When we talk about running local AI these days, the conversation usually either revolves around mini-PCs like the RTX Spark or drifts into wistful thinking about home servers and ludicrously expensive ...
28.9M-parameter model runs on a microcontroller costing less than $10 ...