AblyDocs

Guide: Stream Vercel AI SDK responses using the message-per-token pattern

Open in

This guide shows you how to stream AI responses from the Vercel AI SDK over Ably using the message-per-token pattern. Specifically, it implements the explicit start/stop events approach, which publishes each response token as an individual message, along with explicit lifecycle events to signal when responses begin and end.

Using Ably to distribute tokens from the Vercel AI SDK enables you to broadcast AI responses to thousands of concurrent subscribers with reliable message delivery and ordering guarantees, ensuring that each client receives the complete response stream with all tokens delivered in order. This approach decouples your AI inference from client connections, enabling you to scale agents independently and handle reconnections gracefully.

Prerequisites

To follow this guide, you need:

  • Node.js 20 or higher
  • A Vercel AI Gateway API key
  • An Ably API key

Useful links:

Create a new NPM package, which will contain the publisher and subscriber code:

mkdir ably-vercel-message-per-token && cd ably-vercel-message-per-token
npm init -y

Install the required packages using NPM:

npm install ai@^6 ably@^2

Export your Vercel AI Gateway API key to the environment, which will be used later in the guide by the Vercel AI SDK:

export AI_GATEWAY_API_KEY="your_api_key_here"

Step 1: Get a streamed response from Vercel AI SDK

Initialize the Vercel AI SDK and use streamText to stream model output as a series of events.

Create a new file publisher.mjs with the following contents:

JavaScript

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

import { streamText } from 'ai';

// Process each streaming event
function processEvent(event) {
  console.log(JSON.stringify(event));
  // This function is updated in the next sections
}

// Create streaming response from Vercel AI SDK
async function streamVercelResponse(prompt) {
  const result = streamText({
    model: 'openai/gpt-4o',
    prompt: prompt,
  });

  // Iterate through streaming events using fullStream
  for await (const event of result.fullStream) {
    processEvent(event);
  }
}

// Usage example
streamVercelResponse("Tell me a short joke");

Understand Vercel AI SDK streaming events

The Vercel AI SDK's streamText function provides a fullStream property that returns all stream events. Each event includes a type property which describes the event type. A complete text response can be constructed from the following event types:

  • text-start: Signals the start of a text response. Contains an id to correlate subsequent events.

  • text-delta: Contains a single text token in the text field. These events represent incremental text chunks as the model generates them.

  • text-end: Signals the completion of a text response.

The following example shows the event sequence received when streaming a response:

JSON

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

// 1. Stream initialization
{"type":"start"}
{"type":"start-step","request":{...}}
{"type":"text-start","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","providerMetadata":{...}}

// 2. Text tokens stream in as delta events
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":"Why"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" don't"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" skeleton"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":"s"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" fight"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" each"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" other"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":"?\n\n"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":"They"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" don't"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" have"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" the"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":" guts"}
{"type":"text-delta","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","text":"!"}

// 3. Stream completion
{"type":"text-end","id":"msg_0cc4da489ab9d4d101696f97d7c9548196a04f71d10a3a4c99","providerMetadata":{...}}
{"type":"finish-step","finishReason":"stop","usage":{"inputTokens":12,"outputTokens":15,"totalTokens":27,"reasoningTokens":0,"cachedInputTokens":0},"providerMetadata":{...}}
{"type":"finish","finishReason":"stop","totalUsage":{"inputTokens":12,"outputTokens":15,"totalTokens":27,"reasoningTokens":0,"cachedInputTokens":0}}

Step 2: Publish streaming events to Ably

Publish Vercel AI SDK streaming events to Ably to reliably and scalably distribute them to subscribers.

This implementation follows the explicit start/stop events pattern, which provides clear response boundaries.

Initialize the Ably client

Add the Ably client initialization to your publisher.mjs file:

JavaScript

1

2

3

4

5

6

7

8

9

10

import Ably from 'ably';

// Initialize Ably Realtime client
const realtime = new Ably.Realtime({
  key: 'demokey:*****',
  echoMessages: false
});

// Create a channel for publishing streamed AI responses
const channel = realtime.channels.get('map-cod-cog');
API key:
DEMO ONLY

The Ably Realtime client maintains a persistent connection to the Ably service, which allows you to publish tokens at high message rates with low latency.

