DeepSeek has taken the AI market by storm, and probably everyone has tried the Chinese neural network by now.

However, due to overload, Deepseek often doesn't work. The neural network doesn't respond, sending errors like:

The server is busy. Please try again later.
Deepseek network error. Please try again later.

In this article, we'll look at 4 ways to access Deepseek, even if it's not responding. In the second part of the article, we'll see how to install DeepSeek locally on your computer, which allows you to use it without any internet connection at all!

1. A simple way to solve the 'Deepseek service is busy' problem

The most obvious, but effective, method is to install an application from third-party developers. These applications use the Deepseek API, which is less prone to problems than the official website and mobile app. There are already many of them:

Mobile apps for smartphones (options for iOS and Android)
There are also a large number of Telegram bots.
On a computer: use browser applications or read another option below.

2. Solving the problem via OpenRouter (with access to the DeepSeek API as a bonus)

OpenRouter is another cool third-party service. But through it, you can programmatically connect to the DeepSeek API, and for free!

It also works through a regular chatbot interface:

To connect, simply go to this OpenRouter page and click the Chat button (by the way, the service works without a VPN):

However, if the DeepSeek servers are under too much strain, not only the official applications but also the API can go down.

So let's move on to the most interesting part: installing Deepseek locally.

3. Downloading Deepseek locally via LM Studio

In short, LM Studio is a platform that allows you to deploy DeepSeek R1 and other neural network models directly on your computer. This allows you to use the neural network for free and without an internet connection.

Let's see how to install DeepSeek locally:

Go to the official website of LM Studio.
Install and run LM Studio (works on Windows, MacOS, and Linux).
Go to the Discover section:
There, find the DeepSeek R1 models:
There will be 2 models to choose from: DeepSeek R1 Distill (Qwen 7B) and DeepSeek R1 Distill (Llama 8B).
Download one of the 2 models (it can take from 10 minutes to an hour).
After downloading, click Use in new chat.
Chat with DeepSeek even without an internet connection.

Just in case: the model descriptions state that a minimum of 16 GB of RAM is required, but it worked for me even with 8 GB.

4. Install DeepSeek Coder locally via Jan.ai

You can also deploy Deepseek locally using Jan.ai. This method allows you to install DeepSeek Coder, which is especially useful for developers when programming.

Installing Jan.ai:
- Go to the official Jan.ai website and download the version for your OS (Windows, macOS, Linux).
- Install it as usual, following the instructions. Run it.
Downloading the DeepSeek language model:
- Open Jan.ai and go to the "Models" section:
- Enter DeepSeek in the search bar and choose one of the 2 models depending on your hardware's power (the larger the model's size, the more complex tasks it can solve, but it also requires more computing resources):
- Select a model and start downloading (it can take from a few minutes to a couple of hours).
Configuring DeepSeek for local use:
- After downloading, go to Thread:
- Select DeepSeek and configure the parameters (memory, output temperature, etc.):
- Save the changes.
Running and testing:
- Go to settings and click "Run model", wait for initialization:
- Enter a test query and get a response without an internet connection.

By installing Deepseek locally, we get the following advantages:

No dependence on Chinese servers — it works offline, on a trip, anywhere.
Free – even if Deepseek becomes a paid service someday, the locally installed model will still be free.
Confidentiality — all chat data remains on your computer and is not sent to the internet.

Share your ways of using DeepSeek locally in the comments. We will add them to the article.

Deepseek is not working: Bypassing the 'Deepseek service is busy' error with clever methods