Could we use the new GPT-4o-2024-11-20 in VAPI? it seems to be way faster: https://x.com/sawyerhood/status/1859357797781209294?_bhlid=62ae56157034e7057c9117865ecc31b1fb75fc17