@livekit/agents-plugin-mistralai

Mistral AI plugin for LiveKit Agents

The Agents Framework is designed for building realtime, programmable participants that run on servers. Use it to create conversational, multi-modal voice agents that can see, hear, and understand.

This package contains the Mistral AI plugin, providing LLM, STT, and TTS capabilities via the official @mistralai/mistralai SDK.

Installation

npm install @livekit/agents-plugin-mistralai

Usage

LLM

Uses the Mistral Conversations API.

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

const llm = new mistral.LLM({
  model: 'mistral-small-latest',
  // apiKey defaults to process.env.MISTRAL_API_KEY
});

STT

Supports both batch transcription and realtime streaming (WebSocket).

// Batch transcription
const stt = new mistral.STT({ model: 'voxtral-mini-latest' });

// Realtime streaming (requires a VAD)
const stt = new mistral.STT({
  model: 'voxtral-mini-transcribe-realtime-2602',
});

Realtime models require a VAD for endpointing. If none is provided, Silero VAD is loaded automatically (install @livekit/agents-plugin-silero).

TTS

Text-to-speech with voice presets or reference audio for voice cloning.

const tts = new mistral.TTS({
  model: 'voxtral-mini-tts-latest',
  voice: 'en_paul_neutral',
});

Environment variables

• MISTRAL_API_KEY ; Your Mistral API key (used by all components unless apiKey or client is passed explicitly)

Supported models

See models.ts for the full list of chat, STT, TTS, and voice models.