Developer runs a 28.9M-parameter LLM on an $8 ESP32-S3 chip offline
joemeno · x · 2026-07-25
A developer has forced a 28.9-million-parameter language model onto a standard ESP32-S3 microcontroller that costs about $8.
- The model runs fully offline and produces coherent short stories at around 9.5 tokens per second.
- The chip has only 512 KB of fast SRAM and 16 MB of flash, so the key trick is architectural: most of the embedding table was moved into flash and memory-mapped.
- The runtime only needs to pull about 450 bytes per token, keeping active working memory in SRAM.
- The result is a capable local language model on hardware that draws roughly the power of a single LED.
- The thread also points to early physical-device experiments, including a minimalist voice-controlled universal remote.
Related event: $8 ESP32-S3 Microcontroller Runs LLM Offline(4 posts)→
More from coding & agent
- Looking for a classifier of software engineering task shapes to pick models per task — StewartalsopIII · 2026-09-11
- Steal this idea: prompt-to-hardware where agents assemble custom devices — paraschopra · 2026-09-11
- Model Is the Least Interesting Part: A Guide to Six Core AI Architectures from RAG to Multi-Agent — goyalshaliniuk · 2026-09-11
- Non-coder builds layered memory architecture: 20k tokens tracks a year of agent conversations — matteoianni · 2026-09-11
- Warp's six non-engineering teams all run on Linear and Claude Code — mon__lim · 2026-09-11
- 9-year backend dev: AI code isn't the problem, the rate of making a mess is — Sweaty-Landscape-561 · 2026-09-11