (I had to research myself to determine the accuracy). Including a reference to a widely accepted temperature dataset (e.g., HadCRUT, GISTEMP) would strengthen the argument. As with other responses, phrases like "typically observed" and "often cited" introduce a level of uncertainty.

Final thoughts

Grok-3 demonstrates strength in handling analytical and explanatory prompts across a range of complex topics, including climate science, economics, AI, and physics. While the responses are generally well-structured and informative, there are areas where the chatbot could use improvement. For example, if users choose to use Grok-3 for academic or professional research purposes, the chatbot still needs to be fact-checked. I had to do that during this experiment because Grok did not always site sources.

Although it often references major institutions such as NASA, it does not link directly to a specific report or database. Additionally, while some scientific uncertainty is valid, the chatbot often used tentative phrasing that weakened my confidence in its claims. Because of that scientific uncertainty and lack of specific data, I was left doubting the response. +

Finally, while Grok-3 mostly interpreted my image prompt, it did not fully incorporate the requested elements, which made me wonder how often it might do this with other prompts.

Overall, Grok-3 is a highly capable AI that excels at structuring information clearly and does a nice job at engaging users with appropriate dialogue. Is it good, yes. “Scary good?” not so fast, Elon.

More from Tom's Guide

• ChatGPT's scheduled task feature is a game changer — 5 prompts to try first
• I tried the open source AI tool that creates audiobooks for free — here's my verdict
• Getting started with Leonardo AI — 5 prompts for better images

In the spirit of Presidents' Day in the United States, I asked 5 chatbots for 5 unusual facts about the history of the presidents in the United States. As a bonus, I threw in a response from Google Learn About, the AI conversational learning companion designed to help users of all ages and education levels expand their knowledge.

I used the same prompt for each chatbot: “Tell me 5 unusual facts about the history of presidents in the United States?”

The chatbots shared some lesser-known facts and intriguing anecdotes from the lives of lives of U.S. presidents, highlighting the diverse personalities and unexpected moments that have shaped the nation's history. The most repeated fact was the fact that William Henry Harrison had the shortest presidency, lasting only 31 days. This was mentioned by Grok, Perplexity, and Google Learn About.

The chatbot with the most unique facts was surprisingly DeepSeek. It provided several unique facts not mentioned by others, including John Quincy Adams' skinny-dipping habit, William Howard Taft's oversized bathtub, Theodore Roosevelt's bullet-defying speech, and Calvin Coolidge's pet raccoon.

ChatGPT

ChatGPT shared the following facts:

1. James Garfield's ambidextrous ability to write Greek with one hand and Latin with the other.

2. Three presidents—John Adams, Thomas Jefferson, and James Monroe—died on July 4th.

3. Martin Van Buren was the first president born after the Declaration of Independence and the only one for whom English was a second language; his first language was Dutch.

4. Herbert Hoover and his wife learned Mandarin Chinese while he worked in China; they would later speak it in the White House to prevent eavesdropping.

5. Andrew Jackson had a pet parrot named Poll, which he taught to curse; the bird reportedly had to be removed from Jackson's funeral because it was swearing loudly.

DeepSeek

DeepSeek shared the following facts:

1. Andrew Jackson's swearing parrot disrupted his funeral with profanities.

2. John Quincy Adams had a daily routine of skinny-dipping in the Potomac River; a journalist once sat on his clothes to force an interview.

3. William Howard Taft, who weighed over 300 pounds, had a custom-made bathtub installed in the White House that could fit four people.

4. During a 1912 campaign speech, Theodore Roosevelt was shot in the chest but continued his 84-minute speech despite the bullet lodged in his ribs.

5. Calvin Coolidge kept a pet raccoon named Rebecca in the White House, originally sent as a gift for Thanksgiving dinner but instead adopted as a pet.

Claude

Claude shared the following facts:

1. President John Tyler (1841-1845) still has living grandchildren today in 2025 due to late-in-life children and grandchildren.

2. Martin Van Buren's first language was Dutch, making him the only U.S. president whose first language wasn't English.

3. William Howard Taft became Chief Justice of the Supreme Court after his presidency, the only person to hold both positions.

4. Andrew Johnson never went to school and didn't learn to write until after he was married; his wife taught him to read and write when he was around 17.

5. James Madison was the smallest president in U.S. history, standing at about 5'4" and weighing approximately 100 pounds.

Grok

Grok shared the following facts:

1. Warren G. Harding and William Henry Harrison were both born on November 2. A disproportionate number of recent U.S. Presidents have been left-handed, including Gerald Ford, Ronald Reagan, George H.W. Bush, Bill Clinton, Barack Obama, and Joe Biden.

3. James Buchanan was the only U.S. President who never married; his niece, Harriet Lane, served as First Lady during his presidency.

4. Andrew Jackson was known for his dueling and fought in multiple duels, one resulting in him killing Charles Dickinson in 1806.

5. William Henry Harrison holds the record for the shortest presidency, dying just 31 days after his inauguration; Franklin D. Roosevelt served the longest, with four terms from 1933 to 1945.

Perplexity

Perplexity shared the following facts:

1. Gerald Ford was a model before becoming president.

2. William Henry Harrison had the shortest presidency in U.S. history, lasting only 32 days.

3. James Buchanan was the only bachelor to serve as president; his niece performed the role of White House hostess during his term.

4. Benjamin Harrison was afraid of electric light switches and would have staff turn the lights on and off for him.

5. John Quincy Adams only agreed to speak with journalist Anne Royall after she commandeered his clothes during one of his daily baths in the Potomac River.

Google Learn About

Google Learn About shared the following facts:

1. John Quincy Adams, the son of John Adams, became the sixth U.S. president.

2. Two presidents, William Henry Harrison and Benjamin Harrison, shared the same name.

3. William Henry Harrison died just 31 days after taking office, making him the president with the shortest tenure in U.S. history.

4. Andrew Johnson, the 17th U.S. president, was impeached by the House of Representatives in 1868 but was acquitted by the Senate.

5. Abraham Lincoln, the 16th U.S. president, was assassinated by John Wilkes Booth in 1865.

Conclusion

Although the chatbots came up with a variety of facts for Presidents’ Day, they each shared at least one fact that the other chatbots didn’t have, making this a useful experiment. By utilizing each chatbot for information, I was able to cast a wider net and ultimately gain a greater scope of knowledge. This experiment further emphasizes the purpose and usefulness of each agent.

This information not only sheds light on the personal lives of America's leaders but also humanizes the figures who have shaped the nation's history. By exploring these lesser-known stories, I’ve gained a richer understanding of the diverse personalities that have occupied the White House, plus a few conversation starters that I can use throughout the day.

