llama.cpp adds decision models via new /v1/systemone endpoint

Georgi Gerganov announced that decision models are now available in the latest llama.cpp builds through a new /v1/systemone endpoint. The endpoint supports local, private inference with multiple open models, and more are planned.