Run Deepseek R-1 on your NVIDIA-powered Windows laptop

## All-in-one script, run on WSL

Please read through at least once to understand what it is trying to do before running it.

#!/bin/bash

# Exit script on error
set -e

# Ensure curl is installed
sudo apt update && sudo apt install -y curl

# Install Ollama
if ! command -v ollama &> /dev/null
then
    curl -fsSL https://ollama.com/install.sh | sh
fi

# Download DeepSeek R-1 model
ollama run deepseek-r1:7b -d

# Install uv
if ! command -v uv &> /dev/null
then
    curl -LsSf https://astral.sh/uv/install.sh | sh
fi

# Install and start Open Web UI using Docker
if ! command -v docker &> /dev/null
then
    echo "Docker is not installed. Please install Docker manually."
    exit 1
fi

docker run -d --network=host -v open-webui:/app/backend/data \
    -e OLLAMA_BASE_URL=http://127.0.0.1:11434 \
    --name open-webui --restart always \
    ghcr.io/open-webui/open-webui:main

# Output success message
echo "Setup complete! Open your browser and visit http://localhost:8080 to start chatting."

Prerequisite: Install WSL 2

Reference: https://learn.microsoft.com/en-us/windows/wsl/install

Open your Command Prompt and run this command to have WSL installed

wsl --install

Step 1: Install Ollama on WSL

Reference: https://ollama.com/download/linux

Wait for the WSL installation to complete.

Run the following command on your WSL console:

curl -fsSL https://ollama.com/install.sh | sh

Step 2: Install Deepseek R-1 model

Now you can install your favorite Deepseek R-1 model.

They are listed here: Ollama’s Deepseek R-1

I used the 8B params version on my 3070ti and it works nice.

You should test out different versions to see which works for you.

Run the following command on your WSL console:

ollama run deepseek-r1:7b

Step 3: Install uv

Reference: https://docs.astral.sh/uv/getting-started/installation/

You may have to install curl using sudo apt install curl

Then run this command to install uv

curl -LsSf https://astral.sh/uv/install.sh | sh

Step 4: Install Open Web UI

Reference: https://github.com/open-webui/open-webui

Make sure to check out their troubleshooting section.

Make sure there is no process listening on port 8080 on the windows interface.

Run the following command on WSL:

docker run -d --network=host -v open-webui:/app/backend/data -e OLLAMA_BASE_URL=http://127.0.0.1:11434 --name open-webui --restart always ghcr.io/open-webui/open-webui:main

Step 5: Start chatting!

Go to http://localhost:8080/ on your windows browser, setup a locally-saved account and start chatting with the model.

Have fun!