Transcription
If you're still using websites like Google AntiGravity or Cursor to get your code written, you're making a big mistake. Should I tell you why?
Just look at this. I'm on the Pro plan. Even within the Pro plan, I got it for free from Jio. Everyone got it for free for a year. The ₹2,000 monthly plan is completely free. But I get my code written through it. Still, its limit, that is, it runs out within 7 to 8 days. So, for those who are completely free, don't even mention coding.
After switching to Google AntiGravity, free users can code there. As you can see, the Kinai 3.1 Pro High limit is reached, and the 3.1 Pro Low limit is also reached. The Cloth Sonet OPS limit for all three is reached. Only J3 Flash is working right now. You can use models of level like Jini and Cloth Sonet for free. Use as much as you want 24 hours a day, 7 days a week, brother. Your limit will never end.
So I will tell you how to do all the settings. As you can see, this is Google Antigravity. Apart from this, there is a brother of this which is called VS Code. Right? Visual Studio. If you go to the chat section inside it, you will see below that there is a chat section here. We can get the code written from here also. This is also an ID. Right? And here you can see that if you click on Model Auto, it's asking for an upgrade here too.
You might think, but here we're going to create our own AI agent, where we'll do unlimited coding. So how do we create it? First, go to your browser. After going into the browser, you'll need to download VS Code. Here, search for VS Code and type in "Download." This is from Microsoft. Click on the link above. If you have Windows, click here to download it. If you have a Mac, click here to download it.
What do you need to do first? Go to a website called Agent Rater. You'll find a link in the description below this video. After clicking on that link, you'll sign up and receive $100 as a sign-up bonus. After that, you'll receive $25 per daily login. So, you'll see step-by-step how to use it. As soon as you click on the link, you'll sign up. I'll explain it to you.
Entering your email address and password here won't allow you to sign up. So, what do you need to do? For example, in the description below, there's a link for signing up. Clicking on it will open a page like this. Click on the "Sign Up" option below. Don't enter your email address there. First, create an account on GitHub and click "Continue to GH" with the same Gmail ID. You can connect there. Enter your email address and password along with your GitHub account and you'll be signed up.
After signing up, look up at the top and you'll see four options. Here, look at what we're going to setup in Visual Studio. Here, we're going to create our own AI agent. How do we create that? For example, this is my console account. That means the agent router dashboard is already there. So, look here, you'll get a $100 signup bonus and a $25 daily login bonus. Okay? Look, it's $25. $25 means it's never going to run out. Let me tell you my experience: after creating the API, I used it continuously for about 12 to 14 hours daily. Even then, your daily earnings will never reach $10. That means your daily earnings are more than enough for $25.
Here's how to create the API, and after creating it, how to create an AI agent in Visual Studio. Now, if you open this website, see what models you get. Click on "Printing" and you'll find five top-class models. You only need to select two models from these. I'm telling you. I've used it a lot. I've used each model extensively, 1, 2, 3, 4, 5. And the best models are the Deep Seek R10528 and the Deep Seek V3.2, which is listed third. Why are these two the best? Let me show you something.
Look, I went to the GBT chat and asked, "Here's the Deep Seek R10528." Or is it Deep Seek V3.2? Which model of JNI and Cloud is it equivalent to? So here he said, "The R1 model of DPC, which is called the Thinking Model, is close to Cloud Sonnet 3.5, and if we talk about JNI, it's at the level of JNI 2.5 Pro." As I told you, the Deep Seek [Music] V3.2 model of Deep Seek is equivalent to Cloud Sonnet 4.5, and if we talk about Cloud OPS 4K and JNI, it's equivalent to JNI 3.0 Pro. This chat GBT isn't telling us himself. "This is a benchmark rating, and here the public has compared across models, meaning the public has done the comparison themselves. They've rated these models accordingly, and Chat JNI has told us."
And what's the benefit of getting the code written in Visual Studio? You got some code written. This guy will write the code for you and create all the files. It will also push it to Git Hub. It will also fix all these errors. You don't need to mention much. But most users don't know its settings, and you can watch any video on Agent Rater on YouTube. First of all, you won't find any Agent Rater videos easily, that is, the settings. What happens after that is that you won't be properly guided on how to set it up. Watch this video very closely.
Here, we open Agent Rater. Here, I will open the console and click on the console. First, we will select the API key. Here, I will click on API Token. After clicking, look here, I will delete this API key once. I deleted it. As soon as you open it, you will see this interface. So, to create, click on Create Token. Here, you can give the API key any name. For example, I can write "Test Key." You can give it any name. It doesn't have much meaning. You just need to enter the name here. Don't touch anything else. After entering the date here, you can enter any date, be it 2040, 45, or 25. I've selected 2070, meaning it will never expire. Keep the new quantity the same. For quota, enable the unlimited quota below. This doesn't mean you're getting the code written and if the quota runs out, there's a problem. Okay? Enable this. After that, you don't need to touch anything below or click on any option. Simply click on submit.
As soon as you click on submit, you'll see your API. Now, you'll see which models can be used with this API. Click on printing. You have a green tick on the available models, as you can see. More models are added here. For example, cloud models are sometimes added. But the chances are very slim. But these two models, R1 Deepic and Deepic 3.2, are the best models. Okay?
Let's go back to the console. Click on the token here. What will we do next? Go into Visual Studio. We're going to set it up. Look, there's only one chat option here. There are no other windows here. The fifth option will be extensions. Click on extensions. Here, you have to search for "ru code r double o r c o d e." You just have to type that much. And Ru Code, see which one you want to download. The one that appears at the top says "ru code 1.3m." Currently, it has so many million downloads. Click on it and install it. It's a very easy setup. It will install. And as it installs, you can see that along with chat, this option is also enabled for Rub. Okay? Let's close this window now. Then we'll set it up. This was our chat option. It has no use for us. We'll delete this. We'll chat here. Click on Ru Code. We'll do all the chatting from here.
