DeepSeek Installation Guide for Beginners: Getting Started with DeepSeek V3 and DeepSeek R1

DeepSeek is a Chinese AI model founded in 2023 by Hedge Fund manager Liang Wenfeng. It soared beyond ChatGPT on the US App Store and triggered a historic drop of 17% (almost $600 billion) in Nvidia’s stock on January 27, 2025, all while costing around $5.576 million to develop.

Among DeepSeek’s offerings, models V3 and R1 stand out, each running on 671 billion total parameters with 37 billion activated per token. DeepSeek R1 uses large-scale reinforcement learning for next-level reasoning, while DeepSeek V3 handles multilingual tasks and text generation.

This guide will walk you through everything you need to DeepSeek installation for models V3 and R1: from basic commands to hardware tips, so you can set up and install DeepSeek even if it’s your first time trying it.

Want to check out how DeepSeek and ChatGPT compare? Check out upGrad’s blog, DeepSeek vs ChatGPT: What's The Difference and Which is Better.

What are the Hardware Prerequisites You Should Know Before You Install DeepSeek?

Although DeepSeek V3 is officially geared toward Linux with Python 3.10, it doesn’t come with a rigid, one-size-fits-all hardware checklist. Users have managed to run V3 on Windows or macOS through community frameworks, but you should still have a strong GPU, enough memory, and plenty of storage to deal with its 671B parameters.

When it comes to DeepSeek R1, there are clearer guidelines since R1’s local installation is more common for everyday usage. Here are the different R1 model variants:

• 1.5B or 7B: Lighter models for simpler tasks or machines with limited resources
• 14B or 32B: Mid-range versions that balance performance and system demands
• 70B or 671B: Best for advanced reasoning, but come with steep hardware needs

Here’s a breakdown of the requirements for DeepSeek R1:

Also Read: Top 12+ Best Courses in Artificial Intelligence To Enhance Your Skills for Career Growth in 2025

Now, let’s begin the step-by-step tutorial on how to install and use V3 and R1, starting with DeepSeek V3.

Getting Started With DeepSeek V3: A Step-by-Step Tutorial on DeepSeek Installation and How to Use the V3 Model

If you’re eager to try and install DeepSeek V3 on your system, this section shows you exactly how to download, install, and run it. Follow these steps carefully to get the best results.

Step 1: Visit DeepSeek’s Official Website and Click on ‘Start Now’

Once you’ve clicked on ‘Start Now’, here are the next steps:

• Choose Sign In or Register. You can create an account with your email, phone, or Google login.
• If you run into a busy server message, wait a bit and try again, as user interest is often high.

Step 2: Start Chatting

Once you’re logged in, you’ll land on a chat interface.

• In the text box, enter a prompt or question.
• Hit Enter, and DeepSeek V3 will respond in a conversational format.
• You can ask it to write code, solve math problems, or explain complicated topics. For best results, be specific.

Step 3: Try the Search Mode

• Start a new chat to keep your sessions organized.
• Look for a “Search” or “Internet” toggle if you want V3 to scan the web for up-to-date info.
• If it stalls, click ‘Continue’ until it finishes retrieving data.

Step 4: Use File Uploads

• Look for an Attach or Upload button/ icon in the chat interface.
• Drag and drop images or documents containing text you’d like summarized or interpreted.
• Ask V3 to convert them to text, create summaries, or answer questions about the content.

Step 5: Explore Advanced Settings: DeepThink (R1)

• Choose DeepThink 
• DeepThink prompts the AI to show more reasoning steps, which can help with detailed queries
• Keep an eye out for ways to rename or delete chats to stay organized

Step 6: Switch to the Mobile DeepSeek App (Optional)

• Click the 'Get App' icon at the bottom left, or download DeepSeek from the iOS or Android app store.
• Sign in with the same account you used on the website.
• Chat with V3 on the go, just like you do on the desktop site.

Getting Started With DeepSeek R1 Locally: A Step-by-Step Tutorial on DeepSeek Installation and How to Use the R1 Model

If you’d like to run and install DeepSeek R1 on your own computer without relying on online servers, the steps below will help you install and use the model in a private environment. By choosing a smaller checkpoint (like 1.5B) or a larger one (up to 671B), you can match the model size to the power of your hardware.

Step 1: Install Ollama

1. Go to the Ollama website: Choose the installer for your operating system.

2. Run the installer:

• On Windows: Double-click the downloaded file, then click through each screen until installation completes.
• On macOS: Open the DMG file, drag the Ollama app icon into your Applications folder, and confirm any prompts that appear.

3. Launch Ollama:

• On macOS, you might see a new icon (shaped like a llama) in your menu bar once it’s running.
• On Windows, the program window might open or minimize to the system tray.

Step 2: Open the Command Line for Ollama

1. Look for an “Install” or “Command Line Tools” option in the Ollama app interface. Click it to add Ollama commands to your system.
2. Confirm installation:

• On macOS: Open Terminal and type “ollama version”.
• On Windows: Open Command Prompt or PowerShell and do the same.
• If you see a version number, you’re set.

