Deepseek R1 is shaking up the AI world. It’s a powerful Chinese AI model that recently overtook ChatGPT in search volume and App Store rankings. If you're worried about privacy or just want to explore cutting-edge AI on your own terms, running Deepseek R1 locally is your best bet. And the best part? You don’t need a supercomputer.
Why I suggest you to use it locally. Read Here
And why not have something that you can use anywhere without internet.
This guide will walk you through setting up Deepseek R1 on your computer, even if you’re a beginner.
Why Run Deepseek R1 Locally?
Privacy First: No more sending your data to China or OpenAI.
Faster Processing: No waiting in long queues due to server overload.
Cost-Effective: It’s free to run after setup, unlike cloud-based models.
Customizable: Fine-tune it for legal, coding, or any other specific use case.
Step 1: Install Ollama
Ollama is the engine that will run Deepseek R1 on your machine.
For Windows Users:
Visit the Ollama website and download the Windows version.
Follow the installation steps.
Open Command Prompt and type
ollama
to verify the installation.
Pro Tip: Ollama will run in the background. Check your system tray (Windows) or top menu bar (Mac) to confirm it’s active.
For Mac Users:
Go to Ollama’s official website.
Click Download and choose MacOS Installer.
Once downloaded, open the installer and run it.
Open Spotlight Search, type
Ollama
, and launch it.
Step 2: Download and Install Deepseek R1
Now, let’s get the AI model.
Open Terminal (Mac) or Command Prompt (Windows).
Copy and paste the following command: (For a larger model, replace
7b
with14b
or32b
based on your system’s capability.)
ollama run deepseek-r1:1.5b
(This is just the example, you choose according to your preferences)
Press Enter and wait for the download. This may take several minutes.
Once done, test the model with some prompt.
Choosing the Right Model:
7B – Runs on most modern laptops.
14B – Ideal for powerful machines with 16GB+ RAM.
32B+ – Requires high-end hardware with at least 64GB RAM.
Smaller models are dumber but fast. Larger models are smart but slow.
Higher models need higher specifications. Smaller models can be run of most modern laptops.
Step 3: Set Up a User-Friendly Interface
Running AI in a terminal isn’t fun. Let’s integrate it into a proper chat interface.
Install AnythingLLM
Visit AnythingLLM’s official site or LMStudio AI and download the app.
Install and open it.
Click New Workspace and name it (e.g., "Deepseek Chat").
Open Settings, scroll down, and select Ollama as the backend.
Choose Deepseek R1 from the available models.
Click Save Settings and start chatting!
Bonus 1: AnythingLLM allows chat history, API integrations, and even voice inputs.
Bonus 2: LMStudio allows you to run any open-sourced model locally for free.
Step 4: Test Deepseek’s Capabilities
Now it’s time to see what Deepseek R1 can do.
Try These Prompts:
Coding Test: "Write a Python script for a 3D rotating cube in ASCII."
Reasoning Challenge: "Explain why the sky is blue using a step-by-step logical approach."
Creative Writing: "Generate a sci-fi short story about AI taking over Mars."
Deepseek’s reasoning ability shines compared to OpenAI models because it doesn’t just predict the next word—it thinks before responding. You’ll see a thinking phase before it delivers its final answer, making it ideal for complex tasks.
Congratulations! You now have Deepseek R1 running locally.
Here’s how you can take it further:
Fine-Tune for Your Needs: Modify system prompts for law, medicine, coding, or education.
Experiment with Bigger Models: Try the 32B or even 70B model if your machine can handle it.
Explore Open-Source AI: Follow the latest AI breakthroughs from China, like Qwen and Kimi models.
Stay Adaptable. AI is moving fast—2025 will be the year of change. Keep learning, experimenting, and staying ahead of the curve.
Your Next Move: Join AI communities, contribute to open-source projects, and leverage AI to its full potential. The future is yours to shape.
OpenAI vs. Deepseek isn’t just about better AI—it’s about control.
Running AI locally gives you the power. Take charge today!
🚧 PS: Want more AI tips, tricks, and in-depth tutorials? Stay tuned for the next issue, where I’ll share something more useful for you.
Got a favorite AI prompt? Or an AI tool you swear by?
Let me know (@Manas Moon)—I’m always excited to learn new ways to use AI.
Manas | Founder @Codexai
Cheers xx!! 🥂