Whats the deal with the sudden latency increase? The openai models' latency seems to have greatly increased in vapi. 4.1 mini even has a higher latency than 4o. Is it just me or is this a new thing? Tried searching for a good answer and couldn't find one.