MLCommons released a fresh set of tests and results that rate the speed at which top-of-the-line hardware can run AI applications and respond to users. The results roughly demonstrate how quickly an AI application such as ChatGPT can deliver a response to a user query.