Run Local Uncensored AI from USB - Windows, Mac & Linux (No Internet) 🔥

Hinzugefügt: 2026-05-05

[00:00:00]This tiny USB drive is running a fully uncensored AI. No internet, no account, no filters. And this setup runs on everything. Linux, Windows, Mac, even Android. And here's the best part. Every conversation, every message, every answer saves directly on this USB, not on your computer.

[00:00:20]You unplug the drive, it's gone from the machine like it never happened.

[00:00:25]Now watch this. On Linux, I just ran this one install script and that's it.

[00:00:30]Uncensored AI ready to go. On Mac, I ran start.command file and done. Same chat interface, same AI already there.

[00:00:40]And on Windows, I double-click the start file, the AI just opens up in the browser automatically. And look, all the chats from Mac and Linux are already sitting right there.

[00:00:51]Everything is on the USB. And yes, this setup able run Gemma 4, the newest model from Google. Now let's actually test if this thing is really uncensored.

[00:01:01]So I asked the same question to ChatGPT first.

[00:01:05]And yeah, look at this.

[00:01:07]It just goes around the question, but no clear answer.

[00:01:11]Now I'll ask the same thing to our local AI. And there you go. It just answers it directly. No warnings, no lectures, just a straight response. And I tried a few more questions, too. Just pause and read this yourself.

[00:01:27]So in today's video, I'm going to show you exactly how to install and run an uncensored AI model locally from a USB drive in Linux, Windows, and Mac. So let's get started. All right. Now let's move to the installation. I'll show this for Windows, Linux, and Mac. Let's start with Windows, but Linux and Mac users should also watch this part because the basic setup and files are exactly the same. Only a few steps change later.

[00:01:53]First, plug in your USB drive.

[00:01:56]If you want to use it on Windows, Linux, and Mac, it's better to format it as exFAT since it works on all three. But if you want to continue with your current setup, that's fine, too. No need to format. Next, open the GitHub repo link in the description. This is where all the setup files are. Before you download anything, I recommend just quickly reading through the repo to understand how this setup works.

[00:02:19]Also, it's fully open source, so you can review the code yourself or verify its security before running it on your system.

[00:02:26]Now just go to the code section and click on download zip to get the USB local AI setup files.

[00:02:33]Once it's downloaded, move the zip file to your USB drive. If you're not using a USB, just move it to any drive or folder on your system and follow the same steps.

[00:02:44]After that, extract it there, and you can delete the zip file once extraction is done.

[00:02:49]Now inside the folder, you should see Windows, Linux, Mac, and a shared folder. Let's open the Windows folder.

[00:02:57]Here you'll find install.bat.

[00:02:59]Just double-click it to run.

[00:03:02]Now just press any key to start the installation. All right. First, it will ask you which model you want to download. You can choose any, and you can even download multiple models if you want. But I recommend starting with Gemma 2 since it's lightweight and works well on most PCs. Now the download will start. It will first download the model and then set up the AI engine after that. This step can take some time depending on your internet speed and USB speed. So just let it run and wait.

[00:03:33]All right. The engine is downloaded and the model is imported. Everything is now on the USB drive. Next, we just need to start the chat.

[00:03:41]Go back to the Windows folder and double-click on the start.fastchat file.

[00:03:46]And there we go. The AI engine starts and the chat opens automatically in your browser. This part depends on your PC specs like RAM, CPU, and GPU. All right.

[00:03:58]Let's confirm everything is working. Let me just type hey there and send it. And yes, the AI is responding. Now I'll turn off the internet and try again.

[00:04:08]And yes, it still responds.

[00:04:11]That means our setup is working perfectly offline. You can also tweak settings like temperature or switch between models if you've downloaded more than one. And if you want more flexible responses, just copy the prompt from the description and paste it here.

[00:04:24]You only need to do this once and it will apply to all new chats. One last thing, make sure to stop the AI server when you're not using it. Just press control C to close it.

[00:04:36]Now this shared folder stores all your chats, settings like temperature and prompts. And the models folder contains all the models you downloaded along with their config files. So yeah, that's how this whole setup stays portable.

[00:04:49]Everything is inside the USB. Now if you want to download another model later, just run the installer again, select the model you want, and let it install.

[00:04:58]Once it's done, run the start.chat file again and you'll see the option to switch between models from the drop-down and start chatting with them right away.

[00:05:07]So that's it. This is how you can run a fully portable local AI setup on Windows.

[00:05:13]All right. Now let's see how to set this up on Linux. Just plug your same USB drive into your PC or download the same GitHub repo on your Linux system. Once that's done, the process is almost the same as before. You just need to extract the files just like we did earlier. Now inside the extracted folder, you'll find a Linux folder. This folder contains two important files, install and start files. Next, open the Linux folder in the terminal.

[00:05:40]This part is important because we'll be running commands directly from here.

[00:05:44]If this is your first time or you're using the USB on Linux for the first time, just run the install.sh script once. You don't need to run it again and again. It's a one-time setup. After that, you can directly use the start script. This installer is basically the same as the Windows setup. So it will feel familiar.

[00:06:02]Just select your model and if it's already downloaded, it will skip it.

[00:06:06]This may take some time depending on your internet and USB speed. So just wait. Once it's done, everything will run smoothly on Linux.

[00:06:15]And yes, the setup is now complete. Now we just need to run the start.sh file.

[00:06:21]So press any key to close the installer, then simply run bash start.sh.

[00:06:26]This will automatically start the uncensored AI local server on your Linux machine and it will also open the chat window in your browser. And as you can see, the chat page is already opened.

[00:06:37]And even our previous chat is still there from earlier setup.

[00:06:41]This means our portability is working perfectly.

[00:06:44]You can continue the same chat and it will keep everything in context or you can start a new chat anytime. Let me quickly confirm this. I say, "Hey, I am Jarvis." And it replies. Then I ask again, "What's my name?" And it remembers correctly. But if I ask the same thing in a new chat like, "What's my name?" it won't know because that context is not shared between chats.

[00:07:09]So yeah, our uncensored USB AI setup is working perfectly on Linux, too.

[00:07:16]Now, let's move to the Mac installation part.

[00:07:19]The process is exactly the same.

[00:07:22]Just plug in your USB or download the project.

[00:07:25]Open the Mac folder in the terminal and run bash install.command.

[00:07:32]After that, everything works just like before. Select your model and the script will automatically download and set up everything for you.

[00:07:40]I'll quickly do that now and fast forward this part.

[00:07:44]Setup is now complete. Now we just need to run the start.command file to start the local AI server. Open the terminal and run bash start.command. In a few seconds, it will start the uncensored local AI and automatically open the chat page in your browser. And there we go.

[00:07:59]As you can see, it's running on localhost. Since this is a fresh setup, there are no chats yet. But if you're using the same USB from Windows or Linux, you'll see all your previous chats right here. And that's it. Our uncensored local AI setup is now fully working across Windows, Linux, and Mac using the same USB.

[00:08:19]This means you can carry your entire uncensored AI with you, plug it into any system, and continue right where you left off without losing your chats or context.

[00:08:28]If you found this useful, make sure to subscribe for more videos like this and check the link in the description to download the project and try it yourself.

[00:08:37]Thanks for watching. This is Jarvis signing off.