Grok AI Step-by-Step Guide 2025: Everything You Need to Know

0:00 In the last few videos, we talked about Chat GPT, Google Gemini, and Claude. Today, we're going to take a look at Grock, a large language model chatbot developed by Elon Musk's XAI. Grock was launched in November 2023, about a year later than Chat GPT, but XAI managed to catch up with early players like ChatGpt and Gemini in less than 2 years. From my point of view, Grock's features and performance are already on par with those competitors. And Grock 4, which just came out in July 2025, actually scored the highest in several benchmark tests. I'm going to walk you through some of Gro's features one by one, like web search, deep search, think

0:47 mode, and create image. These can really come in handy for different tasks, whether you're doing research, reasoning, or creative work. All right, let's get started. Right now, there are two ways you can use Grock. One is by going straight to grock.com, and the other is through the Xplatform. On the Xplatform, you'll see a Grock icon on the left side. Just click on it, and Grock is already built right into X. The interfaces are a bit different. On grock.com, you get the full set of features. On the X platform, since Grock is deeply integrated with X, the interface focuses more on searching through X's data. You'll also notice shortcuts for create images and edit

1:32 image, which makes it super convenient for posting content. But honestly, you can do the same things on grock.com, too. So, I'll stick with grock.com for this walkthrough since it has all the features in one place. Once you're on grock.com, you can just type right into the chat box in the center to start talking with Grock. Below the chat box, you'll see a row of example you can try out things like create images, research, edit image, latest news, or personas, which let Grock reply in a preset style. And just keep in mind, if you're not logged in, your number of chats and available features will be limited. So, let's go ahead and log in first. All right, let's jump right in and try a simple chat. You can ask Grock to

2:18 explain a term or concept. For example, what is stablecoin? It'll give you a long detailed answer. Down here at the bottom, you've got options to regenerate response, copy, share, like, or even export it as a PDF. In my case, I feel the answer was a bit too long and formal. So, I asked Grock to explain it in simpler terms, and that makes it much easier to understand. Besides asking it to simpler the answer, you can also click on custom instructions down in the bottom right corner to adjust how Grock replies. By default, it's set to formal style, but you can switch it to concise style or even customize the tone and length further to match what you want. Here's another example. I asked Grock

3:05 for the latest news on Tesla robo taxi. Grock will decide on its own if it needs to browse the web, and it'll automatically turn on the search feature. The sources it pulls from include exposts and regular web pages. Some exposts are just a few hours old, so the info is really up to date. Sometimes for certain questions, Grock will pop up a search deeper option. That's actually the deep search feature, which I'll show you more about in just a bit. All right. Now, let's take a look at how to choose Grock model for you to use. Down here in the bottom right corner of the chat box, you'll see the option to switch models. For free users, the default is the Gro 3 model, which runs pretty fast. Since I have a Super

3:51 Gro subscription, I can also choose the newest Gro 4 model. Gro 4 has much better reasoning skills, especially when you're dealing with complex or specialized topics. Above that, there's even a Gro 4 heavy model, which is part of the Super Gro heavy plan. That one costs $300 a month. That's 10 times the regular Super Grock plan, which is $30. Honestly, most of Grock's features are available to free users, too. The Super Grock subscription mainly gives you four extra perks. First, you get access to the Gro 4 model. Second, you'll have a higher access limit with Gro 3. Third, you get a bigger context memory up to 128,000 tokens, so you can have longer

4:38 and more complex conversations. And fourth, there's the voice with vision feature. This one only works on mobile. iOS users already get it for free, but if you're on Android, you'll need the Super Gro plan to use it. So, for iOS users, it doesn't really make a difference. Next, let's check out the attach feature. Here you can upload a file or add text content by typing in whatever info you want Grock to process. You can even draw something and feed it in which can be super handy in certain situations. Plus, you can also connect your Google Drive and Microsoft One Drive accounts. Uploading a file is perfect when you have a long document you want to read through quickly and save time digesting

