How to Run DeepSeek AI on Raspberry PI 5 – Step by Step Guide

Components

• Raspberry 5 (I use 2GB RAM , Higher the RAM may improve the performance )
• 2A or More power supply ( I use MI 3A/5V phone charger )
• 32 GB SD Card
• USB-UART cable /Ethernet Cable( you can buy from local shops / from aliexpress )
• Linux( I use https://linuxmint.com/) or Windows PC

Steps

1. Download the latest raspberry PI Image & Burn it in the SD Card. Configure the Wifi settings / SSH and user Name /Password https://www.raspberrypi.com/software/

sudo apt install rpi-imager

2. Download and compile llama.cpp

git clone https://github.com/ggerganov/llama.cpp
$ cd llama.cpp
install cmake
sudo apt install curl libcurl4-openssl-dev
cmake -B build
cmake --build build --config Release

3. Download deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B (copy below to ‘pyhug.py’ and run)

from huggingface_hub import snapshot_download 
snapshot_download(repo_id="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",local_dir="DeepSeek-R1-GGUF2",local_dir_use_symlinks=False,revision="main")

4. Convert the model.safetensors file to gguf file

python3 convert_hf_to_gguf.py "/path/DeepSeek-R1-GGUF2" --outfile deep.gguf

5.Connect to Raspberry PI using UART or Network

screen /dev/ttyUSB0 115200

6. Run llama

./llama-cli --model /home/admin/Bookshelf/deep.gguf