Anyone else noticed a significant decrease in the speed of all GPT-4 services, like me?