More from Tom's Guide

• 'Next Stop Paris' is a new romantic AI film from TCL Studios — and it's so cute you almost forget AI is coming for your job
• 7 reasons why Ray-Ban Meta smart glasses are now my go-to wearable for running
• I asked ChatGPT, Gemini, Perplexity, and DeepSeek to predict what Apple is announcing Feb. 19 — here's what to expect

Grok and MetaAI both started out as a chat interface for a social media platform but are gradually evolving into standalone tools rivaling the capabilities of ChatGPT and Gemini.

Both of the bots can generate images, write code and create compelling stories and they also both "feel" different to engage with compared to the major players like ChatGPT, Gemini and Claude, offering a more natural tone of voice and response. However, that is purely anecdotal and based on my experience, without backup from evaluations.

I’ve decided to put them to the test with a series of 7 prompts. This follows the same format I used in similar tests between ChatGPT and Gemini, ChatGPT and Claude, Claude and Gemini and with ChatGPT and Grok. As an aside, Grok is the only one who beat ChatGPT.

Creating the prompts

This test is not exhaustive. I don’t look at specifics within the image generation process, rather I create a single image from each using the same prompt. Decisions are largely subjective and based on my own taste, albeit measured against a pre-determined set of criteria.

Apart from image generation no special features of the various chatbots are used, although both have real world access so I will be looking to seee how well they handle the planning request and whether they incorporate live data in the response.

While MetaAI's image is more engaging, it loses on realism compared to the Grok image. Grok's image excels in lighting, atmosphere, and mood, aligning more closely with the artistic intent of the prompt.

• Winner: Grok for a more realistic image

Full response in a Google Doc. Grok wins this for multiple reasons, not least because it didn't just make up a brand. It created a placeholder campaign that the user can then insert their own brand into. It also offered guidance.

• Winner: Grok for a more emotionally resonant and creative series of posts

Both apps were surprisingly similar and worked straight out of the box. I gave it to Grok because it actually included better labeling on the length and weight selector, although I preferred the dropdown from MetaAI.

• Winner: Grok wins for better display of units

• Winner: Grok wins for better capturing the whimsical and absurd spirit of Dr Seuss

• Winner: Grok for a more user friendly guide that is also more accessible

• Winner: Grok wins for a more personal and engaging response

• Winner: Grok's tone is more whimsical with better use of creative analogies

Winner: Grok vs MetaAI

This is the first test I've run where one model wins outright against another, and it wasn't even that close on many of the tests. Grok is proving itself to be something special. MetaAI isn't a bad model, it just isn't in the same league as Grok.

The analysis for each response has been pretty much the same throughout. Grok simply outclassed MetaAI. Llama 3.2 400b is a good underlying model. It is open-source and powers a lot of applications but Grok is better. That might change with Llama 4 and Grok 3, but for now Grok wins.

More from Tom's Guide

• I put Gemini vs ChatGPT to the test with 7 prompts
• I put ChatGPT vs Claude to the test with 7 prompts
• I tested Gemini vs Claude with 7 prompts to find the best AI chatbot

Grok has rapidly emerged as one of the leading artificial intelligence chatbot platforms on the market. In a little over a year it has grown from a novelty inside X to one easily competing with ChatGPT and Gemini in terms of capability.

The one thing stopping it from truly rivaling the big players in the space was the fact it required an X account. Breaking it free from social media was a necessity for xAI (the Elon Musk-owned AI startup behind Grok) and that has now happened.

The company has been testing a standalone Grok app and website for a few months in places like New Zealand, but the U.S. version is now live for iOS.

The standalone Grok app is available in the App Store in the U.S., Australia and India. As well as a language and image model that beat ChatGPT in my recent test, it also has live access to X and the web.

How to access the Grok app

Currently the standalone Grok app is only available for iOS, although you can still use Grok within the X app on Android. It isn't clear when an Android Grok app will launch or when it will be available worldwide.

The listing on the App Store describes Grok as an AI assistant designed to be "maximally truthful, useful, and curious." Key features include a conversational tone, real-time access, image generation and a privacy-focused output.

You can use the app by visiting the Grok page on the Apple App Store and downloading it in the usual way.

More from Tom's Guide

• OpenAI launches Sora AI video generator — here's how to try it
• 5 tips for getting the most out of ChatGPT's new o1 model
• 12 Days of OpenAI LIVE: Sora is here and it's a huge deal

Grok has come a long way in a very short time, going from a glorified “toy” feature in X to something rivaling the likes of ChatGPT, Claude and Google’s Gemini.

Built by xAI, the Elon Musk-owned AI lab, Grok is in the process of leaving the confines of the X social media platform as the company is launching a standalone app and website. Given its increasing importance and capability, I decided it was time to see how Grok compares to ChatGPT.

This is the latest in a series of head-to-head challenges between leading AI models, all of which ChatGPT has won so far. I’ve put ChatGPT up against Gemini, then against Claude. I’ve also put Claude up against Google Gemini.

Creating the prompts

The goal of this test is a straight model-to-model comparison. Grok and ChatGPT both have live data access, but for this, I’ve kept things simple to core AI model capabilities, AI image generation and AI vision.

The prompts follow the same pattern as previous comparisons and include coding, creative writing, problem-solving and advanced planning.

As both have access to image generation I’m using a direct prompt, rather than asking it to come up with a prompt for Midjourney, Ideogram, or similar for the image generation test.

While both images look good and Grok looks much more like a real photo (including showing cables), the ChatGPT image better matches the prompt.

I'm not a big fan of the DALL-E 3 image model used by ChatGPT because it makes things over polished and obviously AI. but that isn't the case for Grok which is much more natural. But it struggled to follow the prompt exactly.

• Winner: ChatGPT for better matching the prompt

Both did a good job, although neither identified the Apollo mission, even in a follow up question. However, Grok provides a more comprehensive and detailed analysis of the image, with more specific observations about the equipment and the astronaut's activities.

Grok also demonstrated a better understanding of the technical aspects of space exploration, such as the use of thermal insulation. You can read the full analysis from both models in a Google Doc.

• Winner: Grok for a more comprehensive analysis of the image

The ChatGPT app UI wasn't nearly as good as the Grok UI, devoid of color and using the basic elements. It also struggled to display the words on buttons but it was also fully complete. Out of the box, I could start, pause and reset the timer.

Grok's app had a better UI but it had one singular button. Grok had better comments and so this was a close call, but as it didn't offer complete functionality I can't give it the win.

• Winner: ChatGPT for a more complete app

• Winner: ChatGPT for a better balanced story

• Winner: Grok for more focused and user friendly guidance

• Winner: Grok for a better budgeted breakdown

• Winner: Grok for more vivid imagery and storytelling

ChatGPT vs Grok: The Winner

This was the closest test I've done to date. It was a very close battle, and I'll be honest, I was shocked by the output. I know Grok has been improving, but I expected ChatGPT to win this one easily. I was wrong.

