Run LLMs Locally on Windows in One Click with MiniLLM

My 8-Year-Old ThinkPad and Small Language Models

My daily machine is an 8-year-old ThinkPad X280 laptop with only 8 GB of RAM and integrated graphics. The specs are modest by any standard, and running large models on it is out of the question — but small 1–2B models actually run smoothly.

I am not a particularly technical person. I neither understand nor enjoy installing tools through terminal commands. All I want is something simple and fast that just works, without spending hours fiddling with code.

LM Studio: Excellent, but Heavy

At first I used LM Studio to deploy and run models locally. It is a genuinely good and powerful tool: you can download model files matched to your hardware straight from Hugging Face, and start running them right away. Other tools I tried, such as Anything LLM, did not come close to the same experience.

But then I noticed something: LM Studio itself occupies around 500 MB of memory. On a brand-new machine that would be nothing, yet on my aging laptop it is an overhead I simply cannot afford. Perhaps that is the price of being so feature-rich — but there was no lighter alternative offering the same convenience.

Ollama: Great, but Terminal Required

Next I tried Ollama for local deployment. It works well, but pulling models from GitHub requires command-line operations, and that alone put me off. I just want to get up and running quickly, not spend my time on code.

So the question became: is there a tool for Windows that installs, deploys, and runs large language models in one click? I could not find one.

Under the Hood: Everything Is llama.cpp

Here is the thing: whether it is LM Studio, Ollama, or any of the other tools, the engine underneath is llama.cpp. The catch is that installing and running llama.cpp itself also involves a whole series of command-line steps.

So I built a small tool with the help of AI coding assistants — a one-click installer that runs large models on Windows through llama.cpp. I call it MiniLLM.

MiniLLM: One-Click llama.cpp for Windows

MiniLLM is a portable single-file mini app. No installation needed — just open it and use it. The operation is simple, and there are no ads.

What MiniLLM Can Do

• One-click install, upgrade, and uninstall of llama.cpp.
• One-click start and stop of the inference server.
• Built-in builds of llama.cpp for different hardware — whether your PC has an NVIDIA GPU, an AMD GPU, an Intel GPU, or relies on CPU inference only, you can install the right version with one click.
• Automatic model detection: download GGUF model files from Hugging Face into a folder in advance, point MiniLLM at that folder, and it will recognize and run the models automatically.
• One-click access to the WebUI that ships with llama.cpp — once the server is up, start chatting with your local model immediately.

How to Use It

1. Download GGUF-format model files from Hugging Face and place them in a single folder.
2. Open MiniLLM and set that folder as your model directory.
3. Pick the llama.cpp variant matching your hardware and press Install.
4. Press Start Server, then press Open WebUI — and start chatting with your local model.

Who Is MiniLLM For

If your computer is old and underpowered, or if you — like me — would rather not touch the terminal, MiniLLM removes every unnecessary step between you and a locally running model. Download a GGUF file, click a few buttons, and chat. That is the whole workflow.

Local AI should not require a new computer or a developer background — just one small app that does the boring parts for you.