Use Nano

Convenience wrapper for Gemini Nano running in Chrome preview builds with AI features switched on.

Optimized for instant and high-frequency inference. Currently, window.ai does not allow queueing and will cancel any ongoing request when a new request comes in. This package implements a queue that waits for the current request to finish before starting a new one. You can also clear the queue if you want to cancel all current and pending inference requests.

• Streaming response from the model
• Queueing of prompts
• Automatic canceling
• useNano hook for React that immediately submits a prompt and streams the output
• Type declarations for Chrome's new window.ai object

Installation

Make sure you're running Chrome Dev or Chrome Canary with AI features enabled.

npm install https://github.com/freakyflow/use-nano

Usage

Using the useNano hook

Creating a simple, instant inference UI:

export default function TestPage() {
  const [input, setInput] = useState("Who are you?");
  const output = useNano(input);
  return (
    <div>
      <input value={input} 
      onChange={(e) => setInput(e.target.value)} />
      <div>{output}</div>
    </div>
  );
}

Clear the queue with the hook

const output = useNano("Who are you?", { clearQueue: true });

Stream

promptQueue.enqueue("Who are you?", response => console.log(response));

Clear queue

promptQueue.clear();