Ziqi Chen

If you are going through hell, keep going.

How to chat with your local ChatGPT?

AIs like ChatGPT have becoming increasingly inseparatable for our local lives. Everytime platforms like ChatGPT or Claude went down, or our internet connection became unstable, our productivity drop dramatically and we cannot continue to work. With Ollama, you will be free from such panic and enjoy ChatGPT on your own computer even without internect connection!

0. Prerequest

  • Laptop with at least 16GB RAM (For my Macbook Pro 1 with 16GB RAM, it is easy to run Gemma3 4b)
  • Access to GitHub

1. Download Ollama

  • Go to Ollama and download the latest version.
  • Once you finished download, you do not have any AI on your computer yet. So you have to download the one you want by yourself.

2. Download AI

  • Open your Terminal.
  • Run the following command: 
    ollama pull gemma3:1b
  • This command will doanload Gemma3 1b, which is basically the lightest AI.
  • For more models please check Ollama document here.

    Note: The bigger the size is, i.e. the number before b, the more powerful the model is. If your small model cannot handle your request, maybe it is because you feed it too many stuff and it cannot handle. Try with a larger one.

3. Run AI

  • Open your Terminal.
  • Run the following command: 
    ollama run gemma3:1b

4. Setup Web UI (Optional)

If you want to use an interface just like ChatGPT, you may choose Open-WebUI, which is the user interface supported for Ollama.

4.1 Install Docker
  • Go to Docker and download their latest verion.
4.2 Set up Open-WebUI
  • Open your Terminal
  • Run the command 
    docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

    Note: Everytime you need to open the docker container if using Open-WebUI.

4.3 Open the Website
  • Go to your browser and open http://localhost:3000/

Now let’s chat on your own computer!