OpenAI engineers

“OpenAI engineers told colleagues earlier this month that they'd managed to cut inference costs—the expense of running existing AI models—by more than half.”