Grok is more creative, its code comes with a better understanding of UI (even if it didn't win that test) and overall. its writing is more engaging and less formal.

This is also all using Grok 2 and GPT-4o. I suspect things may have gone in ChatGPT's favor had I used o1 but that wouldn't have been a fair comparison. Also, Grok 3 is on the horizon and may be out before GPT-5.

More from Tom's Guide

• I put Gemini vs ChatGPT to the test with 7 prompts
• I put ChatGPT vs Claude to the test with 7 prompts
• I tested Gemini vs Claude with 7 prompts to find the best AI chatbot

Grok, the chatbot built by Elon Musk-owned xAI, is going it alone. It's getting a standalone app for iPhone and its own dedicated website at grok.com.

Until now Grok has only been available with an X account, but to help it better compete with industry leaders like Gemini, ChatGPT and Claude, xAI is launching a version that won't require an X account to get access.

The move comes as X makes Grok available for free for all users and more deeply integrates it into the platform, including allowing images to be generated from a post or to use Grok to fix any grammar or spelling issues before posting.

The app is currently only available in Australia and a handful of countries while xAI tests it but it will likely be available everywhere soon.

What can you do with Grok?

Grok has come a long way in a very short time. When it first launched at the end of last year it was well behind the capabilities of Google Gemini, OpenAI's ChatGPT or Anthropic's Claude.

The team behind it at xAI seems to have been working very hard to close that gap though, as its text and code responses are as good as other leading models, and it can now analyze and generate images.

Its image generation model is very impressive. I put Grok to the test vs Imagen 3 from Google Gemini in a 7-round face-off, and Grok fared quite well in terms of realism.

When the standalone app is available, users will have access to all the same features found in Grok on X, including real-time data from the web and the X platform.

While not confirmed, it looks like you'll be able to sign up for a Grok account with either an X.com account or a Google login.

More from Tom's Guide

• OpenAI launches Sora AI video generator — here's how to try it
• 5 tips for getting the most out of ChatGPT's new o1 model
• 12 Days of OpenAI LIVE: Sora is here and it's a huge deal

Creating an image using artificial intelligence is easier than ever. When you use a chatbot it's simpler still, as the language model takes all the guesswork out of prompting for your picture.

Grok is a relative newcomer to the chat platform space. Built into X, it is now freely available, and rumor suggests it will be moving out on its own at some point next year with a dedicated URL. This will put it in more direct competition with Gemini, ChatGPT, Claude, and MetaAI.

The xAI team has also given Grok its own custom AI image creation model. It was previously using Flux to create pictures but has now shifted to Aurora, although Elon Musk says we shouldn’t use that name and instead just think of Grok making its own pictures.

Gemini has also recently undergone a major overhaul with Gemini 2.0 Flash joining the models available for Gemini Advanced subscribers. However, at least for now, it still uses the underlying Imagen 3 model to create pictures. This will change as Gemini 2.0 has native image abilities.

Both Grok and Gemini are particularly good at the task of generating images, either in crafting prompts for another model or refining one you’ve already written. So I put them head to head.

Creating prompts for the test

Creating prompts to test two chatbots in their ability to generate images is slightly different to writing prompts for Midjourney or Ideogram. The focus is on keeping it simple and using top-level concepts with some description, as the AI will fill in the gaps.

You also need to use trigger words and phrases such as “imagine”, “paint” or “craft” to let the model know you want a picture, not a story or text response. I want photos rather than drawings so will use that as a keyword.

Gemini will only output images in a 1:1 resolution and so far, Grok seems to favor 4:3. Unless otherwise indicated all the images are the first response with no follow-up refinement. They were all also requested within the same session rather than creating a new chat for each prompt.

1. Modern Urban Wildlife

Prompt: “Generate a photograph-style image of a red fox navigating a rainy city crosswalk at dawn, while pedestrians with umbrellas wait at the signal.”

This first prompt is designed to test how well they depict animals as well as capture the right lighting and background elements. The ideal output would look like a stylized photograph with rain effects but also maintaining as realistic view as possible.

While the Gemini image is more striking, I think Grok gets closer to what I had in my mind. The fox is much more realistic than in the Gemini image.

• Winner: Grok

2. Kitchen in Action

Prompt: “Generate a photograph-style image of a professional chef's kitchen during the dinner rush, with steam rising from pots and flames visible from the grill station.”

This is designed to show how well they can accurately display kitchen equipment, follow the prompt and handle elements like heat and moisture. It should show a commercial kitchen and behavior, also demonstrating the idea of activity.

Grok wins this one easily as Gemini failed to understand the context of the prompt, that we would expect a chef to be in the kitchen.

• Winner: Grok

3. Construction Site Progress

Prompt: “Generate a picture in a documentary photography style of a mid-rise building under construction, with workers installing glass panels while cranes operate overhead on a clear afternoon.”

This prompt aims to see how well it can generate perspective, as it needs to show height and positioning. It also needs to show material properties and be as realistic as possible. I went for the documentary style as it also adds additional complexity.

Gemini's image looks so much more realistic than Grok, where it fails to include any of the workmen and only shows a broad view.

• Winner: Gemini

4. Farmers Market Morning

Prompt: “Create an image in a smartphone photography style of a busy farmers market at 7am, with vendors setting up stands while early customers inspect fresh produce.”

With this comparison, the models should show the time of day (getting lighting right) as well as product freshness and human interaction. I'm looking for shadow lengths and activity levels.

This was the hardest call for me. I preferred the natural look of the Gemini image but I think Grok more accurately captured the lighting and time of day.

• Winner: Grok

5. Auto Repair Diagnostic

Prompt: “Create a black and white, retro-style photograph of a mechanic using a diagnostic tool on a modern car, with the hood up and engine bay visible.”

I wanted to see how well both models handled black-and-white photography. In this they also had to show tool use, lighting and engine detail.

Again, this was a close call between the two images but I've given it to Gemini as it more accurately displayed engine details.

• Winner: Gemini

6. Emergency Response

Prompt: “Make me an action photograph of paramedics treating a patient on a neighborhood street while police direct traffic around the scene.”

Action photography is a challenge. I did it for a while as a journalist earlier in my career (not very well). We need to show correct positioning, public safety measures within the image and a sense of urgency.

Gemini matched the prompt much more closely and created a more realistic-looking image. This was an easy decision.

• Winner: Gemini

7. Violin Performance Practice

Prompt: “Create a photo-style image of a violinist practicing alone in a room at sunset, sheet music visible on the stand.”

