API Reference

Gemini Live WebSocket

GET https://api.tu-zi.com/v1beta/live?model={model}

Preview guide

Paste into an AI assistant to help with integration. Documentation and examples only; no real credentials.

# Gemini Live WebSocket

Gemini Live WebSocket

## Endpoint and authentication

`GET https://api.tu-zi.com/v1beta/live?model={model}`

Transport: websocket

Use an inference API key. Examples contain placeholders; read credentials from environment variables.

```http
Authorization: Bearer YOUR_API_KEY
```

## Query parameters

| Field | Type | Required | Description / constraints |
| --- | --- | --- | --- |
| model | string | Yes | Selected model identifier. Example: YOUR_MODEL |

## Minimal request example

```json
{
  "setup": {
    "model": "models/YOUR_MODEL",
    "generationConfig": {
      "responseModalities": [
        "AUDIO"
      ]
    }
  }
}
```

## Before running

Replace https://api.tu-zi.com with this site origin (do not append /v1 twice), and YOUR_MODEL with the exact marketplace model ID. Install requests for Python; use Node.js 20+ and save JavaScript as .mjs. WebSocket samples list their own dependencies.

Create an inference key on the site and set API_KEY in the environment; replace YOUR_API_KEY in cURL. Prepare local input.png/input.wav files for upload examples.

Start with a minimal request, then read the result using the response fields below. For async APIs, a returned task ID requires polling; acceptance does not mean generation has finished.

## Code examples

### curl

```bash
# Gemini Live requires a WebSocket client.
# See the Python or JavaScript example and /docs/api/gemini.
```

### javascript

```javascript
// Node.js: npm install ws
import WebSocket from 'ws';

const ws = new WebSocket("wss://api.tu-zi.com/v1beta/live?model=YOUR_MODEL", {
  headers: { Authorization: "Bearer " + process.env.API_KEY },
  perMessageDeflate: false,
  maxPayload: 2 * 1024 * 1024,
});
let completed = false, sent = false, receivedOutput = false, failed = false;
function fail(error) {
  if (failed) return;
  failed = true;
  console.error(error);
  process.exitCode = 1;
  ws.terminate();
}
const timeout = setTimeout(() => fail(new Error('Timed out before a complete turn')), 120000);
ws.on('close', () => {
  clearTimeout(timeout);
  if (!completed && !failed) fail(new Error('Connection closed before a complete turn'));
});
ws.on('error', fail);
ws.on('open', () => ws.send(JSON.stringify({
  "setup": {
    "model": "models/YOUR_MODEL",
    "generationConfig": {
      "responseModalities": [
        "AUDIO"
      ]
    }
  }
})));
ws.on('message', (message) => {
  if (failed || completed) return;
  try {
    const event = JSON.parse(message.toString());
    if (event.error) throw new Error(JSON.stringify(event.error));
    if (event.setupComplete && !sent) {
      sent = true;
      ws.send(JSON.stringify({"clientContent":{"turns":[{"role":"user","parts":[{"text":"Say hello briefly."}]}],"turnComplete":true}}));
    }
    const content = event.serverContent || {};
    if (content.interrupted) throw new Error('Turn was interrupted; no complete result');
    const text = content.outputTranscription?.text;
    if (text) { receivedOutput = true; process.stdout.write(text); }
    // Consume content.modelTurn.parts[].inlineData for PCM audio.
    if (content.modelTurn?.parts?.some(part => part.inlineData?.data || part.text)) receivedOutput = true;
    if (content.turnComplete) {
      if (!sent || !receivedOutput) throw new Error('Turn ended without model output');
      completed = true;
      console.log(event.usageMetadata);
      ws.close();
    }
  } catch (error) { fail(error); }
});
```