5:23 all that information. For example, let's say I upload a file, a white paper called autonomous vehicle timeline and road map ahead. This PDF is 25 pages long. So if I want to get the main ideas fast, I can just ask Grock to summarize the main points. And just like that, it gives me a quick summary of the whole 25page PDF so I can understand the key takeaways in seconds. Besides text analysis, Grock can also analyze images. Here, I'm uploading two travel photos and asking Grock to tell me where they were taken. Okay, so for the first photo, it says it's near Tokyo Station, and that's spot on. That picture was

6:09 taken right in front of Tokyo station. For the second one, it guesses Joeang or Juja Xiao in Jang Su. The actual spot was Ki Shantang, but since there are lots of similar sites in Djang Su, I'd say it's pretty close. The point here is that Grock can do more than just read text in images. It can also understand what's in the picture and analyze it. All right, now let's move on and talk about the deep search and deeper search features. Deep search lets Grock search across Xplatform posts and web pages, pull all that info together and reason through any conflicting points. Then it gives you a clear structured report. It's kind of like the deep research

6:55 feature in Chat GPT and Google Gemini. So what's the difference between deep search and deeper search? Deep search is faster. It gives you results more quickly. Deeper search on the other hand takes more time because it digs up more sources with higher quality and does a much deeper analysis. So deeper search is better for things like academic research, technical comparisons or strategy analysis. Let me show you an example. I ask Grock to pull together a summary of the major LLM AI chat bots right now along with their pros and cons. Here remember to turn on the deep search button. You'll see an interface that shows Gro's search progress and its chain of thought as it analyzes the

7:41 info. Once it's done, the results show up in this section on the right, kind of like the canvas feature in Chat GPT and Google Gemini. But one thing to note is that you can only edit by yourself here. You can't co-edit it directly with Grock yet. At the very bottom of the chat, you'll see all the X posts and web pages it pulled in. You'll also see a deeper search option. If you want Grock to dig up even more sources and do a deeper dive, just hit deeper search. It'll run through the same process. You'll see the search and analysis progress, but just keep in mind it'll take a bit longer. Okay, so after about 5 minutes, Grock finished this report. Definitely took longer than the 1 minute deep search

8:28 run. If you scroll all the way down, you'll see it pulled in a lot more sources, too. So, if you've got enough time, my tip is start with deep search to get a quick overview, then decide if you want to run deeper search for a bigger, more detailed collection and analysis. Okay, next up, let's talk about the think feature. Think is great for solving complex problems that need multi-step reasoning or super precise answers. For example, here's a financial planning question I ask. Let's say I plan to retire at 50, live until I'm 85, and I'll need about $4,000 a month for living expenses. So, considering inflation and investment returns, how much do I need to save to

9:15 retire comfortably? This kind of math problem needs multiple calculations. So, I definitely recommend using think mode for this. In the middle here, you can actually see Grock's thought process. When you use think, it does take a bit longer to get an answer. And in the end, it gives you the final number along with a clear step-by-step breakdown of how it got there. Let me show you another example. I upload some fictional sales data and ask Grock to analyze the trends and give me an insights report. Again, for this kind of data analysis, it's a good idea to turn think mode on. When it's done, it automatically generates a visual report page. The report includes

10:01 different charts, insights, recommendations, and a conclusion, just like I asked for. All right. Next, let's talk about the schedule task feature. With schedule task, you can set up a schedule for Grock to do specific things at specific times. One of Grock's biggest strengths is how deeply it's integrated with the Xplatform. So, it's especially good at keeping you up todate with real-time info. Let's check out an example. You could set up something like a daily tech digest or a custom newsletter. You just pick the time of day you want and Grock will gather all the latest news from the past 24 hours related to AI and tech developments. Not only does it pull from web pages, but it also grabs fresh X posts, so the info is

10:47 super current. You can choose whether you want to get the update by email or app notification. Then just create the task, and every day at your set time, you'll get your daily text summary. Next up is the create images feature. You can use this to generate visuals for things like marketing ads or just for fun, like making memes or other entertaining content. Here's an example. Let's say I run a restaurant and we're launching a new burger, but all I have is one plain photo of the burger. I can use Grock to turn that simple shot into a more eye-catching promotional image. By default, it'll generate two photos for you to choose from. This one looks pretty good, way better than the plain white background version I started with.

