I have several assistants based on GPT-4.1 Mini, and I've noticed that since this afternoon, the model's latency has increased from 400ms to 770ms. Why is that? The latency of almost all models has been increased significantly: what happened? My assistants went from 900ms of total latency time to 1370ms!

Hey fillipio, I couldn't find enough details to fully answer your question, but I've escalated it to a human agent who can help you out!

To reduce GPT-4.1 Mini latency, shorten prompts, enable streaming if possible..

Ragionato per un paio di secondi We’ve been talking about assistants that have been in production for weeks, and no changes have been made to them. I don’t think it’s a prompt issue, but rather an internal one, and as far as I know OpenAI hasn’t modified the latencies of its models.

I've noticed the same thing.

OpenAI models were never stable in their response times, bigger overall load - longer response times. That's why I never used them for voice AI myself. Just ran my test that does 50 requests to each model/provider and openai has largest stdev https://cdn.discordapp.com/attachments/1369433726294949888/1369569790603034714/image.png?ex=681c56c3&is=681b0543&hm=7461f991d0d522c246b435bff25d828d8daff80c2202b2b26fa8d61a74b471a0&

Hey Fillipio, checking if this is resolved for you.