Making Predictions in JavaScript

Make AI predctions in the browser and in Node.js.

Function provides a JavaScript client for making predictions in the browser, in frameworks like React and Next.js, and in Node.js.

In this example, we'll make a minimal Next.js website that accepts a prompt and generates an image using @samplefxn/stable-diffusion

Creating a Next.js Project

Following the Next.js quickstart , open a terminal and run the following command:

# Clone the Next.js starter app
npx create-next-app@latest nextjs-blog --use-npm --example ""

You should see a nextjs-blog directory. Open the directory in your text editor, then open a terminal and start the development server:

# Start the development server
npm run dev

Once it's started, open http://localhost:3000 in the browser and you should see a page that says "Welcome to Next.js!"

next welcome

Installing Function

Open a terminal and run the following command:

# Open a terminal and run the following command:
npm install --save fxnjs

When building websites, do not specify your Function access key in code or environment variables that run in the browser. Your key will be visible and can be extracted.

Installing without a Bundler

When building websites without a JavaScript bundler like webpack, you can import Function using a <script> tag:

<!DOCTYPE html>
<script src=""></script>
<script type="text/javascript">
const fxn = new fxnjs.Function();
// Do stuff with `fxn` instance

Implementing the Page

Now, let's modify pages/index.js to show an input field for the prompt, a generation button, and the resulting image:

import { Function } from "fxnjs"
import Head from "next/head"
import { useState } from "react"
import styles from "../styles/Home.module.css"
export default function Home() {
const [prompt, setPrompt] = useState("");
const [imageUrl, setImageUrl] = useState("");
const [loading, setLoading] = useState(false);
return (
<div className={styles.container}>
<title>Function in Next.js</title>
<link rel="icon" href="/favicon.ico" />
{/* Title */}
<h1 className={styles.title}>
Welcome to <a href="">Function</a> for JavaScript
{/* Prompt input */}
onChange={e => setPrompt(}
{/* Generate button */}
style={{ marginTop: "16px", fontSize: "1.5rem" }}
{loading ? "Loading..." : "Generate" }
{/* Result image */}
<img src={imageUrl} style={{ marginTop: "16px" }} />
<style jsx>{`
main {
padding: 5rem 0;
flex: 1;
display: flex;
flex-direction: column;
justify-content: center;
align-items: center;
<style jsx global>{`
body {
padding: 0;
margin: 0;
font-family: -apple-system, BlinkMacSystemFont, Segoe UI, Roboto,
Oxygen, Ubuntu, Cantarell, Fira Sans, Droid Sans, Helvetica Neue,
* {
box-sizing: border-box;

Making a Prediction

Once you replace the contents of index.js with the code above, you'll notice that the "Generate" button doesn't do anything. Let's implement the actual prediction. First, add the following generate function:

const [prompt, setPrompt] = useState("");
const [imageUrl, setImageUrl] = useState("");
const [loading, setLoading] = useState(false);
const generate = async () => {
const fxn = new Function({ accessKey: "<ACCESS KEY>" });
const prediction = await fxn.predictions.create({
tag: "@samplefxn/stable-diffusion",
inputs: { prompt }
return (

Then navigate down to the <button> tag and add the following onClick handler:

{/* Generate button */}
style={{ marginTop: "16px", fontSize: "1.5rem" }}
{loading ? "Loading..." : "Generate" }

That's it!

Prediction in Next.js

Predict in the CLI