Thanks — I’m responding to your question about using the new

gemini-2.5-flash-native-audio-preview-12-2025

live model with an external TTS provider. Yes — you can. Configure the model to return text-only by setting its

modalities

to

TEXT

, and then route that text to your external TTS (e.g., Cartesia, ElevenLabs, Google Cloud TTS). Example snippets: • Python

from livekit.plugins import google
from google.genai.types import Modality

session = AgentSession(
    llm=google.realtime.RealtimeModel(
        model="gemini-2.5-flash-native-audio-preview-12-2025",
        modalities=[Modality.TEXT],
    ),
    tts="cartesia/sonic-3",  # example external TTS
)

• Node.js

import * as google from '@livekit/agents-plugin-google';

const session = new voice.AgentSession({
  llm: new google.realtime.RealtimeModel({
    model: "gemini-2.5-flash-native-audio-preview-12-2025",
    modalities: [google.types.Modality.TEXT],
  }),
  tts: "cartesia/sonic-3", // example external TTS
});

If you’d like, tell me which TTS provider you plan to use and whether you need help with the integration steps (auth, format conversions, or streaming vs. file-based output). Sources: Gemini Live API plugin | LiveKit docs