Mistral AI

Voxtral Small 24B

Mistral's audio model: it understands speech

Text

Audio

Tools

Type: Chat
Context: 32K tokens
Russian: medium
Input price: 8 ₽ / 1M tokens
Output price: 24 ₽ / 1M tokens
Release date: October 2025

chat/completions

import OpenAI from "openai"
const client = new OpenAI({
  baseURL: "https://api.yenisei.ru/v1",
  apiKey: process.env.YENISEI_API_KEY,
})
const stream = await client.chat.completions.create({
  model: "voxtral-small",
  messages: [{ role: "user", content: "Hi!" }],
  stream: true,
})
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "")
}

About the model

Voxtral Small 24B is Mistral's model for audio and speech. It recognizes spoken language, answers questions about an audio recording and handles ordinary text chat. The model transcribes speech accurately, and you can hold a conversation about a recording. It is multilingual and handles Russian confidently, which is convenient for voice products in different languages. It follows instructions well and suits voice assistants and audio processing pipelines.