And when this window opens, what do you have to do? Look carefully at these settings. This is a very important setting. Okay? And what do you have to do here? Look down, you'll see some options. Okay? Here, you'll see options like Code, Agent, Debug, or Ask. Along with this, there's an option in the center. You have to click on it. And after clicking on it, you have to go to the Settings icon. Click on it, and there's a plus icon. Click on it, and a New Profile option will appear. So, I'll write Agent here. It's not necessary to write Agent. I'm writing Agent because it's a coding agent, so I've written Agent Name. So, here I'll write V 3.2. 3.2 because in the profile I am creating, I will add the Deep Seek model here. That's why I have written 3.2 here. So click on Create Profile and here at the bottom you will see which is the API provider? Here you have to add the Open AI compatible one. You have to select this here. You should not use Open Rater here. You should not use Open AI. Here you have to use the compatible one. Open AI Compatible. If you click on it, then you have to add this base URL here. It will not be visible to you by default. I had already added it, that is why it is visible. And look here, there is an option for API Key. And where will we get the API Key from? We will go to the website of our agent Rater, where we had generated the API Key. Click here, copy it, and come here, just below the base URL, where it says "API." Paste it here. Where it says "Model," we'll add the model here. Now, where can we get that model from Deep Seek? Go to Pricing, and the models will appear. Okay, click here, and you'll see the models. I'll click on "Deep Seek 3.2," which is equivalent to the 3.0 Pro version of JavaScript. Now, let's go back to Visual Studio. I'll paste it in "Search," and I'll click on "Model," and "Use Custom," and it will automatically be added.
After this, you need to carefully examine the settings below. Okay? Because here's "Maximum Output Tokens," and here's "Context Window Size." You should set the "Maximum Output Tokens" to 8,000, not 8,000. Keep this in mind. And what context window size should you set here? 90000 90000 90000 is its context window size. You should not keep it more than this because if you keep it more than this here. Agent Rater's API will show you error 50404 here. Okay?
Now let's go down here to advanced settings. So everything is already fine here. And after this let's enlarge the window a little here so that we can see all the options. After this we will go to modes. If we click on modes then this is absolutely fine here. It is set to coding. So let's leave the code only. We will come below it here. See, by default print will be written. So you should not touch it here at all. This is absolutely fine. You have to leave it as it is. So we will come at the very last. Look here, this custom instruction, you'll see it blank. The custom instruction will appear empty, meaning it will appear at the very end of the model. After coming to the last one, there's the custom instruction for all modes. What you need to do here is add a custom instruction. If you don't add this, it will show an error because it's very important to add this custom instruction. Otherwise, this setting will remain incomplete.
After that, let's look at more settings. Once we scroll down, it's completely fine. This is optional. There's no need to add anything. We'll go to Skills. Leave this as it is. Command Auto Proof. Okay, look here, all these switches may be off, or all these buttons may be on. Okay? What do you need to do? Leave the question button disabled. Enable the remaining 1, 2, 3, 4, 5, and 6 buttons. That is, these buttons should be blue, and the question button should be grayed out. Okay? After that, we'll go down a bit. Scroll down. Read and Write Include File Outside of Work Allow root to read file outside current. So, don't touch this at all. That's fine. Leave it as is. After that, MCP. If it's disabled, enable it. Okay? After that, don't touch Check Point Context either. Terminal Used Inline Terminal This is also fine. Prop Workspace UI Experimental So, that's all the setup.
Now we'll test our A once. Okay? What will we do to test it now? Click on Save once. This is very important. Click on the Save button at the top. Now let's see if our A is responding or not. So, we'll test it here. What will we do before testing it? Let's go to our console. Let's go to Usage Lock. And now we'll test this model 3.2 here and see if it's responding or not. I can type any message here. I'll just say "Hi." So, look, our app is working fine, and the lock will appear here. I'll refresh it now. So look here, our test key, which we just named "Test Key," is responding well. So, you can get it to write as much code as you want all day long. Now you don't need to use Google AntiGravity. So, we have a perfect setup.
Now I'm going to show you how to get started if you want to get it to write code. Go to File. Click on "Open Folder." And if you click here, you can create any folder inside the admin folder. For example, inside the admin folder, I created a folder called "Visual Studio Pilot." Open it. After opening it, you can create any folder here. For example, I created "TT Toy Game" here. I created it. I selected it. What will it do now? Now, whatever code we get it to create, it will create our files inside that folder. We've created a folder; it will work within that folder. And our model is already correctly selected below. So, if I paste it here and run it, as soon as it starts writing code, our file will be created within that folder. I'll show it to you right now.
Look, it's created the game. And if we go to the file it created, this one, we created a folder for the Ti Tech Toy game in Visual Studio, and this is the game it created. I'll double-click it and see if it's created the game. So, look at its UI design, just by typing a single line, it's created a complete game. Try it out. Now, this guy doesn't just create small games; he can also create large websites. I'll show you this website. I've also set up payments on this website. Okay? Click on it and the payment page opens directly. And the magic of this website is that if you just enter a title, you'll get a thumbnail. I added the title here. I just added the title. You can see it. And here I'll click on "Generate Thumbnail." This will create a stunning thumbnail within 5 seconds. So, look here, I just added a title, and using that title, I created an app using AI. So, I created an Android app using AI. Okay? Whatever was under the title, he added it. So, look here, it's for both Android and iOS, and both Android and iOS icons have been added. Okay? So, you can also get a website created like this. Your prompt just needs to be good.
So, like and subscribe to this channel, brother. Okay? If you like the video,