How to Run DeepSeek R1 Locally Using Ollama (macOS, Windows, and Ubuntu Guide)
Looking to harness the power of large language models without depending on the cloud? Meet DeepSeek R1 — a high-performance, open-source AI model that rivals, if not surpasses, many of OpenAI’s O1 models. In this guide, we’ll show you how to run it safely and locally using Ollama — across Mac, Windows, and Ubuntu.
🔍 What is DeepSeek R1?
DeepSeek R1 is a state-of-the-art large language model trained on a massive 2T token dataset. It’s optimized for reasoning, coding, multi-language tasks, and more.
Why it stands out:
- 🧠 Comparable or better than OpenAI's O1 models in reasoning and dialogue
- 🔐 100% private and local — no data ever leaves your machine
- ⚙️ Great for developers, researchers, and AI enthusiasts
- 💬 Excels in code generation, math, and natural conversation
🧰 What is Ollama?
Ollama makes it dead simple to run powerful LLMs like DeepSeek R1 locally.
- 🎯 One-line install
- 💻 Works on macOS, Windows (via WSL), and Ubuntu
- ⚡ Supports GPU acceleration
- 📡 No cloud, no spying, no latency
✅ System Requirements
- macOS (Intel or Apple Silicon), Windows 11 (with WSL), or Ubuntu 20.04+
- 10GB+ disk space for the model
- 8GB RAM minimum (16GB recommended)
- Optional: GPU for faster inference
🛠️ How to Install and Run DeepSeek R1
🍏 macOS
-
Install Ollama:
brew install ollama -
Run a test model:
ollama run hello -
Pull and run DeepSeek R1:
ollama run deepseek
🪟 Windows (WSL Required)
-
Install WSL and open Ubuntu in WSL.
-
Inside the Ubuntu terminal:
curl -fsSL https://ollama.com/install.sh | sh -
Start the server:
ollama serve -
Run DeepSeek R1:
ollama run deepseek
🐧 Ubuntu (Linux)
-
Install Ollama:
curl -fsSL https://ollama.com/install.sh | sh -
Run a test model:
ollama run hello -
Run DeepSeek R1:
ollama run deepseek
💬 Interact With the Model
You can talk to DeepSeek right from your terminal or integrate it via API:
curl http://localhost:11434/api/generate -d '{
"model": "deepseek",
"prompt": "Explain general relativity like I'm 12."
}'
🔒 100% Local, 100% Private
Running DeepSeek R1 via Ollama means:
- 🧠 No data leaves your device
- 💬 No internet required after install
- 🧘 Full control over your AI stack
Perfect for sensitive work, dev environments, and peace of mind.
🧠 Why Choose DeepSeek Over O1 Models?
While OpenAI's O1 models are impressive, DeepSeek R1 offers:
- ⚖️ Equal or superior reasoning performance
- 🔍 More transparency and control
- 💸 Zero cost per query
- 🚀 Offline, high-speed inference
🎯 Final Thoughts
DeepSeek R1 is a powerful, private, and production-ready model you can run entirely offline. With Ollama, setup is effortless — and you'll be chatting with a top-tier AI model in minutes.
No API keys. No subscriptions. No cloud. Just you and your model.
Try it today and see just how powerful local AI can be.