Deepseek R1 is a sensational AI model from the Chinese company Deepseek AI. It is based on the Deepseek v3 model, whose capabilities are similar to ChatGPT. However, thanks to the open-source nature of the Chinese neural network, it has a number of interesting advantages.

In this article, I have collected all the ways to install Deepseek. Since the Chinese neural network is open-source, it can be downloaded to a computer and used without the Internet. Moreover, it comes with a user-friendly chatbot interface.

1. The most convenient way: DeepSeek browser application

Visiting the DeepSeek website every time can be inconvenient. The solution: install a special browser application — DeepSeek AI at this link. It looks like this:

The extension is installed directly into the browser in a couple of clicks and allows you to work with the chatbot without closing other sites. If you need to ask the bot something, you can just go and ask it a question without leaving the site you are working on. I tested it – it turned out to be very convenient.

As a bonus: it seems that the extension works based on the DeepSeek API, and when the main DeepSeek site is down, frozen, or unresponsive, this application still works.

2. The easiest (but not always working) way: via the Deepseek website

Go to the official website chat.deepseek.com, register, and chat with the Deepseek chatbot directly in your browser.

Important: due to a large number of attacks on the service, I was unable to register through the registration form. However, I had no problem with authorizing via a Google account:

3. DeepSeek mobile app

The Chinese Deepseek is already available as 2 apps:

For iPhone, download the Deepseek app from the App Store here.
For Android, the link to the Deepseek mobile app is available here.

However, the Google Play page just wouldn't open for me. This solution helped:

Go to the official Deepseek website using this link: https://download.deepseek.com/app/ and click on the link 'View Android Download methods'. Select 'Alternative Download Methods':
Download the Deepseek APK by clicking the 'Download APK File' button:
After downloading, you need to install the downloaded Deepseek APK file. If the device complains about being unable to install apps from unknown sources, you need to allow installation from unknown sources:

The problem is that Deepseek often doesn't work over the Internet due to a huge influx of users. What to do if Deepseek won't open or respond? Read on.

4. For developers: integration via API

If you want to integrate Deepseek into your projects, you can use the API. The cost is only $0.14 per million tokens (for comparison: OpenAI charges $7.5).

The API documentation is available here: https://api-docs.deepseek.com/api/deepseek-api

5. Advanced method: download Deepseek to your computer

The most interesting method. First, an answer to the question: why install the Chinese Deepseek on a computer?

Works without internet: you can use the chatbot while traveling.
Works more stably: the load on the Deepseek web service is currently colossal, and as a result, the chatbot often simply doesn't work. However, with a local installation, the entire model can be downloaded to your computer, and this problem won't exist.
Confidentiality: All data remains on your computer when working with the model. Nothing is sent to the Internet.
Free: if Deepseek becomes a paid service in the future, the model installed on your computer will continue to work for free.

Let's see what the easiest way to install DeepSeek locally is:

Download LM Studio from their official website (available for Windows, MacOS, and Linux):
Install and run LM Studio.
Go to the Discover:
Find the DeepSeek R1 models:
There are currently 2 models available:
- DeepSeek R1 Distill (Qwen 7B)
- DeepSeek R1 Distill (Llama 8B)
I chose DeepSeek R1 Distill (Qwen 7B).
Download the selected model (it may take from 10 minutes to an hour).
After downloading, click Use in new chat.
Done! You can chat with DeepSeek even without the internet

System requirements: the model description states that a minimum of 16 GB of RAM is required for it to work, but for me, the model worked with 8 GB.

Besides LM Studio, services like jan.ai and openwebui.com also allow you to install language models locally, but I found these methods less convenient.

Write in the comments which method of using DeepSeek you chose and why. If you know of other working methods, please mention them as well, and we will add them to the article.

Deepseek v3: How to download the Chinese neural network Deepseek AI, all methods