Hey! We're working on a service that is highly dependant on OpenAI completion endpoint and we have noticed that the response time deterirates during evenings in Europe. Anyone else experience any issues with that? Clearly the timing aligns with US day time so my first assumption is that the higher load on GPT just makes it way slower