Running a 28.9M Parameter LLM on an $8 ESP32 Microcontroller

boveyking · hn · 2026-07-26

A developer demonstrated how to run a 28.9M parameter Large Language Model (LLM) locally on an ESP32, an $8 microcontroller.

The project provides a complete code implementation, pushing past the strict memory and compute limitations of low-end hardware. It offers a highly valuable, ultra-low-cost solution for offline deployment in IoT devices and edge AI applications.

Original post →

More from Infra

Infra channel →