TILens turns technical updates into a focused daily brief: official releases,
trusted reporting, and practitioner analysis, deduplicated and organized by topic.
GPU LLM inference on Apple Silicon, packaged as one Java 25 executable JAR, zero dependencies. lightmetal binds a Metal-enabled libllama.dylib through the Foreign Function & Memory API and runs Mistral- and…
Six months ago we shipped the alpha with a promise - a framework-agnostic, provider-agnostic AI toolkit with no lock-in, ever. Today TanStack AI hits Beta with every modality, a hardened protocol, middleware,…