Stream Text

Let's see how we can update our previous generateText example to use streaming.

Open core/stream-text.ts. You should see the following code already:

core/stream-text.ts

import { openai } from "@ai-sdk/openai";
import { generateText } from "ai";
import dotenv from "dotenv";

dotenv.config();

async function main() {
  const result = await generateText({
    model: openai("gpt-4o"),
    prompt: "Tell me a joke.",
  });

  console.log(result.text);
}

main().catch(console.error);

First, change the generateText function to streamText.

core/stream-text.ts

import { openai } from "@ai-sdk/openai";
import { streamText } from "ai";
import dotenv from "dotenv";

dotenv.config();

async function main() {
  const result = await streamText({
    model: openai("gpt-4o"),
    prompt: "Tell me a joke.",
  });

  console.log(result.text);
}

main().catch(console.error);

Next, replace your console.log with an asynchronous for-loop and iterate over the resulting textStream.

core/stream-text.ts

import { openai } from "@ai-sdk/openai";
import { streamText } from "ai";
import dotenv from "dotenv";

dotenv.config();

async function main() {
  const result = await streamText({
    model: openai("gpt-4o"),
    prompt: "Tell me a joke.",
  });

  for await (const textPart of result.textStream) { 
  } 
}

main().catch(console.error);

Finally, write the text to the console.

core/stream-text.ts

import { openai } from "@ai-sdk/openai";
import { streamText } from "ai";
import dotenv from "dotenv";

dotenv.config();

async function main() {
  const result = await streamText({
    model: openai("gpt-4o"),
    prompt: "Tell me a joke.",
  });

  for await (const textPart of result.textStream) {
    process.stdout.write(textPart); 
  }
}

main().catch(console.error);

Note: you are using process.stdout.write rather than console.log because it does not add any new lines.

Run the script in the terminal, and see what happens.

npx tsx core/stream-text.ts

You should see a joke streaming in the console just like ChatGPT!