How to Install Deepseek Locally (Totally Free)

By Caner Öncü - March, 20th 2025

Before diving into how to install DeepSeek locally and run it for free, let’s briefly explain what DeepSeek really is.

What exactly is DeepSeek that people talk about?

Deepseek is an artificial intelligence tool developed by a Chinese team. It was originally built for data analysis and model training, especially used in the field of artificial intelligence and machine learning. It is simply a software miracle that we can ask questions about anything from food recipes to taking over the world. Just like ChatGPT… It’s basically a genie that answers everything.

Now you are thinking, “Dude, this is great! How can I run it on my own computer?”

Well, let’s install it step by step and run DeepSeek on your machine locally. And the awesome thing is, it is totally free!

How to install DeepSeek locally?

Using Ollama we can simplify the setup a lot. Ollama is a platform developed by Facebook that lets you run these models easily. Let’s get started!

1. Install Ollama

First, install Ollama on your system. If you are using Linux or Mac, open the terminal and do the following:

curl -fsSL https://ollama.com/install.sh | sh

If you are using Windows, you can download and install it from here: Ollama Download

Download Deepseek AI (Free)

Once the installation is complete, open a “Windows command prompt” and type this in (or type the following into your Linux/Mac terminal):

ollama run deepseek-r1

This process may take a while because the model files are very large. A little patience is required.

Once the installation is complete, you’re done! You can now chat with DeepSeek on your machine locally, by only typing into the open terminal. Quick and easy, right?

How Can I Access DeepSeek API Locally?

Warning: This part is for software developers. If you are not a software developer, you do not need to go into this detail. If you are not a software developer, it will be enough to chat with DeepSeek via the installation above.

If you want to dig deeper and running DeepSeek locally in the terminal is not enough, you can easily integrate it into your own projects. For that, you can use the DeepSeek API, as in the example below.

Sample Python POST request:

import requests
import json

url = "http://localhost:11434/api/generate"
data = {
    "model": "deepseek-r1",
    "prompt": "How are you Deepseek?",
    "stream": false
}

response = requests.post(url, json=data)
print(response.json())

You can easily run DeepSeek API on your own machine using Ollama.

So Can I Train DeepSeek Myself?

It is possible to train DeepSeek and have it respond according to your desired “context”. One option is to completely retrain a model, and the other is that you can customize an existing model and have it produce appropriate results. Training a model completely from scratch is a very difficult process, so modifying an existing model may be a more logical strategy. I will not go into the details of this in this article, but I plan to write an article about it in the future.

That’s all for now. If you have any questions, please let me know in the comments and I’ll try to help as much as I can.

For the lazy ones, here is a video on how to do it:

Cheers!