Section 05

Articles

Analysis and commentary on Edge AI hardware and open models — measured, opinionated, and shorter than a tutorial. This is also where the multi-hardware, multi-model experiments live: the ones worth publishing that never grow into a full series.

29 September 2026
Three Boards, One Question: Where Should a Small Language Model Run?
Orange Pi Zero 3W, Raspberry Pi 5, and Arduino UNO Q with the same models, the same llama.cpp build, and the same protocol, from chat and captions to agents Read →
7 September 2026
Christmas 1970, Rebuilt in a Browser
All 22 circuits of the Philips Engenheiro Eletrônico EE 20, simulated, bilingual and free Read →
27 August 2026
Open and Local Is Not a Compromise Anymore
What the August 2026 new Macs and open-weight models actually let you run Read →
26 July 2026
Running Small Language Models on a Raspberry Pi 5
Gemma 4 E2B and Qwen3.5 4B with multi-token prediction — how close you can get to the hardware's limit, and what you give up at each setting Read →

← Back to the hub

Articles

Three Boards, One Question: Where Should a Small Language Model Run?

Christmas 1970, Rebuilt in a Browser

Open and Local Is Not a Compromise Anymore

Running Small Language Models on a Raspberry Pi 5