How to Run DeepSeek R1 Locally with Cursor in 4 Minutes (yes completely offline!)
DeepSeek R1 is a groundbreaking open-source AI model developed by DeepSeek AI that rivals the performance of industry leaders like OpenAI’s o1, particularly in reasoning capabilities.
DeepSeek R1 has demonstrated exceptional performance across various benchmarks in mathematics, coding, and logical reasoning tasks.
There are clear challenges associated with using OpenAI or Anthropic’s models on their platforms. By doing so, you’re essentially providing them with free — and sometimes proprietary — data to enhance their models.
Not everyone is comfortable with giving away their data, especially not for free. If you’re one of those individuals, read on for a quick guide on how to run ‘DeepSeek R1:1.5b’ locally, using only your CPU and without an internet connection.
Step 1: Make sure you have Cursor.AI installed. https://www.cursor.com/
Step 2: Install CodeGPT in Cursor
You need to enhance your Cursor environment with CodeGPT, an extension that brings AI assistance directly into your coding workflow. Here’s how you can do it:
- Navigate to extenstions tab in Cursor and search for CodeGPT
Step 3: Install Ollama
Next, you’ll need Ollama, which allows you to download the DeepSeek model.
- Navigate to https://ollama.com/ to download and install Ollama. This tool will be crucial for running the DeepSeek model locally.
Step 4: Accessing DeepSeek Models
With both CodeGPT and Ollama installed, you’re now ready.
- Open the CodeGPT extension in Cursor. You’ll find an interface where you can install various AI models.
- Hit select your AI
- Select ‘Ollama’
- And you should be giving a drop down list to find DeepSeek R1. I went for one which was trained on 1.5 b tokens which downloaded relatively quickly and was ready to prompt with straight away completely offline
In just a few minutes, you can run one of the most efficient AI models available today, right on your local machine for free. Enjoy
