How to Install Ollama and Run Qwen2.5-Coder Locally — Step by Step

Before you start

A few things to know up front so nothing surprises you mid-installation:

• Operating system: Ollama runs on Windows 10 or later, macOS 11 (Big Sur) or later, and most Linux distributions. On Windows, make sure you have the April 2023 update or newer, because Ollama needs a recent version of WSL (Windows Subsystem for Linux — a small Linux layer Microsoft bundles with Windows).
• Disk space: Plan for at least 5 GB free for the 7B model (the version most people start with). The 1.5B version needs around 1 GB.
• RAM (random-access memory, your computer's short-term working memory): The 7B model needs about 8 GB to run smoothly; the 1.5B model runs comfortably on 4 GB.
• Account or payment? None. Ollama is free, and the Qwen models are open-weights (open-weights means the model's learned settings are publicly downloadable — anyone can run them locally).
• Internet: Required for the first download only. After that, the model works offline.

💬 What is Ollama? A free desktop program that downloads AI models and gives you a simple chat window in your terminal (the text-based window where you type commands instead of clicking icons). You don't need to know any commands to start — the steps below walk you through it.

💬 What is a "parameter"? When articles say "1.5B" or "7B," that's the model size in billions of internal settings the AI learned from during training. Bigger generally means smarter but slower.

Step 1 — Download the Ollama installer

Open your web browser and go to ollama.com. On the homepage, look for the large "Download" button near the top of the page. Click it, and the site will detect your operating system automatically — it offers separate files for Windows, macOS, and Linux.

Pick the file that matches your computer, and your browser will save it to your default Downloads folder. The Windows file ends in .exe, the Mac file in .dmg, and the Linux file in .sh.

💬 Example — what you'll see on the page: a row of three buttons labeled "Download for Windows," "Download for macOS," and "Download for Linux." Click the one that matches your system.

You'll know it worked when a file named something like OllamaSetup.exe (Windows) or Ollama.dmg (Mac) appears in your Downloads folder.

Step 2 — Install Ollama on your computer

Locate the installer file you just downloaded (usually in your Downloads folder) and double-click it. Follow the on-screen prompts — the Windows installer shows a short wizard, while the Mac installer opens a window where you drag the Ollama icon into your Applications folder. Accept the default settings; you don't need to change anything.

On Windows, the installer may also ask permission to install or update WSL. Say Yes if prompted — Ollama quietly relies on it. This part can take a couple of minutes.

💬 Example — Mac prompt you'll see: "Ollama" is an app downloaded from the Internet. Are you sure you want to open it? Click Open.

You'll know it worked when the installer finishes without errors and you see the Ollama icon (a little whale-shaped logo) in your system tray (Windows, bottom-right of the taskbar) or menu bar (Mac, top-right of the screen).

Step 3 — Open a terminal window

Ollama is controlled by typing short text commands, so you need a terminal — a plain text window where you type instructions for your computer. Don't worry, you only need a handful of commands, all listed below.

• On Windows: press the Windows key, type cmd, and press Enter. A black window titled "Command Prompt" appears.
• On macOS: press Command + Space to open Spotlight (the built-in search bar), type terminal, and press Enter. A white window titled "Terminal" appears.
• On Linux: open your usual terminal application from your apps menu.

💬 Example — first thing you might see in the window on Mac: a line like YourName@Mac ~ % followed by a blinking cursor. That cursor is where you type.

You'll know it worked when you see a blinking cursor at the bottom of a black (Windows) or white (Mac) window, with no error messages.

Step 4 — Pull and start the Qwen coding model

In the terminal window, type the following command exactly as shown and press Enter:

ollama run qwen2.5-coder:7b

This single command does two things: it downloads the model (called "pulling" in Ollama's vocabulary) and, once the download finishes, drops you straight into a chat with the AI. The first download is roughly 4.4 GB, so it can take anywhere from a few minutes to about 15 minutes depending on your internet speed. You'll see a progress bar made of percentage signs filling up as it downloads.

When the model is ready, the progress bar disappears and you see a >>> prompt. That's your cue — the AI is now waiting for your question.

💬 Example prompt to copy and paste at the >>> symbol: Write a Python script that renames every file in a folder to today's date, like report-2026-06-04.txt.

You'll know it worked when the >>> symbol appears and stays on screen, ready to accept your typing — even if it takes a few minutes to get there.

Step 5 — Ask the AI for code

With the >>> prompt blinking, type a request in plain English. Press Enter when you're done, and the model starts typing back its answer. Wait a few seconds — coding models often pause to "think" before producing code.

Start with something small and practical, like asking it to write a script, explain a snippet, or fix a bug you've been stuck on. The model responds in the same terminal window, line by line.

💬 Other starter prompts you can copy:

• Explain this Python line by line: print([x*2 for x in range(10)])
• Write a shell script that finds duplicate file names in my Downloads folder.
• Convert this Excel formula into a Python one-liner: =SUMIF(A:A, ">=10", B:B)

You'll know it worked when the model prints code (or a plain-English explanation) directly in the terminal, usually within 5 to 30 seconds.

Step 6 — Exit the chat and come back later

When you're done for the session, type /bye and press Enter. The terminal window closes the chat, but Ollama itself stays running in the background. To start a new session, just repeat Step 4 — the model is already downloaded, so it loads in seconds.

💬 Example — what you'll see when you type /bye: a short goodbye message from the model, then the terminal returns to its normal command line (something like YourName@Mac ~ % on Mac or C:\Users\YourName> on Windows).

You'll know it worked when the >>> prompt disappears and your normal terminal cursor returns, which means the chat has ended but Ollama is still installed and ready for next time.

Common mistakes

• "Command not found" after installing Ollama. You opened the terminal before installing, or didn't reopen it afterwards. Close the terminal window completely and open a fresh one — Ollama only adds itself to your system path (the list of programs your computer knows about) in newly opened windows.
• The 7B model feels painfully slow. Your computer probably doesn't have enough RAM, or your processor is older. Quit the chat with /bye, then try a smaller model by running ollama run qwen2.5-coder:1.5b — it uses far less memory and runs on most laptops made in the last five years.
• You closed the terminal and "lost" your chat. The conversation isn't saved between sessions in the default setup. If you want to keep a useful answer, copy it from the terminal and paste it into a notes file or text document before you type /bye.
• On Windows, the install hangs on a WSL step. Restart your computer once after the install finishes, then reopen the terminal and try the ollama run command again. The first launch on Windows sometimes needs a reboot to finish setting up WSL.