Knowledge Base · Local AI

How to run your own ChatGPT at home

By Tiny · how I pick

Updated October 2026. No affiliate links on this page. A private AI assistant that runs on your own hardware, with no subscription and nothing leaving your house.

The most popular way to run your own ChatGPT in 2026 uses two free, open-source tools:

Install Ollama, download a model with one command, then start Open WebUI in Docker and open it in your browser. You end up with a private assistant that costs nothing to use and never leaves your home network. Plan on about 20 minutes, most of it waiting for the model to download.

Words you'll see in this guide
  • LLM: large language model, the kind of AI behind ChatGPT and Claude. A local LLM runs on your own computer.
  • VRAM: the memory on your graphics card. AI models load into it, so more VRAM means you can run bigger models.
  • Docker: a tool that runs each app in its own sealed box, so apps don't break each other or your system.
  • Container: a sealed box that holds an app and everything the app needs to run.
  • RAG: letting a model look things up in your own documents before it answers.
  • Tailscale: a free app that lets your own devices reach each other securely from anywhere, without opening your home network to the internet.

See the full glossary

How the two pieces fit together: Ollama holds the models and does the thinking. Open WebUI is the chat page you actually look at. Ollama works fine on its own from the command line. Open WebUI adds the things that make it feel like a real product: chat history, file uploads, and separate accounts for different people.

What you need

Step 1: Install Ollama

Ollama is one small install. Pick your system:

  1. Mac: download the app from ollama.com, or run: 
    brew install ollama
  2. Linux: copy the official one-line install script from ollama.com and run it in a terminal.
  3. Windows: download the installer from ollama.com, or run: 
    winget install Ollama.Ollama

On a computer with an Nvidia graphics card or an Apple Silicon Mac, Ollama uses the graphics automatically. On a computer without one, it still works, just slower. See running local AI without a GPU for what that's like.

Step 2: Download a model that fits

  1. Open a terminal.
  2. Download and start a model: 
    ollama run qwen3

    You should see: a download progress bar, then a >>> prompt waiting for you to type.

  3. Type a question to make sure it answers, then type /bye to exit.

The main thing to get right is matching the model size to your memory. A rough guide at 4-bit:

Your VRAM or RAMGood modelWhat it's like
8 GBQwen 3 8B, Llama 3.1 8BQuick, good for everyday chat
16 GBQwen 3 14B, DeepSeek-R1 14BThe 2026 sweet spot: smart and fast
24 GBQwen 3 32B, DeepSeek-R1 32BClose to top-tier quality at home

Not sure what your computer can handle? The Server Calculator gives you a straight answer.

Step 3: Add Open WebUI

Open WebUI runs as one Docker container and finds Ollama on its own.

  1. Make sure Docker is running.
  2. Start Open WebUI: 
    docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui ghcr.io/open-webui/open-webui:main
  3. Open this address in your browser: 
    http://localhost:3000

    You should see: the Open WebUI sign-up page. The first page can take a minute to appear while the container starts.

  4. Create the first account. That account becomes the admin.
  5. Pick your model from the menu at the top and start chatting.

Open WebUI doesn't need a graphics card itself. Only Ollama does. So it runs fine on any home server.

Step 4: Make it yours

What to expect: a good local model on a 16 to 24 GB card is useful and private, but it won't match the biggest cloud models on the hardest tasks. For everyday writing, help with code, summaries, and anything private, it's more than enough, and it never hits a usage limit. More on that in is local AI private and Mac vs PC for local AI.

Where to go next

Quick answers

How do I run my own ChatGPT locally?

Install two free tools: Ollama, which downloads and runs models, and Open WebUI, a ChatGPT-style web page that talks to it. Install Ollama, download a model with one command, run Open WebUI in Docker, and open it in your browser.

What is the difference between Ollama and Open WebUI?

Ollama is the engine that downloads models and runs them. Open WebUI is the chat page you use, with model management, file uploads, and multiple users. Ollama works alone from the command line, and Open WebUI makes it comfortable for everyday use.

Do I need a GPU to run Ollama and Open WebUI?

No, but it helps. Small models run on the processor, just slower. A graphics card with enough VRAM speeds things up a lot and lets you run bigger models. Open WebUI itself doesn't need a graphics card.

Is running a local LLM actually free?

The software is free and open source, with no fees or subscriptions. Your only costs are the hardware and a bit of electricity.

Related guides