OpenAILive
OpenAI GPT-Live full-duplex voice model that listens and speaks at the same time. Choose it for the most natural, interruptible speech-to-speech conversations. Authenticates with OPENAI_API_KEY from your environment, or an explicitapi_key.
Fields
str
default:"UNSET"
GPT-Live model to use, e.g. “gpt-live-1”. Defaults to ‘gpt-live-1’.
str | dict[str, str]
default:"UNSET"
Output voice, e.g. “marin”, “cinder” or “vesper”, or {“id”: “voice_…”} for an authorized custom voice. Fixed for the session. Defaults to ‘marin’.
Mapping[str, Any] | None
default:"UNSET"
Where tool calls run, fixed for the session: an OpenAI-hosted Responses model, or an LLM you supply that runs them on your side. Defaults to None (not set).
str | None
default:"UNSET"
Custom OpenAI API base URL, e.g. for a proxy. Defaults to None (not set).