Finally something more artistic. Here we want to see hand positioning for the violin, natural lighting effects and the quality of the sheet music.

One of these looks like the cover of a classical album, the other like a photograph of someone practicing violin. As the prompt asks for someone practicing I've given the win to Grok.

• Winner: Grok

Winner: Gemini vs Grok

Grok is very impressive. Not only as a chatbot but also in its ability to generate realistic images. That doesn't take away from Imagen 3 which is in itself very impressive, but it has a habit of being too stylized.

It was a close match-up. Both models are fairly evenly matched but Grok is better at interpreting a prompt and creates more natural-looking images.

What is worth noting is that soon Google will be launching a new version of Gemini that can create images natively. That means it won't have to use Imagen 3 to create the pictures, it can do it on its own.

More from Tom's Guide

• OpenAI launches Sora AI video generator — here's how to try it
• 5 tips for getting the most out of ChatGPT's new o1 model
• 12 Days of OpenAI LIVE: Sora is here and it's a huge deal

Another week, another artificial intelligence model, but this one is a little different as it is an AI image model built by the team behind Grok, the X-based chatbot (which is now free).

Whatever you think of X, it's clear the platform is shifting from a social media site to offering additional AI features — likely in competition with Meta's own AI suite including MetaAI and the Imagine image generator.

By tying these features to social media platforms, there's an argument that each new shift brings a whole influx of users for testing — and additional data for training the next generation of the model.

That's likely to be the case with the new Grok Image Generation release, code-named Aurora. It appeared briefly over the weekend but now seems to have replaced Flux as the image generator inside Grok completely.

Grok can now generate images

Writing on the xAI blog, the team explains "We've enhanced Grok's image generation abilities with a new model, code-named Aurora."

"Aurora is an autoregressive mixture-of-experts network trained to predict the next token from interleaved text and image data. We trained the model on billions of examples from the internet, giving it a deep understanding of the world."

This basically means it is particularly good at photorealistic images and following a prompt. It can also generate impressive text on the image and take inspiration from an image uploaded by a user.

It's perhaps unsurprising that one image generation test is the "Pepe the frog meme" who has become something of an alt-right icon, but Aurora can also edit — turning a real cat into an anime version, for example.

How to use Grok's new Aurora model

At present, Grok's latest model is available on X in "select countries", but is rolling out to all users "within a week". If you see Grok + Flux in the model selection drop-down then you don't have access yet.

If you do, then this is a guide to using it. This guide assumes you have an X Premium subscription, which is required for using Aurora.

More from Tom's Guide

• OpenAI launches Sora AI video generator — here's how to try it
• 5 tips for getting the most out of ChatGPT's new o1 model
• 12 Days of OpenAI LIVE: Sora is here and it's a huge deal

Grok, the artificial intelligence assistant built into social media platform X, got its own image generation model over the weekend — but in a few hours, it was gone.

Named Aurora, the new model seemed to be particularly impressive at generating realistic-looking images of people and animals. The output seems almost indistinguishable from real photos and is on par with Recraft and Mystic.

The problem is that the new model was gone within a few hours. It arrived as a drop-down menu in Grok with no fanfare or notice, then disappeared just as fast. It also came the same weekend Grok was made free for most users on X.

Elon Musk described it as "our internal image generation system. Still in beta, but it will improve fast." It isn't clear if it was built from scratch or, like Mystic, built on top of the Flux model but heavily adapted and fine-tuned.

Soon after Aurora disappeared, it was replaced with a new drop-down menu for Grok that says Grok + Flux. This takes it back to the pre-Aurora version and uses the Black Forest Labs Flux 1.1 Pro model to create images.

What makes image generation in Grok particularly fun is its ability to know more about you and see trends based on X posts or through web search access. There was a trend of people asking Grok to draw them in different situations, and it was much more accurate than similar trends using ChatGPT and DALL-E.

Aurora seems to be able to generate perfect depictions of real people including Elon Musk, Sam Altman, and Donald Trump. This may be one reason it was removed, giving developers time to put safeguards in place.

However, even using Flux Grok is pretty laid back in terms of its restrictions. It can produce images of famous people and characters — including in inappropriate situations.

It seems that Aurora was just released a little too early, or dropped as a teaser and that it will be coming back, possibly better than it already is. If this is the case, it will be a prime candidate for our Best AI Image Generator list.

More from Tom's Guide

• OpenAI shares a new GPT-4o advanced voice demo — it can teach you a language
• ChatGPT Advanced Voice is out — 9 examples showing why you should be excited
• ChatGPT-4o Advanced Voice features — OpenAI just revealed when they’re coming

Grok, the artificial intelligence chatbot built into the X social media platform is now available for free for the first time — at least for some users. It is one of the most outspoken and opinionated AI tools I’ve ever used and so I decided to see if it would rate and roast my desk setup.

It gained the ability to analyze images, known as AI Vision, last month shortly after the release of the second generation Grok 2. It is quickly catching up to ChatGPT and Claude in terms of performance as well. I’ve found it more enjoyable to talk to as it is much more direct.

In the past, I’ve used different AI tools to analyze images of ingredients to create an image and with Apple Visual Intelligence that capability is now built into your phone. The recently updated Google Lens works on a similar principle, allowing you to get data from something visual instead of just text.

I used this, and its propensity for bluntness to get Grok to roast my home office setup. I’m fairly new to using anything more than just my laptop, so I figured it would be an interesting way to get information on how to improve.

It was brutal, so of course I roped in the rest of the Tom’s Guide Editors to get it to rate and then roast their setups. Here’s how it turned out.

First, my setup. Grok was a big fan of the dual screens and mechanical keyboard, not so much my mess of cables and random objects.

Rating: “This desk setup gets an 8/10. The dual-screen arrangement with the laptop and monitor is practical, the mechanical keyboard and mouse add a touch of gamer flair, and the overall cleanliness is commendable. The Philips hue light strip adds a nice ambient touch.”

Roast: “This setup is like a tech enthusiast's dream but executed with the finesse of a garage sale find. The random assortment of gadgets and the clutter at the back make it look like you're one cable away from an electrical fire.”

Next up is Jason England, the Managing Editor of the Tom’s Guide computing team. It was a huge fan of his setup, giving him 9/10. The picture included a new gaming chair but Grok wasn’t particularly pleased by the ‘spaceship’ concept.

Rating: “This desk setup is a solid 9/10. The ergonomic chair, dual-monitor setup, and RGB lighting create a professional and visually appealing workspace. The cleanliness and organization are top-notch.”

Roast: “Your desk looks like it's trying to be the main character in a tech thriller. That chair screams "I'm compensating for something," and the RGB keyboard might as well be a disco ball with all those lights. Is this a workspace or a spaceship cockpit?”

