The GitHub entry for vLLM describes the software as: “A high-throughput and memory-efficient ... “All of this translates to ...