How to Run DeepSeek-R1 Models Locally Using Ollama

In today’s rapidly evolving AI ecosystem, the ability to run powerful language models locally offers a range of compelling benefits—enhanced data privacy, cost-efficiency, offline functionality, and greater control over the deployment environment. The DeepSeek-R1 family of open-source language models delivers robust reasoning, code generation, and problem-solving capabilities, rivaling proprietary solutions—all while running on your hardware.

This guide outlines how to deploy DeepSeek-R1 models locally using Ollama, a user-friendly framework for managing and running LLMs on personal systems. We also demonstrate how to optionally expose your local model as a web API using Pinggy, and explore graphical interaction through Open WebUI.

Key Advantages of Running DeepSeek-R1 Locally

• Data Privacy: Your inputs and outputs remain entirely on your machine.

• Cost-Free Operation: No API usage limits or monthly fees.

• Offline Accessibility: Operate fully without an internet connection.

• Flexible Model Sizing: Select from lightweight to high-capacity models based on your hardware.

Step-by-Step Installation Guide

Step 1: Install Ollama

1. Visit the Ollama download page.

2. Select the version compatible with your operating system (Windows, macOS, or Linux).

3. Follow the installation instructions for your platform.

4. Confirm installation:

ollama --version

Step 2: Download a DeepSeek-R1 Model

Choose a model variant suited to your system's memory capacity:

Check available models:

ollama list

Step 3: Run the Model

To launch a model in your terminal:

ollama run deepseek-r1:1.5b

Try it with reasoning tasks:

User: Solve this step by step: If I have 12 apples and give away 1/3, then eat 2, how many are left?

Using DeepSeek-R1 via API

1. Start the Ollama server:

ollama serve

2. Make requests using curl:

curl --location 'http://localhost:11434/api/chat' \
--header 'Content-Type: application/json' \
--data '{
  "model": "deepseek-r1:1.5b",
  "messages": [{"role": "user", "content": "Hello"}]
}'

You can also integrate the API into Node.js by cloning the RunOllamaApi repository, installing dependencies, and running your script with:

node main.js

Graphical Interface with Open WebUI

If you prefer a user-friendly GUI, install Open WebUI, which offers a ChatGPT-like experience.

Installation via Docker:

docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

Access the interface at http://localhost:3000. Select your DeepSeek model and begin chatting, uploading files, and organizing conversations.

Key Features:

• Multi-modal input (text + images)

• Session and chat history

• Preset prompts and system instructions

• Mobile compatibility

Sharing Access with Pinggy (Optional)

Pinggy allows you to expose your local Ollama API over the internet.

Steps:

1. Ensure the Ollama server is running:

ollama serve

2. Create a public tunnel using Pinggy:

ssh -p 443 -R0:localhost:11434 -t qr@a.pinggy.io "u:Host:localhost:11434"

You’ll receive a URL such as https://abc123.pinggy.link, which can be used to access your API remotely.

Verify Remote API:

curl https://abc123.pinggy.link/api/tags

Performance Optimization Tips

• Use quantized models to reduce memory footprint:

ollama run deepseek-r1:1.5b-q4_K_M

• Tweak parameters for more dynamic responses:

ollama run deepseek-r1:1.5b --temperature 0.7 --top_p 0.9

• For slow performance, reduce context size:

ollama run deepseek-r1:1.5b --num_ctx 1024

About DeepSeek-R1 Models

DeepSeek-R1 is a suite of open-source language models available under the permissive MIT License—ideal for both research and commercial use.

Architecture Overview:

• Qwen-based Models: 1.5B, 7B, 14B, 32B

• LLaMA-based Models: 8B, 70B

Conclusion

Running DeepSeek-R1 locally via Ollama provides a seamless, secure, and high-performance LLM experience. Whether you're a developer, researcher, or enthusiast, you can now harness cutting-edge AI capabilities on your machine, without compromising on privacy or control.

Get Started Now:

ollama pull deepseek-r1:1.5b && ollama run deepseek-r1:1.5b

References

1. How to Run DeepSeek Locally

2. How to Easily Share Ollama API and Open WebUI Online

3. Pinggy’s Official Website

4. DeepSeek documentation

5. Ollama GitHub repository