LiveKit - Phaseo Docs

Point the LiveKit Agents OpenAI plugin’s LLM at https://api.phaseo.app/v1 with your Phaseo key, and your voice agents use Phaseo model IDs and routing. The plugin calls Phaseo’s Chat Completions endpoint.

Install

uv add "livekit-agents[openai]"

pnpm add @livekit/agents @livekit/agents-plugin-openai

Point the agent’s LLM at Phaseo

import os
from livekit.agents import AgentSession
from livekit.plugins import openai

session = AgentSession(
    llm=openai.LLM(
        model="openai/gpt-6-astra",
        api_key=os.environ["PHASEO_API_KEY"],
        base_url="https://api.phaseo.app/v1",
    ),
)

import { voice } from "@livekit/agents";
import * as openai from "@livekit/agents-plugin-openai";

const session = new voice.AgentSession({
	llm: new openai.LLM({
		model: "openai/gpt-6-astra",
		apiKey: process.env.PHASEO_API_KEY,
		baseURL: "https://api.phaseo.app/v1",
	}),
});

Start a session, say something, and confirm the request appears in Settings → Usage → Logs.

Known limits

Use the plugin’s Chat Completions mode for custom endpoints. LiveKit’s Responses-specific OpenAI tools don’t become Phaseo server tools.
This only routes the LLM. Configure speech-to-text and text-to-speech separately, unless the LiveKit interface you use accepts a custom OpenAI-compatible URL for that modality.
Test interruption handling, tool calls, and streaming latency before production voice traffic.

Last modified on November 5, 2026