Inconsistent GPT-4.1 Mini API Performance Over Time

Lately, I’ve noticed that OpenAI has performance spikes — some days it’s very slow, and other days it works fine. Is anyone else experiencing this? It really impacts the user experience at the API level. With the GPT-4.1 mini model, I can clearly feel the difference: on good days, it might take around 20 seconds to respond with function calls, while on bad days, it can take 80 seconds or more.

2 Likes