Best Settings for conversational voice agent
# support
b
I am looking to deploy a voice agent to call potential home buyers and sellers, and I am looking to start testing with what is the current best model, and voice model. I see a ton of options on VAPI - initially I am not worried about cost per minute as it is all cheaper than our current costs. What should I start with for testing? Thank you!
a
Hi @bentheguy Start your testing with: LLM: OpenAI GPT-4o – best for natural, low-latency cold call conversations. (You can compare later with Anthropic Claude 3.5 Sonnet for longer qualification calls.) Voice: ElevenLabs Rachel (seller outreach) or Adam (buyer/investor tone). A good baseline is GPT-4o + Rachel to start. This combo works very well for outbound real estate testing before worrying about cost.
b
thank you!
c
Hi, For testing, here's what I'd start with: Model: GPT-4o - it handles longer, nuanced conversations well which is exactly what you need for buyers and sellers. Voice: ElevenLabs - best quality voices on the market, very natural sounding. You can browse their voice library and pick one that fits the tone you're going for. This combo will get you up and running quickly and gives you a solid baseline to test from. Once you've done a few calls you'll have a much better sense of what to tweak.