Will vapi support any voice mode models like this or realtime voice from openai? Or are they essentially competing solutions