Splash: a local LLM inference stack built for each model
The article remains in its original language.