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

Melvin Vivas · X post · 2026-08-02 · Open on X

Topics: LLM Fundamentals, Industry Trends & Job Market · Level: intermediate

Summary

A developer used a trick borrowed from Google's Gemma models to fit a 28.9M-parameter language model on an $8 ESP32 microcontroller. It shows that very small LLMs can run on very cheap hardware at the edge. The linked Hackster.io article has the details.

Key points

• A 28.9M-parameter LLM was made to run on an ESP32 chip that costs about $8.
• The approach borrows a technique from Google's Gemma model family.
• Tiny LLMs can run on microcontrollers, without a GPU or the cloud.
• Read the Hackster.io write-up for how it was done.

Resources mentioned

• Running a 28.9M Parameter LLM on an $8 Microcontroller (Hackster.io) · article · hackster.io · free
  Hackster.io news article about fitting a 28.9M-parameter LLM onto an ESP32 using a trick from Gemma.
• ESP32 · tool · espressif.com · paid
  A low-cost microcontroller from Espressif, used here to run a tiny LLM.
• Gemma 4 · tool · ai.google.dev · free
  Google's family of open-weight models in several sizes, built to run on devices and offline, with multimodal and agentic abilities, and open to fine-tuning.
  Also in: Gemma 4 Runs Locally On-Device in the Antigravity SDK (Melvin Vivas on X · notes), On-Device AI: Running Gemma 4 E2B Offline on an iPhone with LiteRT (Melvin Vivas on X · notes), Running Gemma 4 Models Offline on an iPhone (Melvin Vivas on X · notes), Fine-tuning Gemma4-E2B on your own tweet style with Unsloth (Melvin Vivas on X · notes) and 25 more

Try this

• Read the Hackster.io article to learn the technique used.
• Run a tiny language model on a cheap microcontroller such as an ESP32.