### python

```python
# Python 3.11+: pip install 'websockets>=14,<16'
import asyncio, json, os
from websockets.asyncio.client import connect

async def main():
    async with asyncio.timeout(120):
        async with connect("wss://api.tu-zi.com/v1beta/live?model=YOUR_MODEL", additional_headers={
            'Authorization': "Bearer " + os.environ["API_KEY"]
        }, max_size=2*1024*1024, max_queue=4, compression=None) as ws:
            completed = sent = received_output = False
            await ws.send("{\"setup\":{\"model\":\"models/YOUR_MODEL\",\"generationConfig\":{\"responseModalities\":[\"AUDIO\"]}}}")
            async for message in ws:
                event = json.loads(message)
                if 'error' in event:
                    raise RuntimeError(event['error'])
                if 'setupComplete' in event and not sent:
                    sent = True
                    await ws.send("{\"clientContent\":{\"turns\":[{\"role\":\"user\",\"parts\":[{\"text\":\"Say hello briefly.\"}]}],\"turnComplete\":true}}")
                content = event.get('serverContent', {})
                if content.get('interrupted'):
                    raise RuntimeError('Turn was interrupted; no complete result')
                text = content.get('outputTranscription', {}).get('text', '')
                if text:
                    received_output = True
                    print(text, end='', flush=True)
                # Consume content.modelTurn.parts[].inlineData for PCM audio.
                if any(part.get('inlineData', {}).get('data') or part.get('text') for part in content.get('modelTurn', {}).get('parts', [])):
                    received_output = True
                if content.get('turnComplete'):
                    if not sent or not received_output:
                        raise RuntimeError('Turn ended without model output')
                    completed = True
                    print(event.get('usageMetadata', {}))
                    break
            if not completed:
                raise RuntimeError('Connection closed before a complete turn')

asyncio.run(main())
```

## Response fields

| Field | Type | Description / constraints |
| --- | --- | --- |
| setupComplete | object | Setup-complete event. |
| serverContent.modelTurn.parts[] | object | Model audio or other content parts. |
| serverContent.turnComplete | boolean | Turn completion. |

Illustrative response; some endpoints show essential fields only. Optional fields may be absent.

```json
{
  "setupComplete": {}
}
```

## Streaming

Connect with wss, send setup, then realtimeInput or clientContent. Handle audio/transcription events, turnComplete, interrupted and errors. The example sends one request after setupComplete and requires model output followed by turnComplete. Errors, timeout, premature close or interrupted mean incomplete; do not automatically resubmit.

## Errors and retries

Check HTTP status and protocol-specific errors; do not mix error formats across protocols. For 401/403 check credentials and permissions, for 429 back off, and for 5xx check gateway/upstream status. Error bodies may contain error/message/code depending on the protocol.

Use bounded exponential backoff for reads. Do not blindly retry task creation, generation, or key mutations, to avoid duplicate billing or execution.

## Constraints

This page describes the base protocol. Use the selected model quick start for supported fields, enums, limits and asynchronous behavior. Replace YOUR_MODEL. Form initial values are not upstream defaults.

## Documentation source

Definition version: 1

Reviewed: 2026-11-02

Source: https://api.tu-zi.com/en/docs/api/reference-relay-gemini-live

Related protocol guide: https://api.tu-zi.com/docs/api/gemini

This export contains documentation and examples only, without real credentials, account data or user requests.

Updated 2026-11-02 · Reviewed 2026-11-02

Endpoint and authentication

GET https://api.tu-zi.com/v1beta/live?model={model}

Transport: websocket

Use an inference API key. Examples contain placeholders; read credentials from environment variables.

Authorization: Bearer YOUR_API_KEY

Minimal request example

{
  "setup": {
    "model": "models/YOUR_MODEL",
    "generationConfig": {
      "responseModalities": [
        "AUDIO"
      ]
    }
  }
}

Before running

Replace https://api.tu-zi.com with this site origin (do not append /v1 twice), and YOUR_MODEL with the exact marketplace model ID. Install requests for Python; use Node.js 20+ and save JavaScript as .mjs. WebSocket samples list their own dependencies.

Create an inference key on the site and set API_KEY in the environment; replace YOUR_API_KEY in cURL. Prepare local input.png/input.wav files for upload examples.

Start with a minimal request, then read the result using the response fields below. For async APIs, a returned task ID requires polling; acceptance does not mean generation has finished.

Constraints

This page describes the base protocol. Use the selected model quick start for supported fields, enums, limits and asynchronous behavior. Replace YOUR_MODEL. Form initial values are not upstream defaults.

Query parameters

Field	Type	Required	Description / constraints
model	string	Yes	Selected model identifier. Example: YOUR_MODEL

Response fields

Field	Type	Description / constraints
setupComplete	object	Setup-complete event.
serverContent.modelTurn.parts[]	object	Model audio or other content parts.
serverContent.turnComplete	boolean	Turn completion.

Illustrative response; some endpoints show essential fields only. Optional fields may be absent.

{
  "setupComplete": {}
}

Streaming

Connect with wss, send setup, then realtimeInput or clientContent. Handle audio/transcription events, turnComplete, interrupted and errors. The example sends one request after setupComplete and requires model output followed by turnComplete. Errors, timeout, premature close or interrupted mean incomplete; do not automatically resubmit.

Errors and retries

Check HTTP status and protocol-specific errors; do not mix error formats across protocols. For 401/403 check credentials and permissions, for 429 back off, and for 5xx check gateway/upstream status. Error bodies may contain error/message/code depending on the protocol.

