v3 voices are not currently designed for low-latency communication. So they probably won't be available until ElvenLabs releases v3 models designed for that