Map Vercel AI SDK streaming events to Ably messages

Choose how to map Vercel AI SDK streaming events to Ably messages. You can choose any mapping strategy that suits your application's needs. This guide uses the following pattern as an example:

  • start: Signals the beginning of a response
  • token: Contains the incremental text content for each delta
  • stop: Signals the completion of a response

Update your publisher.mjs file to initialize the Ably client and update the processEvent() function to publish events to Ably:

JavaScript

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

// Track response ID across events
let responseId = null;

// Process each streaming event and publish to Ably
function processEvent(event) {
  switch (event.type) {
    case 'text-start':
      // Capture response ID from text-start event
      responseId = event.id;

      // Publish start event with response ID
      channel.publish({
        name: 'start',
        extras: {
          headers: { responseId }
        }
      });
      break;

    case 'text-delta':
      // Publish each text delta as a token
      channel.publish({
        name: 'token',
        data: event.text,
        extras: {
          headers: { responseId }
        }
      });
      break;

    case 'text-end':
      // Publish stop event when stream completes
      channel.publish({
        name: 'stop',
        extras: {
          headers: { responseId }
        }
      });
      break;
  }
}

This implementation:

  • Captures the responseId from the text-start event
  • Publishes a start event at the beginning of the response
  • Filters for text-delta events and publishes them as token events
  • Publishes a stop event when the response completes using the text-end event
  • All published events include the responseId in message extras to allow the client to correlate events relating to a particular response

Run the publisher to see tokens streaming to Ably:

node publisher.mjs

Step 3: Subscribe to streaming tokens

Create a subscriber that receives the streaming events from Ably and reconstructs the response.

Create a new file subscriber.mjs with the following contents:

JavaScript

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

import Ably from 'ably';

// Initialize Ably Realtime client
const realtime = new Ably.Realtime({ key: 'demokey:*****' });

// Get the same channel used by the publisher
const channel = realtime.channels.get('map-cod-cog');

// Track responses by ID
const responses = new Map();

// Handle response start
await channel.subscribe('start', (message) => {
  const responseId = message.extras?.headers?.responseId;
  console.log('\n[Response started]', responseId);
  responses.set(responseId, '');
});

// Handle tokens
await channel.subscribe('token', (message) => {
  const responseId = message.extras?.headers?.responseId;
  const token = message.data;

  // Append token to response
  const currentText = responses.get(responseId) || '';
  responses.set(responseId, currentText + token);

  // Display token as it arrives
  process.stdout.write(token);
});

// Handle response stop
await channel.subscribe('stop', (message) => {
  const responseId = message.extras?.headers?.responseId;
  const finalText = responses.get(responseId);

  console.log('\n[Response completed]', responseId);
});

console.log('Subscriber ready, waiting for tokens...');
API key:
DEMO ONLY

Run the subscriber in a separate terminal:

node subscriber.mjs

With the subscriber running, run the publisher in another terminal. The tokens stream in realtime as the AI model generates them.

Step 4: Stream with multiple publishers and subscribers

Ably's channel-oriented sessions enables multiple AI agents to publish responses and multiple users to receive them on a single channel simultaneously. Ably handles message delivery to all participants, eliminating the need to implement routing logic or manage state synchronization across connections.

Broadcasting to multiple subscribers

Each subscriber receives the complete stream of tokens independently, enabling you to build collaborative experiences or multi-device applications.

Run a subscriber in multiple separate terminals:

# Terminal 1
node subscriber.mjs

# Terminal 2
node subscriber.mjs

# Terminal 3
node subscriber.mjs

All subscribers receive the same stream of tokens in realtime.

Publishing concurrent responses

The implementation uses responseId in message extras to correlate tokens with their originating response. This enables multiple publishers to stream different responses concurrently on the same channel, with each subscriber correctly tracking all responses independently.

To demonstrate this, run a publisher in multiple separate terminals:

# Terminal 1
node publisher.mjs

# Terminal 2
node publisher.mjs

# Terminal 3
node publisher.mjs

All running subscribers receive tokens from all responses concurrently. Each subscriber correctly reconstructs each response separately using the responseId to correlate tokens.

Next steps