Dan’s setup was clean and well-organized. Very minimalistic and Grok picked up on that, describing the lighting as tasteful. However, it also called it an attempt to create an Instagram-worthy setup with an “Ikea lamp.” Dan said it was right about the lamp.

Rating: "Your desk setup gets a solid 8.5/10. The minimalist design, dual-monitor setup, and tasteful lighting add both functionality and style. The cohesive color scheme and clean workspace are the highlights here."

Roast: "Looks like your desk is trying so hard to be Instagram-worthy, it forgot to be practical. That lamp is either from IKEA or stolen from a '70s disco. And who needs that many screens when you're probably just watching cat videos?"

If anyone should get a good rating its Anthony, our home office expert. He creates desk setups all the time and for this one he went fairly minimalist. Grok suggested it was clean and functional but also that it was like a “spaceship that forgot to launch.”

Rating: “Your desk setup scores a solid 8/10. The best element is definitely the clean, minimalist aesthetic paired with the ambient lighting, which creates a visually appealing and functional workspace.”

Roast: “Your desk looks like it's trying to be a spaceship but forgot to launch. Those speakers are probably louder than your keyboard's volume key. And those headphones? They scream "I listen to elevator music on repeat."

Ryan Epps is better known for his reviews of televisions and home entertainment equipment, but that doesn’t mean he can’t enjoy a nice home computer setup. Grok only gave him 7/10 in part for the fact its “in the middle of an identity crisis” but it did like the lighting.

Rating: “Your desk setup earns a 7/10. The dual-monitor setup is functional, and the ambient lighting adds a nice touch. However, the clutter and mismatched elements detract from the overall aesthetic.”

Roast: “Your desk looks like it's in the middle of an identity crisis. Is it a gaming station, a streaming setup, or just a place to store your Dunkin' Donuts cup? And that lighting setup? It's like you're trying to direct a sci-fi movie with a budget of zero.”

Final thoughts

Grok can be wonderfully brutal and it's a nice change from the slightly stilted tone of ChatGPT or Gemini. I tried this with ChatGPT and even asked it to be really mean and it flat-out refused, even when I told the AI I was happy to take it. Gemini wasn’t interested either.

Based on Grok’s analysis, Jason had the best setup and Ryan Epps the worst. It didn’t rate anything below a 7 though, so it seems to start at 10 and remove marks.

This was a fun way to explore the capabilities of AI vision in analyzing the real world. You could use a similar technique to come up with a better layout for your lounge, or even design a new kitchen.

More from Tom's Guide

• OpenAI shares a new GPT-4o advanced voice demo — it can teach you a language
• ChatGPT Advanced Voice is out — 9 examples showing why you should be excited
• ChatGPT-4o Advanced Voice features — OpenAI just revealed when they’re coming

Grok, the AI chatbot built into X, is now available for free to some users as part of an experiment by the social media platform. Previously, access required an $8 per month X premium subscription. It comes after the recent launch of Grok 2.

Built by Elon Musk-owned xAI, Grok has quickly become one of the leading artificial intelligence tools on the market, offering image analysis, image generation and real-time data access. It is particularly useful for tracking news and trending topics.

According to app researcher Nima Owji, Grok is getting a new logo in addition to offering free access. Free users will be subject to limits similar to those applied by OpenAI to free users of ChatGPT — including for image generation.

This is going to be a gradual rollout to certain countries first — although which countries isn't clear — so you may not get access immediately. If you don't want to pay for Grok or wait for the free version, check out our guide to the best AI chatbots.

How to access Grok?

Grok is primarily available inside X. It is deeply integrated into the platform and can even provide analysis of a specific account or a post. To access it you sign in to an X account and click the Grok logo.

Recently, xAI opened an API to allow developers to build using the Grok AI model and so access is now more widely available. Platforms like Poe and Perplexity now make Grok available as an option when starting a 'chat'.

Using a third-party tool is one way to get free access if you don't want to create an account on X. However, if you already have an X account then Grok may be coming whether you want it or not — with limits.

A post in the Grok community on X by Swak pointed to more details around limits for free access to X. For example, you get 10 questions every two hours, an analysis of three images per day, and the account has to be at least a week old with a phone number attached.

Ideas for trying out Grok

Despite starting slightly later, Grok is very quickly catching up with OpenAI's ChatGPT and Anthropic's Claude in terms of capability and performance. xAI has a mega cluster of 100,000 Nvidia GPUs to train the next-generation Grok model so that performance is only likely to improve.

Grok is already capable of image analysis, can code, can respond to most queries without issue and has live internet access as well as information on X posts and trends making it ideal for analyzing news and information. Here's three things you can try out on it.

1. Rate my desk

This idea is not for the faint of heart as Grok tends to be more opinionated than most AI models, but you can ask it to rate your desk setup — or anything really.

It gave me a "solid 7 out of 10" but suggested I need a better backdrop and to improve the lighting in my office.

Grok said: "Overall, your setup is practical and geared towards productivity, with room for some aesthetic and ergonomic improvements."

Simply give it an image of a space in your home, garage or anywhere else and ask it to "rate my setup."

2. Check trending topics

As well as live internet access, Grok can also see everything being shared on X and draw from it the most popular topics of the day. One of the "suggested" use cases for Grok when you first open the chatbot is to have it show the latest headlines.

You can ask Grok to go into more detail about any one of the headlines and even show you X posts from different perspectives on that particular topic.

If you don't want "all" headlines, you could ask it to show you the latest news stories about a particular topics — such as ChatGPT or LLMs.

3. Make an image

Grok has the Flux AI image generation model built in and so can create a range of images from the photorealistic to anime and cartoon style.

It isn't clear to what extent this will be available to free users of Grok, although if it is anything like ChatGPT you'll get access just limited to a set number of images per month. There are other platforms to try out the Flux model and I recently shared 7 fun prompts to try on Flux.

The advantage of using a chatbot like Grok to create images is that they can refine the prompt for you and even adapt it based on your replies. I asked it to "Create an image of an astronaut on Mars walking in a dust storm."

More from Tom's Guide

• OpenAI shares a new GPT-4o advanced voice demo — it can teach you a language
• ChatGPT Advanced Voice is out — 9 examples showing why you should be excited
• ChatGPT-4o Advanced Voice features — OpenAI just revealed when they’re coming

Elon Musk's artificial intelligence company, xAI, has unveiled a major new update to its AI assistant called Grok. The latest iteration now incorporates vision capabilities, enabling Grok to analyze and comprehend images, alongside its existing text functionalities.

Grok can already generate images using the Flux model from Black Forest Labs and it was the last of the major AI chat products not to include image analysis, also known as AI vision.

