How to Use DeepSeek: Web, App and Local Install (Step by Step)

Option 1 — Use DeepSeek in your browser (easiest)

1. 1

   Open the official site

   Go to chat.deepseek.com (check the address carefully — copycat sites exist).

2. 2

   Sign up

   Create an account with your email or phone number, or sign in with a supported account.

3. 3

   Choose the mode

   Use the normal mode for quick answers. Turn on the reasoning / “DeepThink” option for maths, logic and coding, and web search when you need up-to-date information.

4. 4

   Start chatting

   Type your question, or attach a file or image where supported. Use the prompt formula from the Prompting module for better results.

Option 2 — Install the official mobile app

1. 1

   Open your official app store

   Google Play Store on Android or the App Store on iPhone.

2. 2

   Search “DeepSeek”

   Pick the app published by DeepSeek itself. Check the developer name and download count — avoid look-alike apps that ask for payments or unusual permissions.

3. 3

   Install and sign in

   Use the same account as the website so your chats sync.

4. 4

   Never install from random links

   Do not download APK files from social media links; fake AI apps are a common way to spread malware.

Option 3 — Run DeepSeek on your own computer with Ollama (free, private, offline)

Ollama is free software that downloads and runs open-weight AI models. The DeepSeek-R1 models available in Ollama come in several sizes. As a rough guide: 8 GB RAM → try the 1.5b or 7b size; 16 GB RAM → 7b or 8b; 32 GB RAM or a good graphics card → 14b; larger sizes need workstation hardware. You also need several gigabytes of free disk space per model.

1. 1

   Download Ollama

   Go to ollama.com and download the installer for Windows, macOS or Linux. On Linux the site gives a one-line install command.

2. 2

   Install it

   Run the installer like any other program. On Windows and Mac, Ollama then runs quietly in the background.

3. 3

   Open a terminal

   Windows: open PowerShell or Command Prompt. Mac: open Terminal. Linux: open your terminal.

4. 4

   Download and start DeepSeek

   Type: ollama run deepseek-r1 — this downloads the default size the first time and then starts a chat. For a specific size type e.g. ollama run deepseek-r1:7b or ollama run deepseek-r1:1.5b for older/low-RAM computers.

5. 5

   Chat

   Type your question at the >>> prompt and press Enter. You will see the model’s reasoning followed by its answer. Type /bye to exit.

6. 6

   Optional: a chat window instead of the terminal

   Newer Ollama versions include a simple app window, or you can install a free interface such as Open WebUI. Alternatively use LM Studio (next section).

ollama run deepseek-r1:7b     # download (first time) and chat
ollama list                    # see downloaded models
ollama rm deepseek-r1:7b       # delete a model to free space
ollama pull deepseek-r1:7b     # download without starting a chat

Alternative — LM Studio (no terminal needed)

1. 1

   Install LM Studio

   Download it from lmstudio.ai for Windows, Mac or Linux and install it.

2. 2

   Search for DeepSeek

   Open the Discover / search tab and search “deepseek”. LM Studio marks which versions fit your computer’s memory.

3. 3

   Download and load

   Download a recommended size, then load it in the Chat tab and start typing.

Option 4 — DeepSeek API for developers

1. 1

   Create a developer account

   Sign up on DeepSeek’s developer platform (platform.deepseek.com) and add credit if required.

2. 2

   Create an API key

   Generate a key in the API keys section and store it securely — never put it in public code.

3. 3

   Call the API

   DeepSeek’s API follows the OpenAI-compatible format, so many existing libraries work by changing the base URL and model name. Check the official documentation for current model names and prices.

Troubleshooting

• “Server busy” on the website: popular times get crowded — try again later or use a local model.
• Local model is very slow: choose a smaller size (e.g. 1.5b or 7b) and close other heavy programs.
• “ollama” is not recognised: restart the terminal (or the computer) after installing so the command is found.
• Model runs out of memory: pick a smaller size; the number in the name (7b, 14b) is roughly how heavy it is.