TensorFlow.js

What it does

TensorFlow.js is a machine learning library for JavaScript. Developers supply their own code, existing pre-trained models, or Python TensorFlow models converted to run on the web, and get working ML models that run directly in a browser page or in Node.js. It can also produce newly trained or retrained models built with its JavaScript APIs.

Common uses are running official off-the-shelf models, retraining an existing model on your own data through transfer learning, or building and training models from scratch in JavaScript. Because ML runs where the web app already lives, no separate Python serving step is needed. Google's demos include a Webcam Controller trained with images in the browser, a real-time neural-network piano performance and a web ML photo experience called Holobooth. Tutorials with end-to-end examples are provided.

It suits JavaScript and web developers adding ML features to browser or Node.js applications, and learners approaching ML through web technologies; a video course is linked. It is a developer library rather than a hosted app, so using it means writing code. Check the tutorials and the GitHub repository for performance, hardware acceleration and version requirements before committing to a design.

What it can do

• Run pre-trained, official TensorFlow.js models for common use cases
• Convert Python TensorFlow models to run in the browser or Node.js
• Retrain existing models on your own data using transfer learning
• Build and train models directly in JavaScript with its APIs
• Run ML in the browser or under Node.js
• Learn from tutorials and live in-browser demos

Example workflows

Add a pre-trained model to a web page

Illustrative example. Input: a web app and an official TensorFlow.js model. Steps: choose a model from the models page, load it in the browser, and pass it user input such as webcam frames. Output: predictions computed client-side in the page.

Customize a model with your own data

Illustrative example. Input: a pre-existing model and your own labelled examples. Steps: load the model, apply transfer learning in JavaScript, and run the retrained model in the browser or Node.js. Output: a model tuned to your task.

Before you choose

• It is a code library for developers, not a no-code app
• Python TensorFlow models can be converted, so test your model's conversion early
• Check the GitHub repository for licence terms and hardware acceleration details

Listed under

• Coding
• Browser ML
• Developer library
• Javascript
• Machine learning
• Node.js
• Open source

Get started

Follow the Get started with TensorFlow.js tutorials

1. Open the Get started tutorials and pick an end-to-end example
2. Install the library for browser or Node.js use via the Install and GitHub links
3. Load an official model or convert a Python TensorFlow model

The source code is available under Apache 2.0. You supply browser or server compute; hosted services may charge separately.

Source: github.com/tensorflow/tfjs, checked 28 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 28 Oct 2026. Each source supports part of the overview; availability and prices can change.

• TensorFlow.js | Machine Learning for JavaScript Developers ↗www.tensorflow.org
• Contribute to TensorFlow ↗www.tensorflow.org
• GitHub - tensorflow/tfjs: A WebGL accelerated JavaScript library for training and deploying ML models. · GitHub ↗github.com

Company website

tensorflow.org: product, pricing and policy pages

YC directory mirror

Batch and company name

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 28 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.