this was raised during office hours yesterday. Nothing could be officially stated on the matter. It will require a bit of R&D on their part before they can confidently assure us responsibly. This is not already an option that i've seen, which is logical as it takes some labor hours to sort out and we're still within one day of the openai announcement However its easy to see how the VAPI team would want to incorporate such functionality. Anytime you dont have to stream data to/from random vendor colos then you're gonna shave off a couple hundred milliseconds. Since the geographical distances involved in any particular tech stack are the biggest factor in latency, its without doubt safe to say that if its possible and if it works well then the VAPI team would want to take advantage of making that available to their users