Run DeepSeek R1 Locally: Your Private AI Chatbot

Bryan Wilks
5 hours ago
3 min read

Want to run a powerful AI model like DeepSeek R1, which rivals OpenAI's GPT-4, right on your own computer? And keep it private? You can! This guide breaks down how to get it set up, making it as simple as possible.

Key Takeaways

Local Control: Run advanced AI models like DeepSeek R1 entirely on your machine, ensuring privacy.
Cost-Effective: Avoid hefty monthly subscription fees for premium AI services.
User-Friendly Interface: Utilize OpenWebUI for a chat experience similar to ChatGPT.
Model Flexibility: Easily switch between different AI models, including DeepSeek R1 and Llama 3.3, within the OpenWebUI.
Hardware Matters: Performance depends on your computer's specs, especially the GPU, with larger models requiring more power.

Step 1: Install Ollama

The first step is to install Ollama. You can download it from ollama.com. It works on both Mac and PC. Just hit the download button and choose your operating system. If you're on Windows, make sure it's Windows 10 or later.

Once installed, Ollama will give you a command-line prompt. You can run your first model here. For example, you can install Llama 3.3 by typing into your terminal. This is a good way to test if your system can handle these models, though it's not the DeepSeek R1 model itself.

Step 2: Install DeepSeek R1 Model

Now, let's get the DeepSeek R1 model. Go back to the Ollama website and find the models section. You'll see DeepSeek R1 listed there. It comes in different sizes, from 1.5 billion parameters (1GB) to much larger ones. For most users, starting with the 7 billion parameter model (around 4.7GB) is a good balance.

To install it, you'll use the terminal again. The command is . If you want to try a different version, like the 14 billion parameter model, you'd use . Just copy the command from the Ollama website and paste it into your terminal. It will download the model, which might take a minute or two depending on your internet speed.

Once downloaded, you can technically interact with the model directly in the terminal. Just type your prompt after the model responds. However, this isn't the most user-friendly way to chat.

Step 3: Install Docker

To get a nice, chat-like interface, we need a couple more things. First is Docker. Head over to docker.com and download the application for your computer. Install it, and then just make sure it's running in the background. You don't need to do anything else with it; just having it open and minimized is enough.

Step 4: Install OpenWebUI

The final piece is OpenWebUI. This is what will give you that familiar chat interface. Go to docs.openwebui.com. If you already have Ollama installed, you'll see a command to run. Copy this command and paste it into a new terminal window. This command sets up OpenWebUI to work with your installed models.

After running the command, you might see a new container listed in your Docker application. Open your web browser and go to . You'll need to create an account for OpenWebUI the first time you access it.

Using DeepSeek R1 with OpenWebUI

Once you're logged into OpenWebUI, you can select the AI model you want to use. Look for the dropdown menu at the top. You should see DeepSeek R1 listed there, along with any other models you've installed, like Llama 3.3. You can set DeepSeek R1 as your default model if you plan to use it most often.

Now you can start chatting! Try a prompt like, "Give me 10 challenges only a reasoning AI model can solve." You'll notice the speed is quite good, especially with the 7B model, because it's running locally. The model will show its thinking process in brackets before giving you the answer.

One cool feature of OpenWebUI is the ability to run multiple models side-by-side. You can add another model, like Llama 3.3, and ask both models the same question. This lets you compare their responses directly. Keep in mind that larger models, like the 32GB or 70GB versions of DeepSeek R1, will require more powerful hardware, particularly a good GPU, to run smoothly.

This setup gives you a private, powerful AI chatbot right on your computer, without any ongoing costs. You can even install larger versions of DeepSeek R1 by simply finding the new command on the Ollama website and running it in your terminal.