Field notes from the desk where the models actually run: local LLMs, the gear that feeds them, and the 3D printer humming in the corner. No filler, no hype. If it ships, I ran it.
Colibrì is a pure-C, zero-dependency inference engine that runs GLM-5.2, a 744B MoE model, on a consumer machine with 25GB of RAM, by keeping dense weights in RAM and streaming routed experts from disk.
read the story →⚡ Signal over noise
Apple's trade secret suit names OpenAI's hardware chief and claims candidates were told to bring actual Apple parts to job interviews. The AI phone might be built on Cupertino's homework.
Joshua Bird's custom clay extruder uses an air compressor and stainless steel auger to print pottery. The process works, but shrinkage, abrasion, and a two-fire ceramic workflow make it genuinely hard.
Build b9963 lands proper multi-tile support for DeepSeek-OCR v1 with dynamic resolution, and unifies the image preprocessor for both v1 and v2. Multi-column documents now handled correctly.
