MLCommons has announced the release of the MLPerf Inference v5.0 benchmark suite, which saw participation from 23 organizations and yielded over 17,000 performance results. The new benchmark suite introduces advancements in benchmarking large language models (LLMs), including two new benchmarks: the Llama 3.1 405B Instruct benchmark and an updated Llama 2 Chat 70B Interactive benchmark. Additionally, the v5.0 release includes a new Graph Neural Network (GNN) benchmark, aimed at providing a standardized method for evaluating the performance of GNN implementations and hardware accelerators. These updates reflect the increasing demand for AI applications, which require models to respond to a greater number of queries, such as those used in chatbots and search engines.