Chrome AI Test Page

This project was inspired by lightning-joyce/chromeai.

Visit the page alexcheng1982.github.io/chromeai/.

View screenshot

Implementation

promptStreaming is used for streaming output. During testing, I found out that prompt often gave incomplete results.

Use Chrome AI

Prerequisites

Latest version of Google Chrome (127+) or any compatible Chromium-based browser.

Tested Chrome versions:

• 129.0.6640.0

Set Up Built-in Gemini Nano in Chrome

1. Install Chrome Canary: Ensure you have version 129. Download Chrome Canary.
2. Enable Prompt API: Open chrome://flags/#prompt-api-for-gemini-nano, set it to "Enabled".
3. Enable Optimization Guide: Open chrome://flags/#optimization-guide-on-device-model, set it to "Enabled BypassPerfRequirement". Restart the browser.
4. Download Model: Go to chrome://components/, find Optimization Guide On Device Model, ensure it’s fully downloaded. If the version is "0.0.0.0", click "Check for update".
5. Verify Setup: Open a webpage, press F12, and check window.ai in the console.

Test Code:

const model = await window.ai.createTextSession();
await model.prompt("Who are you?");

View flags

Known issues

1. When message history is enabled, the model may not function properly.