Google Gemini CLI is an INSANE AI coding tool (full tutorial)

Google just released what might be the most insane AI coding tool ever built. But what if I told you it's not because of how powerful it is. It's because of how much they charge you for it. In this video, I'll walk you through what Gemini CLI is, how it works, how you can build your first app with it, and why it might be the only AI coding tool I use moving forward. Let's get into it. So, Google

quietly released today, just in the blog post I'm showing you here, a brand new AI coding tool completely powered by Gemini 2. 5 Pro, which many people consider the most powerful AI coding model out there, uh, called Gemini CLI. Now, what is Gemini CLI? Gemini CLI is Google's answer to Claude Code. For those who haven't been paying attention, claude code has been getting insane amounts of hype over the last few weeks, and for good reason. It is probably the best AI coding tool I've used in my entire life. But Google's trying to change that. So, Gemini CLI, just like you see here, is basically the same exact thing as Claude Code. It runs in your terminal. You don't need to download another program for it. a whole development environment. You just open your terminal and you can run it and start writing code with AI. But here's where the cool part comes in. Here's what makes Gemini CLI completely different than Claude Code is the pricing model. They just give you a thousand requests a day for free. For those who don't know, Claude Code is insanely expensive. Yes, you can get it on a $17 per month plan, but you're not getting many requests out. You got that's basically enough requests for like two hours. So, if you really want to use Claude Code, you need to move up to these two tiers. And if you're a serious coder and you're looking to build an actual app, you pretty much have to do the $200 a month version. So, very, very pricey. Although, again, the most powerful tool out there. So, in my opinion, was worth it. But Gemini CLI, that's different. This is a different pricing model. Google's saying, "Hey, not only can we match you on your product, but we're going to beat the hell out of you with our pricing model. " And that's what they did here. As you can see here, you get 1,000 requests per day for free from Gemini Pro 2. 5. That is pretty amazing. Gemini 2. 5 Pro is very, very good. And the fact you get 1,000 requests per day free is basically Google just saying, "Hey, we got billions of dollars. We're worth trillions of dollars and we don't need to nickel and dime you for a couple hundred dollars a month. Just have it for free. Just have at it. We want you on our model. " Now, why would Google do that? Well, they're probably doing that for a few reasons, right? One, they want to get you using their product, right? They want to move you over from Claude Code because everyone's been on Claude Code. So, hey, they're willing to invest a few dollars to get you to use their product. Two, it probably helps them with training their model, right? The more data they get in their model, the more usage of their model, the better their model gets. So, if they can get a whole bunch of people to come over and start using Gemini CLI, their model will just improve that much faster. And three, they're doing it because they can afford it, right? They have tons of money. They can afford it. They can afford to just give things away for free. That's the strength of Google, right? that they'll have over these smaller companies like Anthropic, like OpenAI, is they have an absolute war chest of money they can throw around and make sure they win the AI war. So, that's absolutely massive, free of

charge. Let's get into using it. I'm going to show you how to set it up, how to use it, and we'll see just how powerful it is. Does it match up to Claude Code? We're about to find out. If you stick here, you're going to be a master of Gemini CLI. I promise you that. So, what I'm going to do is I'm going to open up cursor and we're going to run this in cursor. So again, you don't need You can just run this in your terminal. But I like to run this in cursor because then you can combine it with your other tools you have in cursor like the background agents for instance. So if you want to follow along with me, feel free to pause here, download cursor. I'm sure you probably already have cursor. Download it. Get the free tier. Doesn't really matter. Open that up. Then hit control tillday to open up your terminal you see here. And I'm about to show you how to install Gemini. So to install Gemini CLI, pretty simple. You want to copy and paste this command here. I'll put this down below. So feel free to pause, grab this from down below. You take that, you go on your terminal, you just paste that in, you hit enter, Gemini CLI installs. Then all you have to do once you install it is type in Gemini, hit enter, and Gemini is good to go. You're ready to cook. This is what you get. You get the screen right here with Gemini. Now we're ready to build some cool things. So here's what we're going to do. Let's start building out an app, and let's see how well this performs. I'm going to show you exactly how to interact with Gemini CLI so you can get the most out of this. So there is no planning mode with Gemini CLI. That's a downside here. Cloud code has that. In Claude Code, you can just hit shift tab and you go into plan mode where you just go back and forth with the model to build out a plan for your app. That doesn't exist here. So if we want to plan here, what we're going to want to do is basically just say, "Hey, don't code. Build us a plan first. " Which isn't the biggest deal in the world. So here's the initial prompt I'm going to use. And I'll put this down below as well. Let's do some planning. Right? So I'm making sure it knows that first. I don't want you to write any code. Let's do

