When are other models besides speech 2.0 turbo going to be available for Minimax? and does changing the api model through the vapi api make it use other models or does Vapi have to add specific support for them?