I've been wondering about OpenAI's response quality throughout the day. It seems like the answers I get are noticeably poorer during the night (US time) and on weekends. Do they actually reduce their server capacity during these times to save costs? Or is the demand higher at those hours? I'm not criticizing them for optimizing; I just want to know if anyone has some insight on this, and if others feel the same way.
6 Answers
I really believe that cost-cutting measures are in place. OpenAI is spending money like there's no tomorrow!
Nah, I think it’s just that you might be tired and distracted when using it at those times.
Keep in mind that night for you is morning in Asia, where a lot of their users are active.
I think it's possible they're doing some updates during those times. I've noticed that on weekends when they aren't deploying anything, the service is usually better overall.
I thought it was just me! Every time I use any LLM like ChatGPT or Gemini in the evening, it feels like they dumb down significantly.
What exactly do you mean by 'worse'? Are you talking about slower responses or lower quality answers?
I mean lower quality! I feel like the response times might get better, but the answers seem rushed and less thoughtful. It’s all just my experience, really, but I notice more hallucinations and thinner reasoning.
Wow, that sounds a lot like how utilities function. Access to AI might really become a commodity!