some planning. I want you to build an AI habit tracking app. I enter in a few sentences about myself. Then an AI generates daily habits I should perform to better my life. I can then check off those habits and the app shows a visualization of my habits and my streaks. I want to build this with Nex. js. So, a few things we're doing here. We're telling it we want to do planning. We're describing the app we want to build. So, in this video, we're going to build an AI habit tracking app where it generates habits for us to perform. I want to get this a little bit more complex, right? So that we can see just how powerful this is and how much it can build out. And I'm also saying I want to build this with Nex. js, one of the most popular uh web frameworks out there for building apps. I'm also going to say do not write any code just so it knows we're in planning mode. So I'm going to hit enter. Let's see how this performs. So it's deconstructing my query. So the so Gemini is going in understanding what I'm requesting. Now let's see how good this plan is. Okay, looks like the plan's done. That was rather quick. I would say is honestly a little bit quicker than claude code. Of course, building a clear plan is the most important first step. Here's a highle plan. All right, let's see the plan. Core concept and visualization. The application will serve as a smart coach. Okay, technology stack. We're going to use Nex. js. It's going to have APIs. They're going to do styling with Chad CN. That's good. AI integration, we're going to use an LLM. Doesn't tell me what LLM we're going to use, but that's fine. We'll use Postgress. That's cool. And visualizations. Okay, so we're going to have some charting visualizations. It's going to have an entire onboarding, habit tracking, generation, daily habit dashboard. Okay, so this is actually a pretty good plan and it has an entire road map for us as well. So it's going to do it in four phases. Now, if you want to go back in here and edit this plan, you can do that. So you can just say, "Hey, I don't like this or change this. " You can use this time to make sure the plan is perfect. I actually really, really like this plan. This is a very wellthoughtout, detailed plan. Again, Gemini 2. 5 Pro is a very good model. So let's hit enter for this. I can't believe all this is free. They're just giving me all this for free. Shout out Google. Thanks for just giving stuff away for free. Pretty amazing. Here's where the first test is. Is okay. How much can it actually start building out? One of my biggest issues with Gemini 2. 5 Pro and why I stuck with Claude is it has it just for me it had a lot of issues calling tools where it would try to write files and then it would mess up and not actually do that. Let's see how good it is actually using its tools and building out files and all that. Okay, so it looks like it has the same kind of UI as cloud code as well where I can just have it go on YOLO mode. I'm going to have it go in YOLO mode where it just

builds out everything for me and we'll see how well it performs. So it's installing all the technology. It's doing all the project setup. So it's determining what the file structure should be, what technology should install. It's going through all that here. All right, looks like it built out the next. js app. So it has the kind of boilerplate code ready to go. So far, honestly does feel a little bit faster than cloud code. No issues with calling tools just yet either. All right. Looks like the app finished building this phase one. So, let's see what phase one of this looks like here. All right. So, we have a pretty general UI. Tell us about yourself. Generate habits. This probably is not going to work cuz I didn't enter in any sort of APIs or anything. Although, I wonder if they could automatically just implement Gemini API into this or something like that. And then they have some boilerplate to-dos. You can't even check it off. Okay, that's solid. Listen, just phase one. We're building out the UI. The UI is kind of ugly. we can make it better. Let me show you how to communicate with Gemini CLI so we can build and improve our apps that we're building here. Okay, so I just did this. I asked Gemini CLI what phase 2 is, right? We had that really long plan. Now I'm just going back, hey, what's phase two? Let's see what it built out. What's different about Gemini CLI from like say cursor for instance is when you're working with cursor, it's very much like an engineering intern. You have to guide it, handhold it, show it what to do. But with claude code, with Gemini CLI, what makes it really, really amazing, it's like a senior developer. It's guiding you. I'm just having it guide me. I'm just saying, "Hey, continue to the next thing. Can you can trust it to build out its on its own. " So, I'm just saying, "Okay, what's phase two? " It's telling me it's going to build out the API routes so I can communicate with the AI to implement AI functionality. It's going to build out AI prompts so that it can generate the habits for us. And it's going to connect the front end to the back end. I'm going say, "Okay, let's move on to phase two. " All I'm saying is, hey, continue with your plan. That's how I'd recommend working with Gemini CLI is you have it upfront. Build a really detailed plan. Spend a good amount of time here and then once it has a really in-depth detailed plan, you're going to save a lot of time after because you can just keep going, okay, continue, continue, continue because it had such a good plan. If you spend more time upfront with these AI tools, you'll get a lot better results and save a lot more time on the back end, right? Because before when we were using cursor it was like okay do this next do this and you're handh holding each step of the way. Now because we did good planning with Gemini CLI I can say continue and just like that as I was talking this completes phase two that took about what 20 seconds and now phase two is done. Let's see what it did here. All right let's see what we got for phase two here. All right so now we have

