GLM-Realtime

Overview

Best for: Chinese-market voice or video-call assistants with short interactions and predictable per-minute cost.

At a glance

Price is glm-realtime-flash audio, 0.18 CNY/min converted at 7.1 CNY per USD (air 0.3 CNY/min; video 1.2 to 2.1 CNY/min). Concurrency 5 at account tier V0, up to 20 at V3. Context 8K for audio (about 2 min memory), 32K for video. China platform only.

Features

• server VAD or client VAD
• interruption (interrupt_response, response.cancel)
• function calling (voice calls only)
• built-in web search (auto_search)
• passive video mode (video_passive)
• near/far-field noise reduction
• greeting config
• singing

Pricing

Free tier: Not stated on the GLM-Realtime page.

Source: docs.bigmodel.cn

Setup

1. Register on open.bigmodel.cn and create an API key (real-name verification may be required).
2. Connect to the realtime WebSocket with Authorization: Bearer <key> (a JWT also works).
3. Send session.update with session.model set to glm-realtime-flash or glm-realtime-air.
4. Stream input_audio_buffer.append (pcm16) and play response.audio.delta (24 kHz PCM).

Endpoint

wss://open.bigmodel.cn/api/paas/v4/realtime

Authentication

Authorization header with API key or JWT

// npm i ws
import WebSocket from "ws";

const ws = new WebSocket("wss://open.bigmodel.cn/api/paas/v4/realtime", {
  headers: { Authorization: `Bearer ${process.env.ZHIPU_API_KEY}` },
});

ws.on("open", () => {
  ws.send(JSON.stringify({
    type: "session.update",
    session: {
      model: "glm-realtime-flash",        // or glm-realtime-air
      modalities: ["text", "audio"],
      instructions: "You are a helpful voice assistant.",
      voice: "tongtong",
      input_audio_format: "pcm16",        // 16 kHz mono 16-bit
      output_audio_format: "pcm",         // 24 kHz mono 16-bit
      beta_fields: { chat_mode: "audio" },
    },
  }));
});

// ~100 ms frames, at most 50 messages per second
export function sendPcm(buf) {
  ws.send(JSON.stringify({ type: "input_audio_buffer.append", audio: buf.toString("base64") }));
}

ws.on("message", (raw) => {
  const ev = JSON.parse(raw.toString());
  if (ev.type === "response.audio.delta") play(Buffer.from(ev.delta, "base64"));
  if (ev.type === "error") console.error(ev);
});
function play(pcm) {}

Written from the current docs. Check the vendor's SDK version before you ship.

Warnings

Plus 14 warnings that apply to all voice-to-voice APIs. See category warnings.

Limits

• Context 8K for audio calls (about 20 turns per docs) and 32K for video calls
• Conversation memory up to about 2 minutes
• max_response_output_tokens up to 1024
• Client VAD mode: max 30 s per upload; send at most 50 messages per second (100 ms frames recommended)
• Concurrency by account tier: V0 5, V1 10, V2 15, V3 20

Models and products

Docs and sources

Docs

• GLM-Realtime guide

Sources used

• docs.bigmodel.cn/cn/guide/models/sound-and-video/glm-realtime