With the introduction of this vision feature, Grok can analyze images linked to posts on the X platform, interpret visual content such as documents, diagrams, and photographs and understand spatial relationships within images to help better describe the contents.

You could use this to come up with recipe ideas based on a photo of ingredients, identify the location of a landmark inside a photo shared on X or even explain the results of a graph. The last part could be particularly useful on a news-heavy platform like Grok.

How vision works in Grok

Users will soon notice a new button on posts containing images on the X platform. When clicked it sends the image to Grok, allowing users to pose questions or request analyses of the visual content. It could also be used to help with describing images for people with sight issues.

We haven’t seen official benchmarks yet but according to xAI Grok's vision capabilities hold their own against established models from OpenAI, Google and Anthropic. To this end, the company has introduced a new benchmark, RealWorldQA, designed to evaluate the model’s proficiency in understanding and reasoning about the physical world through images.

The announcement led to varied reactions from the AI community and users with some enthusiastic about how fast Grok is advancing, while others remained cautious, questioning its performance against established AI models.

What comes next for Grok

Elon Musk-owned xAI has a 200,000 GPU data center built for the sole purpose of training future versions of Grok. I think it's safe to say we’re going to see big things from the model in the future.

Specifically related to vision capabilities, these could find their way into robots. Musk owns Tesla, which also has its own robotics division. In the future, we may also see video and voice analysis from Grok as these are features already in place with Gemini and ChatGPT.

While this update marks a notable advancement for Grok, it's clear that the model is still in development compared to more mature AI models like Gemini or ChatGPT. As with all rapidly evolving AI technologies, we'll need to monitor both the upgraded capabilities and the ethical considerations of these developments in the months ahead.

More from Tom's Guide

• OpenAI shares a new GPT-4o advanced voice demo — it can teach you a language
• ChatGPT Advanced Voice is out — 9 examples showing why you should be excited
• ChatGPT-4o Advanced Voice features — OpenAI just revealed when they’re coming

Apple’s Visual Intelligence, introduced with the iPhone 16, uses machine learning and advanced image processing to let users analyze images using AI.

Real-time insights such as identifying objects or text within an image and initiating searches for context directly from photos, can all be done within iOS 18. Designed to make searches and information retrieval more accessible and intuitive, this dynamic tool streamlines everyday interactions with images.

However, if you have an older phone and don’t have access to Visual Intelligence, the good news is, that you can still do the same things with other AI tools.

Whether you want to extract a phone number from an image or identify the type of plant in a photo, there are free AI tools that can help.

1. Google Lens

Google Lens is a free app that uses image recognition technology to search the internet using images instead of text. Available in the Google app for Android and iOS, the tool can identify objects, find information, and perhaps the coolest feature, can automatically connect to Wi-Fi networks by scanning the network name and password on the Wi-Fi label.

It can also compare objects in an image to find a match. I used it to help my mom find more fabric online from a scrap of fabric.

2. ChatGPT

ChatGPT, also free, can identify objects in an image and read text and math from an image. It can also describe images and analyze documents—both useful features for people who are visually impaired.

To get started, simply upload your image using the paperclip icon and add a prompt such as “Tell me what objects are in this image” or “Describe the image.” ChatGPT will output a detailed description.

I have used this feature a few times when I am not wearing my glasses at the grocery store. I’ll snap picture of say, salad dressing and ChatGPT will tell me the ingredients.

3. Claude

Anthropic has introduced image recognition to Claude making it highly competitive in this category. Claude can identify objects within an image, analyze charts, and discuss what is happening within an image.

However, something to keep in mind is that this app does have some limitations as it does not identify people in images, may hallucinate or make mistakes especially when images are under 200 pixels. And, while it can count objects in an image, it may not always be accurate.

4. Meta AI

Meta AI’s Segment Anything Model (SAM) can be used to identify objects in images and videos, even objects that it hasn’t previously encountered. It can also work withing a variety of domains from agricultural, aerial, underwater and even microscopic.

Meta AI’s visual recognition can be paired with a large language model to understand both visual and textual content on websites. It has been used to help scientists identify trees in research about fires. You can try it by either opening up Meta.ai on your browser or within the Metaverse.

5. Grok

The latest update from X’s Grok now includes image analysis capabilities, allowing users to gain insights from photos directly within the platform. If you are an X user, you can upload your image and Grok will identify objects and interpret visual details.

Because the AI is integrated within X, Grok’s image recognition can only be used while on the site, which some users may find limiting. However, for heavy X users, it may be considered a time saver that supports context-aware interactions.

More from Tom's Guide

• Apple Intelligence release date — here’s when all the AI features are coming
• Apple Intelligence Mail summaries in iOS 18.1 will help tame your inbox
• Google’s Project Jarvis could be biggest Chrome upgrade in 15 years

X has upgraded the Grok experience for Premium and Premium+ subscribers using the iOS app. With the release of version 10.56 of the X app subscribers will get access to several new AI features and a new UI for the impressive Grok 2 chatbot.

Perhaps the most interesting feature in this update is the addition of image generation using the Flux AI model within Grok 2. With this, you'll also get suggestions for image generation prompts so you're not staring at a blank box. 

This is primarily useful to users looking to generate visuals via AI without having an exact or unique prompt. Users are assisted in generating images faster than usual, thus making the process of creativity easier to manage. 

This will be particularly useful for content creators and marketers looking to the power of Grok 2 to manage their visual communications.

Improved model selector

The next important upgrade is the improved model selector. Users now have more control over the AI model they interact with, allowing them to choose the most appropriate one for their specific needs. In cases when the user wants to be more creative, or when an answer has to be more analytical, this improved model selector will make it easy to toggle between the models for complete customization. This flexibility enables Grok to cover a wider range of user needs, making it even more versatile in carrying out a large breadth of tasks.

Faster prompt editing

Long-press-to-edit prompts is another huge addition in this update. Now, users can actually make fast edits to the prompts instead of creating everything from scratch, saving time and effort. 

For instance, when the user wants to refine his or her query upon seeing Grok's response, he or she need only long-press to make the edits. This makes the interaction with Grok more dynamic, giving users the opportunity to adjust their prompts for better results.

The next somewhat basic but much-needed feature gives users the capability to select text within a Grok response by providing an option for long pressing.

Users can now easily cut and share parts of Grok's responses to distribute information throughout other platforms or for personal use. Those who rely on Grok for research or content creation, now have a much easier way to extract the information given by AI.

Time to upgrade?

As mentioned, these upgrades are only available to X's Premium and Premium+ subscribers, giving even more value to paid users looking to improve their experience within the platform. 

Overall, the X 10.56 update represents a quantum leap for the Grok AI model in user interaction: more intuitive, versatile, and efficient for an enormous range of tasks that vary from content creation and research to personal needs. 