just the input which is what we wanted right we didn't want fake data in there for the habits. So let me tell them a little bit about myself. I am a content creator that makes YouTube videos and vibe codes apps. All right, generate habits. Let's see if it works here. Okay, so it generates uh review daily SC. Okay, so it has uh it has template habits in there that show up. The AI is a little bit nicer. You can check things off now. Okay, I think for the next step, I would imagine the next step is going to be it uses AI. So let's see that. So again, as I keep telling you, you know, you built your plan, now you do it phase by phase. I just ask what phase three is. And now it's going to tell me what the phase three is in that plan. And then we're just going to okay and have it continue. Phase three is about making the application persistent. Right now, if you generate habits and refresh the page, they disappear. Habit 3 will introduce a database to save all the data. Okay, looks like it continued. Uh the next step it did was update the API route to call open AI. Okay. So, all we need to do is put in the Open AI uh API key, which I'll show you how to do in a second. And then once it does that, it should be generating habits for us, which is pretty cool. So, let's get the OpenAI API key. I'll show you how to do that now if you're building this alongside me. So, if you go to platform. openai. com, then sign up there. Really straightforward and simple to do. Then click API keys, you can create a new API key that we can use for the AI. And basically, the way this works, it just allows you to communicate with chat GPT. So you can generate AI content, which is awesome. So I click create new secret key up here in the top right. And now we can name this habit tracker. And then click create the secret key. And that'll give you the key that we'll copy and paste in a second. Just as a side note, if you're going to build and launch an app, do not put your API keys uh in anything but aenv file. So this app, the Gemini CLI is having us put it in an actual like customer clientfacing code file, which is not good. But for the sake of this video, we'll do it. But if you ever to ship an app live, do not put it in like your code files. Put it in av file, which makes it private. So it's having me put the API key here, which I'll do that now. All right. So I refreshed the page. Let's see this. Let's see if this actually generates AI habits for us. I am a YouTuber who creates videos. All right. Looks like it is doing some work here. Let's see what it comes up with. Okay, let's see. Okay, it came up with some habits. Create a content calendar for your channel. Record and edit at least one video script per day. Engage with your audience by responding to comments and messages daily. Research and learn about new AI coding trends regularly. That's pretty amazing. It worked. The AI worked here. I've had basically zero issues. It's worked very quickly and I can check these off. It's pretty amazing, right? We did this in just a few minutes through Google CLI. Let's recap what we've learned so far and how to use Gemini CLI most effectively. So, it is basically claude code. It performs very much like claude code from what I see. Performs very well. If anything, it performs a little faster. The most amazing part of this though is the pricing model. You get a,000 requests a day for free. Two, how to use Gemini CLI the most effectively. Well, you go in, it doesn't have a planning mode. So, you want to say, "Let's build a plan. Do not write code. " Then, you give it a very detailed description of what you're building. And then, it's able to give you a great plan. So you go through the plan, make sure that looks good based on what I'm seeing. It is very, very smart. You give that the go-ahad and then all you do is just say, "Okay, build this one step at a time. " Right? time. So here is how you get the most out of Gemini CLI, right? You go in, you

have it build out a plan for you first. So go say, "Hey, since it doesn't have a plan mode, you say, "Let's make a plan. Let's not write any code. Here is my idea of what I want to build. " You give it a detailed description of what you want to build. It then builds out your plan. You go through that plan step by step. Just make sure it looks good. If you need to edit anything, just say, "Hey, change this. " Once that looks good, you give the keys over to Gemini CLI. You say, "Hey, build this out for me. " Step by step. Each step, you go in, you test it out, you make sure it works. Then you give it the go-ahhead to move on to the next step. If you do this very carefully and if you use a co-pilot LLM, like for me, I'm going to probably use Claude as my co-pilot here and I'll go in and ask questions along the way around planning. You will probably have an amazing app at the end of it. I showed you what I did here. We built an AI powered habit tracker, which you probably know this is going to be a real app in the next couple months that a lot of people use as an AI generated habit tracker. I'm really liking Gemini CLI. If I'm going to be honest you, I need to do a little bit more testing against Claude Code to see which one's smarter, which one's better. I'll probably do a deep dive comparison between the two that I'll put out in the next couple days. Make sure to subscribe and turn on notifications for that. But based on what I'm seeing, I think I could probably downgrade my Clawude subscription from $200 a month probably to $100 a month and then use all my free calls with Gemini as I start coding out with it. And then once I get through the thousand requests, if I get through that, then switch over to Claude Code. But if I don't get through those thousand requests, I probably can just turn off Claude Code completely. But we'll see after I do the full Gemini verse Claude comparison. By the way, if you want to get started with Claude Code, I have a tutorial video on that that's going viral right now. That's down below. Make sure to click on that. If you learned anything at all, leave a like. I will see you very soon with a deep dive Gemini CLI versus Claude Code video. See you then.