Ternlight brings 7 MB embedding model to the browser via WebAssembly

Ternlight is a compact embedding model that runs entirely in-browser using WebAssembly, eliminating the need for backend inference calls. For indie developers building AI features, this means lower latency, zero server costs for embeddings, and offline-capable applications—all with a model small enough to load instantly.