Use bounded exponential backoff for reads. Do not blindly retry task creation, generation, or key mutations, to avoid duplicate billing or execution.

Document information and sources

Documentation source

Definition version: 1

Reviewed: 2026-11-02

Source: https://api.tu-zi.com/en/docs/api/reference-relay-gemini-live

Related protocol guide: https://api.tu-zi.com/docs/api/gemini

This export contains documentation and examples only, without real credentials, account data or user requests.

Request examples

# Gemini Live requires a WebSocket client.
# See the Python or JavaScript example and /docs/api/gemini.

# Python 3.11+: pip install 'websockets>=14,<16'
import asyncio, json, os
from websockets.asyncio.client import connect

async def main():
    async with asyncio.timeout(120):
        async with connect("wss://api.tu-zi.com/v1beta/live?model=YOUR_MODEL", additional_headers={
            'Authorization': "Bearer " + os.environ["API_KEY"]
        }, max_size=2*1024*1024, max_queue=4, compression=None) as ws:
            completed = sent = received_output = False
            await ws.send("{\"setup\":{\"model\":\"models/YOUR_MODEL\",\"generationConfig\":{\"responseModalities\":[\"AUDIO\"]}}}")
            async for message in ws:
                event = json.loads(message)
                if 'error' in event:
                    raise RuntimeError(event['error'])
                if 'setupComplete' in event and not sent:
                    sent = True
                    await ws.send("{\"clientContent\":{\"turns\":[{\"role\":\"user\",\"parts\":[{\"text\":\"Say hello briefly.\"}]}],\"turnComplete\":true}}")
                content = event.get('serverContent', {})
                if content.get('interrupted'):
                    raise RuntimeError('Turn was interrupted; no complete result')
                text = content.get('outputTranscription', {}).get('text', '')
                if text:
                    received_output = True
                    print(text, end='', flush=True)
                # Consume content.modelTurn.parts[].inlineData for PCM audio.
                if any(part.get('inlineData', {}).get('data') or part.get('text') for part in content.get('modelTurn', {}).get('parts', [])):
                    received_output = True
                if content.get('turnComplete'):
                    if not sent or not received_output:
                        raise RuntimeError('Turn ended without model output')
                    completed = True
                    print(event.get('usageMetadata', {}))
                    break
            if not completed:
                raise RuntimeError('Connection closed before a complete turn')

asyncio.run(main())

// Node.js: npm install ws
import WebSocket from 'ws';

const ws = new WebSocket("wss://api.tu-zi.com/v1beta/live?model=YOUR_MODEL", {
  headers: { Authorization: "Bearer " + process.env.API_KEY },
  perMessageDeflate: false,
  maxPayload: 2 * 1024 * 1024,
});
let completed = false, sent = false, receivedOutput = false, failed = false;
function fail(error) {
  if (failed) return;
  failed = true;
  console.error(error);
  process.exitCode = 1;
  ws.terminate();
}
const timeout = setTimeout(() => fail(new Error('Timed out before a complete turn')), 120000);
ws.on('close', () => {
  clearTimeout(timeout);
  if (!completed && !failed) fail(new Error('Connection closed before a complete turn'));
});
ws.on('error', fail);
ws.on('open', () => ws.send(JSON.stringify({
  "setup": {
    "model": "models/YOUR_MODEL",
    "generationConfig": {
      "responseModalities": [
        "AUDIO"
      ]
    }
  }
})));
ws.on('message', (message) => {
  if (failed || completed) return;
  try {
    const event = JSON.parse(message.toString());
    if (event.error) throw new Error(JSON.stringify(event.error));
    if (event.setupComplete && !sent) {
      sent = true;
      ws.send(JSON.stringify({"clientContent":{"turns":[{"role":"user","parts":[{"text":"Say hello briefly."}]}],"turnComplete":true}}));
    }
    const content = event.serverContent || {};
    if (content.interrupted) throw new Error('Turn was interrupted; no complete result');
    const text = content.outputTranscription?.text;
    if (text) { receivedOutput = true; process.stdout.write(text); }
    // Consume content.modelTurn.parts[].inlineData for PCM audio.
    if (content.modelTurn?.parts?.some(part => part.inlineData?.data || part.text)) receivedOutput = true;
    if (content.turnComplete) {
      if (!sent || !receivedOutput) throw new Error('Turn ended without model output');
      completed = true;
      console.log(event.usageMetadata);
      ws.close();
    }
  } catch (error) { fail(error); }
});