Step 3: Download and Run a DeepSeek R1 Model

1. Pick a model size: DeepSeek R1 is available from 1.5B parameters up to 671B.
2. Open your terminal or command prompt.
3. Install and run this command: ollama run deepseek-r1:1.5b

Please note: In the command above, replace 1.5b with 7b, 14b, 32b, 70b, or 671b if your hardware can handle a larger model. Ollama will download the required files and start DeepSeek R1 locally.

Step 4: Chat with DeepSeek R1

• Type a prompt right in the terminal window, then press Enter.
• Wait for DeepSeek R1 to respond. It may take some time, especially if you've chosen a bigger model or are on a CPU-only setup.
• Stop output at any point by pressing Control + C (on macOS) or Ctrl + C (on Windows).

Step 5: Run DeepSeek R1 as a Background Service

• Open a second terminal or command prompt window.
• Start the Ollama server using this command: ollama serve
• This keeps DeepSeek R1 running in the background, ready to answer API calls or power other apps on your system.

Step 5: Use the Local API

• Call the API from the script below or another terminal

curl http://localhost:11434/api/chat -d '{
  "model": "deepseek-r1",
  "messages": [{"role": "user", "content": "Hello! How are you?"}],
  "stream": false
}'

• Integrate DeepSeek R1 with any software that can send HTTP requests (like a custom app or web service).

Step 6: Explore Python Integration

• Install the Ollama Python library

pip install ollama

• Create a Python script

import ollama

response = ollama.chat(
    model="deepseek-r1",
    messages=[{"role": "user", "content": "Write a short poem on monsoon."}],
)

print(response["message"]["content"])

• Use this approach if you plan on building AI features into a larger Python project.

Step 7: Build a RAG App (Optional)

• Install extra packages

pip install langchain chromadb gradio
pip install -U langchain-community

• Create a Gradio interface that asks questions, loads PDFs, and retrieves relevant text chunks before passing them to DeepSeek R1.
• Run the app to see a local webpage where you can upload files and chat with R1 about their contents.

Step 8: Shut Down When Done

• Exit the chat by typing “/bye” if you’re in interactive mode.
• Stop the background service by closing the terminal or pressing Control + C again in the server session.

By following these steps, you get a full DeepSeek R1 setup. You’ll have privacy (no cloud storage) and a quick way to integrate R1 into your code or projects. Feel free to start small (1.5B parameters) and move to a larger version later if you need more power.

Pro Tips for Beginners Using and Installing DeepSeek for the First Time

Here are some quick pointers to get you comfortable with DeepSeek right from the start:

• Start Small: If you’re new to AI, pick a smaller model (1.5B or 7B) to avoid overloading your system.
• Keep an Eye on GPU Temperature: Running large models can heat up your GPU, so use a monitor tool (like nvidia-smi on Linux) to avoid overheating, especially if you’re in a warm environment.
• Try Quantization: Tools like Ollama or community scripts sometimes reduce model size to int8 or int4, lowering VRAM needs without losing much accuracy.
• Give Clear Prompts: The more detail you provide, the better the model’s replies will be.
• Take Breaks to Prevent Lags: If your system feels slow after a long session, pause DeepSeek, free up RAM, and relaunch.
• Double-Check Results: Even if DeepSeek’s replies look correct, always confirm math or code outputs since occasional errors can happen.
• Check Your Hardware: Confirm you have enough RAM and storage space to handle larger parameter sizes.
• Mind Your Privacy: Avoid typing personal data since any online usage can contribute to training.
• Use DeepThink Wisely: This feature shows how the AI is reasoning step by step, which helps with homework or coding.
• Plan Storage Wisely: Large checkpoint files can quickly eat up space. Back them up on an external drive or a high-capacity partition if you’re low on disk space.
• Stay Organized: Save your best prompts and solutions in a text file so you can refer back later and make each new session more effective.
• Experiment Gradually: Test advanced features like local RAG apps once you’re familiar with the basics.

Conclusion

DeepSeek gives you the freedom to explore artificial intelligence on your own terms, whether you stick to online chats or run the model locally for full privacy. DeepSeek V3 caters to broader conversations and coding support, while DeepSeek R1 shines with step-by-step reasoning.

Both models can handle a lot but require a fair amount of resources. If you’re short on hardware, don’t hesitate to try smaller model sizes first. Once you feel confident, you can push towards heavier checkpoints, tinker with advanced features like DeepThink or RAG, and even build your own AI systems and apps.

As you explore these possibilities, you might also be interested in taking your AI skills further. That’s where upGrad can come in — our AI and ML online courses dive into real-world applications and help you build on what you’ve learned here. For any further enquiries regarding your career path, book a call with upGrad’s experts for free.

Expand your expertise with the best resources available. Browse the programs below to find your ideal fit in Best Machine Learning and AI Courses Online.

Best Machine Learning and AI Courses Online

Discover in-demand Machine Learning skills to expand your expertise. Explore the programs below to find the perfect fit for your goals.

In-demand Machine Learning Skills

Discover popular AI and ML blogs and free courses to deepen your expertise. Explore the programs below to find your perfect fit.

Popular AI and ML Blogs & Free Courses

Reference Links:
https://timesofindia.indiatimes.com/technology/tech-tips/how-to-use-deepseek-on-laptop-system-requirements-steps-to-download-the-software-and-more/articleshow/117619851.cms
https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf
https://github.com/deepseek-ai/DeepSeek-V3