Last released Oct 2, 2026
Give an open LLM a fast, calibrated decision path and serve it natively with vLLM.