Implementation:Openai Openai node ToReadableStream

Knowledge Sources	openai-node Web Streams API
Domains	Streaming, Server_Architecture
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete tool for converting ChatCompletionStream to a Web ReadableStream for HTTP proxying provided by the openai-node SDK.

Description

The toReadableStream() method on ChatCompletionStream converts the async iterable stream into a Web Streams API ReadableStream. Each chunk from the OpenAI stream is serialized to JSON, appended with a newline, and encoded as UTF-8 bytes. This ReadableStream can be used directly as an HTTP response body.

The base implementation lives in Stream.toReadableStream() (src/core/streaming.ts), with ChatCompletionStream delegating via its EventStream base class.

Usage

Call .toReadableStream() on a ChatCompletionStream instance to get a ReadableStream suitable for HTTP response bodies. Set appropriate response headers (Content-Type: text/event-stream or application/octet-stream).

Code Reference

Source Location

• Repository: openai-node
• File: src/lib/ChatCompletionStream.ts (ChatCompletionStream.toReadableStream), src/core/streaming.ts (Stream.toReadableStream base)
• Lines: ChatCompletionStream.ts:L603-606, streaming.ts:L191-215

Signature

// On ChatCompletionStream
class ChatCompletionStream<ParsedT> {
  toReadableStream(): ReadableStream;
}

// Base implementation on Stream
class Stream<Item> {
  toReadableStream(): ReadableStream {
    // Creates ReadableStream where each chunk is:
    // new TextEncoder().encode(JSON.stringify(item) + '\n')
  }
}

Import

import OpenAI from 'openai';
// Access via: stream.toReadableStream()

I/O Contract

Inputs

Name	Type	Required	Description
(method on)	ChatCompletionStream	Yes	Active stream from client.chat.completions.stream()

Outputs

Name	Type	Description
readableStream	ReadableStream	Web Streams ReadableStream of UTF-8 encoded JSON lines

Usage Examples

Express Server Proxy

import OpenAI from 'openai';
import express from 'express';

const app = express();
const client = new OpenAI();

app.post('/api/chat', async (req, res) => {
  const stream = client.chat.completions.stream({
    model: 'gpt-4o',
    messages: req.body.messages,
  });

  res.setHeader('Content-Type', 'text/event-stream');

  // Convert to ReadableStream and pipe to response
  const readableStream = stream.toReadableStream();
  // Use Web Streams pipeTo or manual piping
  for await (const chunk of readableStream) {
    res.write(chunk);
  }
  res.end();
});

Next.js Route Handler

import OpenAI from 'openai';

const client = new OpenAI();

export async function POST(req: Request) {
  const { messages } = await req.json();

  const stream = client.chat.completions.stream({
    model: 'gpt-4o',
    messages,
  });

  return new Response(stream.toReadableStream());
}

Related Pages

Implements Principle

• Principle:Openai_Openai_node_Server_To_Client_Stream_Proxying

Requires Environment

• Environment:Openai_Openai_node_Node_20_Runtime