11:32 So, that's how you can enhance an existing image. Now, let's try an example of making something completely from text. I ask it to draw a cat winning an Olympic gold medal. Basically, you can use create images to come up with all sorts of fun, silly, or creative pictures. When you click on the photo you like more, you can enter a new prompt to tweak it more. And you'll also see some handy preset options to pick from. This is one thing I really like about Grock. It gives you suggestions you might not think of, so you can experiment with different styles. One extra thing to mention, Grock has way fewer restrictions when it comes to generating images. Unlike chat GPT, which sometimes blocks images for certain reasons, Grock is a lot more

12:17 flexible. That's definitely one of its strengths. Next, just like other AI chatbots, you can definitely use Grock for some basic coding, too. Here's an example. I want to make a simple loan calculator web page. It should let me input things like the loan amount, loan term, and interest rate. After hitting calculate, it should list out the monthly payments and also show a chart of monthly repayment. Since there's some math involved, I turn on think mode to get a better result. You'll see Grock's step-by-step reasoning along the way. Then it pops up a code generation block where you can see the code being built step by step. And once it's done, you can preview it right there. So here I try out the loan calculator Grock built

13:04 for me. Let's say I want to borrow 1 million, pay it off over 25 years with an interest rate of 5%. Then I can choose equal principal and interest or equal principal. Hit calculate and it lists out each month's repayment amount showing how much goes toward interest and how much pays down the principal. If you scroll down, there's also a chart it made based on what I asked for. You can even download the result. It saves as an HTML file. So, when you open that, you get a complete little web page. Well, the design is pretty basic, but it works and that's pretty cool. Next, let's click on files in the sidebar and switch over to the files page. All the files you've generated before, whether they're

13:50 documents, images, or code, will be saved here. Any files you've uploaded yourself will show up here, too. So, it basically works like a content manager. If you look further down in the sidebar, you'll see projects. The projects feature is pretty straightforward. You can group related chats, documents, code, images, and more into a single project. This makes it super easy to keep track of your progress. Plus, all your past chats and generated content stay inside the project will be the background info here. So, when you start a conversation, Grock will automatically refer back to previous content, which saves you from having to repeat the same background every time. There's also an

14:36 instructions section here. You can set up custom instructions for the project and Grock will follow those guidelines in any future chats. So, for example, if I want to do a deep dive on a specific topic, projects are perfect for keeping all my files and chats in one place, making it way easier to stay organized. All right, to wrap up, let me quickly cover a few extra little features. Over on the left sidebar, there's a history option. This lets you go back and review your previous chat sessions whenever you want. Up in the top right corner of the page, you'll see a private button. When you turn this on, your conversations won't be saved or used to train the model. And down at the bottom left, you'll find your profile icon. Click on

15:23 that to open settings where you can tweak some basic preferences. Under data controls, you can also turn off data sharing if you don't want your info being used for model training. So that's my walkthrough of how to use Grock and here to share some final thoughts. Since Grock can pull realtime info straight from the Xplatform, it has a big advantage when it comes to analyzing social buzz and tracking trends. Also, compared to chat GPT, I feel like a lot of the small design details in Gro's interface just make it feel smoother and more intuitive to use. And one more thing I really appreciate, most of Grock's features, like deeper search, projects, and tasks still have pretty generous limits for free users. So, it's

16:10 actually really open-handed in that sense. All right, that's it for today's intro. If there's any part of Grock you'd like me to dive deeper into, just leave a comment below and let me know. By the way, XAI is planning to roll out updates for Grock's coding model, multimodal agent, and video generation features in August, September, and October. So stay tuned. I think there's a lot to look forward to. If you found this video helpful, don't forget to subscribe, like, and hit that bell icon so YouTube can recommend more useful videos to you in the future. See you next time. Bye.