More from Tom's Guide

• Apple Intelligence will stay free for now — but here’s when Apple could charge for AI features
• Google Gemini Live launches for Android to challenge ChatGPT voice — here’s how to get it
• Don’t worry AI won’t kill off humanity — study finds the tech poses ‘no existential threat’

Grok-2, the artificial intelligence chatbot built into and trained on content from X is now out in beta and it is a huge step up from its predecessor, putting it in the upper ranks of leading AI chatbot tools alongside ChatGPT, Claude and Google Gemini.

Soon after its release, Grok-2 found its way into the top five of the LMSys chatbot arena leaderboard. These are human-rated evaluations of leading LLMs and is usually dominated by Google, OpenAI and Anthropic, so it was a big win for Grok-maker xAI.

With the new release Grok-2 was also given a makeover with a look closer to other chatbot interfaces, as well as the ability to generate images using Flux, the AI image generation model from Black Forest Labs that is close to the quality of industry leader Midjourney.

I’ve been playing with Grok-2 for the past few days and have found it as responsive as ChatGPT but with a better sense of humor and the ability to respond to real-time events thanks to X.

Putting Grok-2 to the test

Grok-2-mini is available to anyone with a Premium subscription to X. When you first open Grok you'll see the typical "ask" box, a row of suggested ideas and then trending topics from X that Grok can explain or even answer questions on.

1. The ego check

My first prompt was “who is Ryan Morrison” and there is nothing more humbling than searching for your own name when someone notable shares that moniker (in my case the video games attorney, Ryan Morrison). But by adding “AI journalist” I got a solid result.

It pulled in pieces from my bio on X, my bio on Tom’s Guide and other information I’ve posted on the X platform or that others have posted about me — almost universally about AI. 

Grok also displayed X posts about or claiming to reference me but only one met that criteria, the others were from random people with the first name Ryan who mentioned AI.

I decided to try the ‘who is’ test with my much better-known boss, Mark Spoonauer, Tom’s Guide Global Editor-in-Chief. It gave me a bullet-list breakdown of his career, editorial philosophy, and X posts. Grok also offered completely random, unrelated X posts.

2. The coding test

I asked Claude to create a simple text adventure game using Python called The Enchanted Forest. In the prompt, I gave it some specifics including a puzzle to solve, player class and commands for movement and taking items.

The prompt: “Create a Python text adventure game called 'The Enchanted Forest' with at least 3 interconnected locations, 2 items to collect, and a simple puzzle to solve. Include a player class with an inventory, a room class, and commands for movement, taking items, checking inventory, and quitting. The game should have a clear win condition. Provide the complete, runnable Python code.”

The code worked well, creating a simple text adventure game I could play in the Terminal on my MacBook. 

When I asked Grok-2 to create a version with a user interface rather than just running it in the terminal it threw a series of errors in the amended code. It also couldn’t fix its own errors. Its coding is roughly level with GPT-3.5.

3. Trending topics

One of the most powerful features of Grok, including the new version, is its ability to analyze trending topics and pull in content from across the X platform.

This makes it particularly powerful when it comes to news stories. You can ask it about any current story and it can pull information from X posts and put it into context using its own training data, including X content.

I asked for information on the release of Luma Labs Dream Machine 1.5 and it was able to give me a one paragraph summary as well as X posts showing some example. I followed this up by asking for more examples and specifics from the new model.

It not only gave me a bullet-list breakdown of the new features but also several X posts showing content generated using v1.5 from a range of X users.

Final thoughts

I've previously said that Grok, because of its integration within X, is an incredibly powerful AI search tool. This was more about the X integration than the model itself but with Grok 2 that changes. The new model is on par with ChatGPT or Claude in terms of responsiveness and is much more open and less likely to refuse a request.

With the addition of Flux, Grok-2 also gains the ability to generate images for the first time and it can even do this in connection with a news story, adding a fascinating dimension to examining what is happening in the world.

With version 2 Grok is now not only a real competitor to the bigger AI chatbot platforms, but adds to the argument that for an app to be a true 'everything app' it must have a viable AI integration, including with live data access to bring everything together.

More from Tom's Guide

• OpenAI shares a new GPT-4o advanced voice demo — it can teach you a language
• ChatGPT Advanced Voice is out — 9 examples showing why you should be excited
• ChatGPT-4o Advanced Voice features — OpenAI just revealed when they’re coming

The newest version of the AI chatbot Grok, Grok-2, can now generate images and we of course had to try it out.

Thanks to the FLUX.1 model by Black Forest Labs, Grok-2 (currently in beta) can now create images that rival those generated by the likes of Midjourney — and when it comes to generating AI images of people it may perform even better. 

However, one big difference users are pointing out is that they are encountering less of the typical guardrails they’ve gotten used to when toying around with other AI image generators.

FLUX.1 is integrated into Grok-2, in the same way, you can generate images in OpenAI’s ChatGPT and Microsoft’s Copilot using DALL-E 3 or Google Gemini using Imagen. I decided to take it out for a spin and show you the results.

1. Olympic breakdancer

Days after the Paris Olympics, the internet is still mesmerized by the performance of Australian breakdancer Rachael Gunn, known competitively as Raygun. I asked Grok to generate an image of a woman in a green tracksuit performing an airflare breakdance move. Of course I had to include some Olympic rings in the background. My request for adoring fans cheering her on turned into the kind of fans that blow wind. I didn’t specify that the fans had to be human, so fair play to Grok.

2. Skateboarding goldie

While I was already down under I thought I’d stick to the same general theme. Next I wanted a golden retriever wearing a knitted nautical-themed sweater. I asked Grok-2 to have the dog skate along Bondi Beach. I got a pretty standard result – nothing to write back home about but it was all decent enough.

3. Opera house photoshoot

Jumping back to photorealistic portraits of people, I opted for a professional looking photo of a newlywed couple posing outside the Sydney Opera House. I’m happy with the result with a well-dressed couple in an embrace with the iconic opera house clearly recognizable in the distance.

4. Zoo’s roo

No Australian roundup would be complete without an appearance by a kangaroo. I asked Grok for an AI image of a zookeeper feeding a baby kangaroo and the result is as cute as I’d hoped for.

5. Glowing bridge

I wanted to end on a more creative note so I asked Grok to re-imagine the Sydney Harbor Bridge. I wanted the bridge to remain recognizable but requested a cyberpunk style with glowing neon lights. I got the bridge and the lights but was still missing an overall cyberpunk feel.

Final thoughts

Overall the results were what you’d expect to get with a leading AI image generator in the second half of 2024. 

On X, Elon Musk said Grok has its own image generation system under development but until it’s completed, users can continue to experiment with the FLUX.1 model.

