How to Deploy DeepSeek Locally on Android

Today, I want to talk about how to install DeepSeek on an Android phone using Termux and Ollama.

If you also want to experience powerful AI capabilities on your mobile device, keep reading. The steps are actually not complicated, and I will guide you through them one by one.

Step 1: Install Termux

First, we need to install Termux, a tool that provides a complete Linux environment on Android phones.

You can think of it as a “virtual machine” on your phone, allowing you to run various command-line tools, install software, and even set up server-like environments.

For your convenience, I have organized all the software resources used in this tutorial into a cloud drive, so you can download what you need.

Click the public account card below and reply with the keyword:deepseek, to get the cloud drive link.

Next, remember to grant Termux storage permissions so that you can manage files and download models smoothly. Without storage permissions, some steps may encounter issues.

Enter the following command to obtain storage permissions for your phone:

termux-setup-storage

Step 2: Install Environment Dependencies

In the Termux environment on your Android phone, we also need to install some dependencies. Common tools include git, cmake, and golang, which are essential for installing DeepSeek later.

You can enter the following command in the Termux terminal:

pkg install git cmake golang

The functions of these tools are as follows:

• git is used to download source code;
• cmake is used to build and compile software;
• golang is a programming language used to run specific applications.

Step 3: Download and Install Ollama

After installing the environment dependencies, the third step is to install Ollama. Ollama is an AI model running platform that allows us to efficiently run various deep learning models on Android phones, including the DeepSeek we want to use.

First, we need to download the source code of Ollama using git. You can use the following command to complete the download:

git clone --depth 1 https://github.com/ollama/ollama.gitcd ollamago generate ./...go build .

After downloading, you will need to compile and install Ollama using some commands. During this process, you may encounter some network issues (such as slow download speeds). In that case, you can try using a proxy or patiently wait for the download to finish.

Step 4: Start Ollama

After installation, the next step is to start the Ollama service to run in the background. You can enter the following command:

./ollama serve &

This command starts the Ollama service, and it will continue to run in the background. You can check its status or manage it at any time.

Step 5: Download and Run the DeepSeek-R1 Model

Now, the next step is the most critical—downloading the DeepSeek-R1 model and running it on the Ollama platform. You can use the following command to download the model and run it:

x ./ollama run deepseek-r1:1.5b --verbose

When running the model, ensure that your device has sufficient performance, as such deep learning models typically require high computational power. Test to see if the model can work properly.

Notes

1. Phone Performance Requirements: It is recommended to use a newer Android device, preferably with strong memory and CPU, so that the operation runs more smoothly. If the phone is too old, running the model may lag or crash.

2. Storage Space: Deep learning models are usually large, and files like the DeepSeek-R1 model may occupy dozens of GB of storage space, so make sure your phone has enough available space.

3. Network Connection: Downloading software and model files requires a stable network, especially since model files can be quite large. If your network is unstable, consider using a proxy or other acceleration tools.

Accessing Related Resources

To make it easier for everyone, I have provided download links for all the necessary software resources, which you can directly download from the cloud drive. This way, you don’t have to search for these files online, saving you a lot of trouble.

Click the public account card below and reply with the keyword:deepseek to get the cloud drive link.

In summary, although installing and running DeepSeek on an Android phone requires some steps, it is not particularly complicated. As long as you follow the steps and debug slowly, you will find that your phone has become a powerful AI tool, capable of conducting deep learning experiments at any time.

I hope this article helps you, and I wish you all the best on your journey into AI!