I Ran Deepseek R1 on Raspberry Pi 5 and No, it Wasn't 200 tokens/s

DeepSeek AI was tested on Raspberry Pi 5 with various model sizes. The 1.5B model performed well for basic tasks, while 7B and 8B models were slower and less accurate. The 14B model couldn't run due to memory constraints. Tests included queries about Docker and generating HTML/CSS code. The experiment demonstrated the feasibility of running AI models on single-board computers, with the 1.5B model being most suitable for the Pi 5.