Claude Code is Getting WORSE — Here's How I Made It 10x Better

Added: 2026-05-06

[00:00:00]Cloud code is powerful, but when you combine it with ChatGPT 5.5 and Gemini, you unlock the capabilities that don't exist in any model, and it's completely free. Here's the thing nobody wants to say out loud. Cloud code just got worse, and Anthropic confirmed it. But instead of jumping out of it, I found a solution for this. In this video, I'm going to show you how to add Cloud Code, Gemini, and the ChatGPT inside your Antigravity Terminal. Plus a free skill that automatically selects tasks for each model so you make the best results out of them. Let me just show you how to do it. So, basically, one terminal, three brains. Combine Cloud Code with ChatGPT 4.5 and Gemini 3.1 inside Antigravity.

[00:00:41]Each model handles what it's best at at zero extra costs. So, here's how it's going to work. We're going to connect Cloud Code, we're going to connect ChatGPT, and we're going to connect Gemini into Antigravity. And the best thing about it, it's three AI models in single place, no extra cost, the token context is over 1 million, and the setup time is less than 5 minutes. So, three models, three superpowers. Each model excels where others fell short, so together they cover blind spot. Cloud Code is the builder. It's best at coding, creating apps, and components, and it's also able to handle the heavy lifting. Now, moving on to ChatGPT, it's going to be the reviewer, your code reviewer and your second opinion. So, Codex catches bugs, spots hallucinations, and reviews architecture decisions. So, when Cloud Code gets stuck in a loop, GPT breaks it out with fresh eyes. And moving on to the Gemini, it's going to be the analyst, your multimedia specialist. With 1 million token context window, the Gemini handles video analysis frame by frame, processes audio, and reads 200-page PDF that would choke other models. Now, let's just compare the models. So, first is the Cloud Code. The Cloud Code is best at code generation, but if we look at the code review, as you can see, it's decent. It's not the best at it, but overall, moving on to the ChatGPT, it's the best at code review. That's why we're going to use the ChatGPT for code reviewing. And moving on to video analysis, as you may know, Cloud Code cannot analyze videos, and same goes with ChatGPT, but the Gemini is good at it, and it's best at it. So, that's why we're going to use this for the Gemini.

[00:02:12]And audio processing is the same, we're going to use the Gemini for this. And on the context window, as you can see, the Gemini has over 1 million tokens, while other models has less than 200,000.

[00:02:25]And moving on to long documents, also Gemini's best at it. And on the PDF analysis, also the Gemini's best. And the last thing is agentic coding, which Cloud Code excels at. So, if you take a look at this comparison, you can actually see that no there's no perfect model. Every model has its flaws, but that's not bad, because we're going to fix this by combining all three inside one single terminal, so we can use all three of those with no extra costs, and the results are going to be better than ever. So, here's how the three-brain router works.

[00:02:57]So, one skill that automatically tags the right model for every task. You're going to drop it into Antigravity, and Cloud Code handles the routing. So, so this is the flow. First, we insert the prompt, then it's going to be auto-routed to the different models. So, for example, if it's coding, it's going to be sent to Cloud Code. If it's bugs, it's going to be sent to the Codex. And if it's a analysis task, video, PDF, audio, for example, it's going to be sent to the Gemini. So, all of those three will be used to their own power.

[00:03:26]Now, on the cost breakdown, the three models at minimal cost. So, there's a high chance you already have at least two or one subscription. So, for example, if you're using the Cloud Code, then you're going to save money for these two. If you're using the ChatGPT, you're going to save money from the ChatGPT. So, in total, if you are already using these models, so for example, I have the Cloud Code and the Gemini. So, the extra cost is just 20 bucks, because I need to get ChatGPT Plus, but it's not mandatory. You can actually have the free version, because it's not going to be your main horse, it's going to be like just on the side, so it's really up to you. But if you have the money, you can purchase all of those three subscriptions. So, $60 at most per month, and you get the capabilities of every model, and you don't get logged out of your session for using too much credit so fast. So, let's just get started. So, I want to make your life easier, that's why I have my free community. The link is in this video description. And if I go on my free community, I go to classrooms, I go to the vault, and I can see everything that's needed for this video. So, it's the three-model brain. So, here's what we're going to do. We're going to use all three models, as you can see right here, the Cloud Code, Codex, and Gemini.

[00:04:38]And there's the part one. You have to open up your Antigravity, create a new folder. And it's kind of hard to do stuff in Antigravity, but trust me in this one, I'm going to do it slow, so you will understand everything. Now, if you already don't have it, you can see on the extensions on the extension tab on the left corner, we have Cloud Code for VS Code. So, just click on there and download it. So, once you have it, you will just need to open it up. And as you can see, you have this option, the web or the local. So, select on local, and click on new session. And now you have a Cloud Code inside Antigravity. How cool is that? So, when you have it, there's this little button right here. You click on it, and you have to find the terminal. So, that's super easy, just scroll, open Cloud in terminal. That's it. Boom, we have Cloud in terminal. And once we have it, now we're going to go back to the free community. So, we're going to start off by copying this Codex NPM code. So, I'm just going to copy it real quick. And let's head to Antigravity.

