AI

Deploy DeepSeek Locally with Ollama

Install Ollama

Ollama is a lightweight environment for running LLMs locally. Download and install it from the official website.

Windows users: After installation, log out and log back in to apply system variables correctly.

Run DeepSeek 14B

Ollama will automatically download the model upon first execution.

  • Run DeepSeek 7B (default model):

    1
    ollama run deepseek-r1

  • Run DeepSeek 14B (requires 16GB VRAM):

    1
    ollama run deepseek-r1:14b

Configure Environment Variables (Optional)

To enable external access, set the following environment variables:

  • Windows:

    1
    2
    setx OLLAMA_HOST "0.0.0.0:11434"
    setx OLLAMA_MODELS "%USERPROFILE%\.ollama\models"

  • Linux/macOS:

    1
    2
    export OLLAMA_HOST="0.0.0.0:11434"
    export OLLAMA_MODELS="$HOME/.ollama/models"

Start Ollama Server

To run Ollama as a background service and allow remote access:

1
ollama serve

Ollama will now listen on port 11434, making it accessible locally and remotely.

Test Your Deployment

After launching, run:

1
ollama run deepseek-r1:14b

Alternatively, use Ollama API:

1
2
3
4
5
curl http://localhost:11434/api/generate -d '{
    "model": "deepseek-r1:14b",
    "prompt": "Hello, introduce the features of DeepSeek 14B",
    "stream": false
}'

Exit Ollama

To stop Ollama, you can simply close the terminal or type:

1
/bye

Summary

Ollama provides a lightweight local environment for running LLMs
DeepSeek 14B requires 16GB+ VRAM
Set OLLAMA_HOST to enable remote access
Run ollama serve to start the API server