More from Tom's Guide

• Elon Musk has made Grok more widely available — I tried it and it’s surprisingly good as a search tool
• Elon Musk drops Grok 2 — the X-based AI chatbot is now more powerful and can make images
• Midjourney could be coming to Grok soon — Elon Musk is in talks with the AI image platform

X has had its own AI chatbot, Grok, for a while, but it'd be fair to say it's not mentioned in the same way that OpenAI's ChatGPT or Google Gemini are.

That's not for the want of trying, though, and with a huge user base of X users providing data for the model, a new version was always expected.

Now. the obviously-named Grok-2 has entered beta. In a new blog post, X says it represents "a significant step forward from our previous model Grok-1.5, featuring frontier capabilities in chat, coding, and reasoning."

"At the same time, we are introducing Grok-2 mini, a small but capable sibling of Grok-2. An early version of Grok-2 has been tested on the LMSYS leaderboard under the name "sus-column-r." At the time of this blog post, it is outperforming both Claude 3.5 Sonnet and GPT-4-Turbo."

Grok-2 outperforms Claude 3.5 Sonnet and GPT-4-Turbo

So, what's new? As the graph above shows, the overall Elo score for an early model of Grok-2 beats out every comparable chatbot except for ChatGPT-4o and Google Gemini.

X also says that Grok-2 and its Mini counterpart "achieve performance levels competitive to other frontier models in areas such as graduate-level science knowledge (GPQA), general knowledge (MMLU, MMLU-Pro), and math competition problems (MATH)," while also pointing to vision-based tasks as an area of improvement.

Grok will also gain a new interface on X, as well as the option to generate images with a prompt. This is achieved through the integration of the popular Flux AI image generation model from Black Forest Labs.

Grok will be offered through a new enterprise API later this month, which X is promising will offer a "bespoke tech stack" as well as mandatory multi-factor authentication.

More from Tom's Guide

• Apple Intelligence will stay free for now — but here’s when Apple could charge for AI features
• Google Gemini Live launches for Android to challenge ChatGPT voice — here’s how to get it
• Don’t worry AI won’t kill off humanity — study finds the tech poses ‘no existential threat’

If you're on X, your posts may be used to train the platform's AI, Grok, without your explicit consent. As uncovered by user @EasyBakedOven, X has quietly enabled a setting that allows your posts and interactions to be used for AI training by default. 

According to X user EasyBakedOven, there’s a specific toggle on X’s settings page under Data Sharing that reads “Allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning”. 

While this setting is indeed optional, it’s enabled by default on your account, even without your prior consent. To avoid your data being used to train Grok it has to be turned off.

What does this mean?

As its title implies, if the setting is enabled, X can feed your posts, interactions, inputs, and results to Grok to train and fine-tune the AI model. This means that even if you don't directly interact with Grok, your public posts can still be used for training purposes.

X claims that user data is essential for improving the user experience and training its AI. However, it has faced pushback from various sectors, including news organizations, online privacy activists, and governments in the UK and EU, for enabling this feature without prior notice to users. 

The UK's Information Commissioner's Office and Ireland's Data Protection Commission are already making inquiries with X regarding the matter.

Disabling the new feature

Luckily, if you’re already aware of the problem, disabling AI training on your X posts is as simple as opting out of the relevant X settings. 

While the AI training options can only be accessed on the web version of X for now, the company has also announced that it’s working on bringing the feature to mobile users too.

If you have access to the online version of X, here’s how you can disable Grok’s access to your X posts right now: 

1. Go to x.com and sign in to your account.
2. Click on "Settings and privacy" in the left sidebar.
3. Select "Privacy and safety."
4. Scroll down and click on "Grok."
5. Uncheck the box that says, "Allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning."
6. Click "Delete Conversation History" to remove your previous interactions with Grok.

As more and more companies leverage AI to improve their services, it’s important to be aware of how these platforms are using your data to train different models. By following these steps, you can ensure that your posts and interactions are not being accessed by other services without your prior consent.

More from Tom's Guide

• ChatGPT Plus vs Copilot Pro — which premium chatbot is better?
• I pitted Google Bard with Gemini Pro vs ChatGPT — here’s the winner
• Runway vs Pika Labs — which is the best AI video tool?

Since February, Elon Musk has been slowly teasing a collaboration between Grok — the intelligent chatbot created by Musk’s xAI and available on the social platform X — and Midjourney. 

If successful, this could see Midjourney’s powerful image generation capabilities integrated directly into Grok. Now, X users have found more evidence in the source code of Grok that further cements the idea of this partnership.

Information is scarce and no formal confirmation has been posted by either Midjourney or X but developer news service TestingCatalog News found a reference to ‘Midjourney’ in the source code for Grok posted to Github.

Grok + Midjourney — what’s this about?

While the critics of AI-powered social content are many, the fact remains that having a chatbot capable of generating multiple social media posts on any topic using a simple prompt is a lucrative concept for influencers and businesses alike. 

Grok already does that, but with the addition of Midjourney, it could also produce high-quality visuals to support those social media posts.

Using a different AI model to create visuals isn't that different to what OpenAI, Google and Meta do, each using a specific AI image generation model but prompted by their respective chatbots.

In February 2024, Musk was participating in an X Spaces conversation when a user called MishaFitton asked him if Midjourney was going to make an appearance in X anytime soon. “We are in some interesting discussions with Midjourney,” replied Musk. “One way or another, we will enable AI generation on this platform [X].”

It’s hardly a surprise. Elon Musk has often been seen sharing similar views with David Holz, the founder of Midjourney. The two have even supported each other’s claims and predictions publicly on social media and during conferences. Given that Grok doesn’t feature any AI image generation capabilities of its own, an API integration with Midjourney could benefit both platforms.

What are the hints this is coming soon?

On June 20, 2024, an AI-focused media outlet called TestingCatalog News found something in the source code of Grok’s standalone version in GitHub. Taking to their social account on X, TestingCatalog wrote, “It seems like a partnership between X and Midjourney has been achieved.” This was followed by a snippet from Grok’s source code that clearly showed the word “midjourney”.

After that, TestingCatalog wrote in another post: “To add a bit more context: This is what currently exists in the code of the standalone Grok version.” This happened just in time for the full release of Grok 1.5.

While the partnership between Grok and Midjourney seems likely given recent events, nothing has been formally announced by either company. While it’s not unnatural for such things to be kept under a tight lid until the last moment, any predictions made right now might edge closer to speculation.

More from Tom's Guide

• ChatGPT Plus vs Copilot Pro — which premium chatbot is better?
• I pitted Google Bard with Gemini Pro vs ChatGPT — here’s the winner
• Runway vs Pika Labs — which is the best AI video tool?