How to run your first open‑source LLM from DeepSeek on a home PC

Download and install Ollama

To run an AI model on your computer, you first need a special helper application that translates between your computer's hardware and the AI's digital brain. We'll use a free, trusted tool called Ollama. Think of Ollama like a video player, while the DeepSeek AI is the movie file you want to play.

Visit the official Ollama website in your web browser. You'll see a prominent "Download" button or options for different operating systems like "Download for Mac" or "Download for Windows." Click the appropriate button for your computer. Once the download is complete, find the downloaded file (usually in your "Downloads" folder) and double-click it to start the installation. Follow the on-screen prompts, accepting the default options, much like installing any other software. If the website layout looks different, simply look for a section titled "Download" or "Get Ollama" and select your computer's operating system.

You'll know it worked when: The Ollama application is installed on your computer, and you might see an Ollama icon in your system tray (bottom-right on Windows) or menu bar (top-right on Mac).

Open your computer's command centre

To tell Ollama to fetch and start the DeepSeek model, we need to use a simple text command. This is done through a special window where you type instructions directly to your computer, often called the "command centre" or "terminal."

• On Windows: Click your Start menu (the Windows icon, usually bottom-left), type "cmd" into the search bar, and then click to open the "Command Prompt" app.
• On a Mac: Press the Command key and the Spacebar on your keyboard at the same time (this opens Spotlight Search), type "Terminal," and then press Enter.

A new window will open on your screen. This window usually has a black or dark background with white text and a blinking cursor. It might look a bit old-fashioned or "retro," but it's very powerful for specific tasks. If you see "PowerShell" open on Windows instead of Command Prompt, that's perfectly fine too, as it works similarly.

You'll know it worked when: You see a dark window on your screen with a line of text ending in a blinking cursor, typically showing something like C:\Users\YourName> on Windows or YourName@YourComputerName ~ % on Mac.

Download and run DeepSeek

Now we will fetch a lightweight version of DeepSeek's model. This is an LLM (Large Language Model), which is the digital engine or "brain" that powers a chatbot like ChatGPT. We'll use a smaller version of DeepSeek-R1 that is specifically designed to run smoothly on everyday home computers.

In your Command Prompt or Terminal window, type the following command exactly as shown. Pay close attention to spacing and punctuation, then press Enter. Your typed command will appear on the screen.

After you press Enter, your computer will automatically start downloading the DeepSeek model files. You'll see text indicating the download progress, often showing a percentage or a download speed. Depending on your internet connection, this could take a few minutes. If it seems to get stuck or you see an error, double-check your spelling of the command and ensure your internet is connected.

You'll know it worked when: The download completes, and the command line prompt changes to a chat interface, often showing a line like >>> or Send a message (/? for help), indicating the AI is ready for your input.

Start your private conversation

You are now connected to an offline AI running entirely on your local machine! This means your conversations stay on your computer and are not sent to any external company's servers.

At the >>> prompt, type your question or a command (this is called a "prompt") and press Enter. For example, you could ask it for ideas, summaries, or explanations. After you send your message, the AI will begin to "think" and then generate its response word by word, just like you might see in online chatbots.

If the AI's response seems to miss the point, try rephrasing your question with more detail. This AI is running entirely on your computer, so its speed will depend on your hardware.

You'll know it worked when: The AI finishes typing its full response, and a new prompt (like >>>) appears below, ready for your next input.

Finish your chat session

When you're finished chatting with DeepSeek and want to close the conversation, you can simply tell the AI you're done. This will return you to your computer's regular command line, where you started.

At the >>> prompt, type the special command /bye and then press Enter. The chat session will then end, and the AI will stop responding.

After typing /bye and pressing Enter, the chat interface (the >>> prompt) will disappear. You will then see your standard command line prompt again (e.g., C:\Users\YourName> or YourName@YourComputerName ~ %), indicating that you have exited the DeepSeek chat and are back to your computer's general command centre. If /bye doesn't seem to work, you can usually force quit by holding down the Control key (or Ctrl on Windows, Command on Mac) and pressing the C key a couple of times.

You'll know it worked when: The >>> chat prompt disappears, and you see your regular computer command line prompt again, ready for you to close the window or type other computer commands.