[00:05:40]And inside here, let's just paste this code. So, let's just click on run. And as you can see, Cloud Code is building it for us.

[00:05:49]So, it's connecting the OpenAI API key.

[00:05:54]So, do you want to proceed?

[00:05:56]Yes. So, we want to proceed.

[00:06:00]And now it's adding all of the packages, and it's installed successful. Now, we need to install the Google Gemini.

[00:06:09]So, let's go to Antigravity. Let's paste it in.

[00:06:13]And let's click on run.

[00:06:17]So, it should be fast. Do you want to proceed? Yes. So, now everything is installed successfully. And if you want to check it, I have both of these. So, for example, if you want to check Codex, if it's successfully installed, you can head back to Antigravity, paste this thing in. So, Codex version.

[00:06:37]And it should be able to give you an answer. So, yes, and don't ask again.

[00:06:45]Let's take a look at it. And as you can see, everything is working and installed. So, I got this result. Now, the logging in part. So, we have to do this thing. So, let's just copy the Codex login.

[00:07:01]Now, head back to Antigravity again.

[00:07:03]Let's just paste this quickly in.

[00:07:06]And it's pretty much that. You just have to go to this cool community. You just have to copy these and paste them in.

[00:07:16]Now, as you can see, we have to sign up with the Google. So, I'm just going to do that real quick.

[00:07:25]It should not take long.

[00:07:27]So, now it's done. I may close this page. Now it's logged in. Now we have to log in to the Gemini. So, we just have to copy the Gemini. That's it. Let's get back again, and just simply paste that.

[00:07:40]So, now you will be able to connect to the Google Gemini right here, because we are using the Cloud Code, as you can see.

[00:07:49]So, do you want to proceed?

[00:07:55]And let's go.

[00:07:57]That's it, it's pretty simple. We can sign up with Google. So, let's just do that.

[00:08:03]Yes. So, now we sign up to the Google for the Gemini, because we're using the Cloud Code inside the Antigravity, and it's successful. We can close it now. That's it. So, as you can see, everything is successful.

[00:08:20]And now, as you can see, it gives the instructions. So, press Alt to restart.

[00:08:24]Okay, so let's press So, we just restarted it. So, and you just have to do all of this for all of those. So, this is the one step, step two, step three. So, just do all of those things. You just have to install these plugins. And when we come to the next part, you can do a test. So, for example, if you type {slash} Codex setup, you should get true. So, it goes same for the Gemini, say hello, and it should be greeting. If both pass, you are fully wired up. And now, once it's finished, you install all of these inside the Cloud Code in your Antigravity. What you need to do once it's installed. So, I'm just going to give you a skill that you can place inside the Cloud Code. So, if you scroll down, you can see a skill. So, what we're going to do, we're going to copy all of this text. It's pretty simple.

[00:09:22]Okay, so just like this. Let's just click on copy. Now we get back to the Antigravity.

[00:09:31]We ask law, inject this skill.

[00:09:37]And we're just going to paste it in.

[00:09:40]That's it.

[00:09:41]So, it's going to inject this skill.

[00:09:43]Let's make the terminal smaller. So, that's it, it's installed. And as you can see, both of them are present and up to date. The skill is installed at this and dot MD, and it will be available at free brain globally across all of your projects. So, as you can see, what's live, the codex, and the Gemini both above minimum versions. So, here's how we're going to test it. We can either upload a video and see how does it do.

[00:10:07]So, for example, if I paste this YouTube link, it's one of my videos, and I'm just going to ask, "Analyze this video."

[00:10:16]That's it. So, I'm just going to give it to the cloud, and let's see what it's capable of with that skill. And I just really forgot to mention, so we just have to insert the free brain skill and just paste that now. So, analyze this video. So, now it will understand what you are thinking about. You want to use this skill, and let's wait just for the results.

[00:10:39]What it actually will do, it will use all of the models. So, the Gemini, the cloud code, and also the chat GPT to do your task. So, it's going to be way more powerful than ever before. So, imagine how cool is that. Gemini analyzed the video, and here's the breakdown. And funny thing, we are inside cloud code, but we are also inside the anti-gravity.

[00:11:00]So, we are using all of the models, and as you can see, the cloud skill actually decided which model to use for the analysis, as we expected. It's going to use Gemini for the analysis because we wrote that in the skill, and here's the breakdown. So, the video was hired to build and sell these websites, and we even got the timestamp timestamp breakdowns. We got the core claims. And when I will we can do with this, for example, we just analyze something with the Gemini, but now we want to build something. So, if we ask the cloud to build something, it will actually use the cloud now, and the codex will be the one for reviewing the script and seeing if everything is all right, and fix any issues if cloud code glitches somewhere or gets in a loophole. So, this is like just new era of AI tools because you can now have all of them inside one single place, and they are able to do different tasks which which they excel at. So, I think this is pretty insane because the capabilities are unstoppable. So, for example, if you have a very hard task, you have to analyze something, then you have to build it, and you don't want to make any mistakes, so every tool comes in handy because Gemini analyzes, cloud builds it, and codex checks it. So, you have no mistakes left, and the result is going to be better than ever. So, if you are really serious, I have a new course coming up in my paid group, so feel free to check it all out, and see you in the next video.