Meet Window AI: A New Way To Use Your Own AI Models On The Web – Including Local Ones

Window AI is a new experimental tool launched to introduce a new kind of AI app based on a browser extension. This allows users to configure their AI models in one place and use them on the web. The tool works as a browser extension and JavaScript API, allowing users to maintain visibility on what data is requested by the apps they use. Currently, many AI apps require users to use their specific models or API keys, which can be limiting and can expose the keys.

One significant advantage of Window is that users can fine-tune their own local model on private data and use it on the web, with full visibility of the prompts. The tool is also a great use case for privacy and becomes portable. 

Window offers users a flexible and secure way to use AI models on the web. Users can manage all their model setups in one place, whether it’s an external model like OpenAI, proxied, or local. This ensures they can protect their privacy and use their preferred models without being limited by specific vendors or models. Users can also save their prompt history across apps.

Developers can utilize Window to create multi-model apps without worrying about API costs and limitations. Using the injected library, they can easily develop apps free from vendor lock-in and create decentralized apps. This allows developers to use their own AI models without being limited by specific vendors or models.

In addition to the Window extension, there is also an aggregator of Window apps called This aggregator features a variety of apps, including three games, a ChatGPT clone, a Toolformer implementation, and a template for users to create their own apps using Next.js.

How does it work?

The ‘Window’ extension allows users to set up their keys and models only once. Subsequently, apps can request permission to send prompts to their chosen model through the library. This process is transparent to the users as they are always informed about the prompts they receive and when they receive them.

Window currently works with the following models – OpenAI’s GPT-3.5, GPT-4, Together’s GPT NeoXT 20B, Cohere’s Xlarge, and open models like Alpaca, which can run locally.

Why should you build with Window?

There are several reasons to consider building apps with the Window extension. Firstly, it eliminates the infrastructure burden associated with model API costs, timeouts, and rate limiting, resulting in reduced server billing time. Additionally, Window enables easy integration of multiple models and takes care of model upgrades and support for other providers.

Furthermore, Window allows users to build privacy-conscious apps that communicate solely with the user’s preferred model, which significantly reduces the liability for the model’s output. 


The developers of Window AI hope that the JavaScript API will become an evolving standard for various AI apps, even multimodal ones. The tool is inspired by the way crypto wallets work, and to make it work, the developers believe that a variety of JavaScript apps built on top of the standard is needed. Additionally, private, niche models must plug into it from the other side, which could be a two-sided challenge.

All in all, Window offers a flexible and secure way to use AI models on the web, empowering users to manage their models in one place without being limited by specific vendors or models. Developers can easily build multi-model apps, free from infrastructure burdens and model API costs. However, adopting the JavaScript API standard may require further development and private, niche models to plug into it.

Check out the Web and Github. Don’t forget to join our 19k+ ML SubRedditDiscord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more. If you have any questions regarding the above article or if we missed anything, feel free to email us at

🚀 Check Out 100’s AI Tools in AI Tools Club

🐝 Join the Fastest Growing AI Research Newsletter Read by Researchers from Google + NVIDIA + Meta + Stanford + MIT + Microsoft and many others...