Loading video…
All breakouts
38.8App DemoHow-To hookAI Tools
Hook
RUN OWN AI MODEL FROM USB - FULLY OFFLINE
@cubesetups
26.5K followers · usually 2.1K views a post
Views
81.0K
Likes
4.7K
Comments
65
Shares
1.3K
Promoted product
Llamafile
LLM inference in C/C++. Contribute to ggml-org/llama.cpp development by creating an account on GitHub.
Why it worked
The video provides a practical, step-by-step guide to a complex technical process (running an AI model offline), making advanced technology accessible to a wider audience. The use of a USB drive for portability and offline functionality is a compelling feature that likely resonates with users concerned about privacy or internet connectivity.
Summary
This video demonstrates how to set up and run a private AI model offline using a USB drive. It guides viewers through formatting a USB, downloading the Llamafile executable and a specific AI model from Hugging Face, creating a batch file to run the model, and accessing the AI interface via a local IP address.
Structure
- 1Prepare USB drive
- 2Download Llamafile and AI model
- 3Create batch file
- 4Run AI model locally
Original caption
100% working way to run your own AI from a USB drive. #aitools #offlineai #ai #TechTips #usbai How to Run a Private AI Model From a USB Drive Want to carry your own AI with you? Here’s the step-by-step setup. 1. Prepare Your USB Drive Plug a USB drive with at least 8GB into your computer. Right-click the drive in File Explorer → Format, select exFAT, enable Quick Format, and click Start. 1. Download Llamafile Open Google and search “llamafile GitHub.” Open the official GitHub page, go to Releases, and download the main llamafile executable from a release such as v0.10.5. Once downloaded, rename it so the file ends with .exe, for example: llamafile-0.10.5.exe Copy this file directly to your USB drive. 1. Download the AI Model Go to Hugging Face → Models and filter the results by the GGUF library. Search for a lightweight model such as Qwen/Qwen3. Open the model, go to Files and versions, and download the appropriate .gguf quantized model file, around 2–4GB. Once downloaded, copy the GGUF file to your USB drive. 1. Create the Startup File Open Notepad and enter the following command, replacing the filenames with your exact files: .\llamafile-0.10.5.exe --server --model qwen3-4b-thinking-2507.Q4_K_M.gguf Now go to File → Save As. Name it: start.bat Set Save as type to All Files, then save it directly on your USB drive. 1. Start Your AI Open the USB drive and double-click start.bat. A command window will appear. Look for the local address shown near the bottom, usually something like: http://127.0.0.1:8080 Open that address in your browser. And that's it—you now have an AI model running locally from your USB drive, without needing an internet connection for the model itself. Save this guide so you can set it up later.
More from @cubesetups
More breakout videos from this creator.
