Docs | NPM | FAQ

Features

• 🍪 Multi-Cookie Use: Have multiple Bard instances
• 🖼️ Google Lens Support: Powerful image recognition
• 📷 Image Output: Access images from Google
• 🌐 Contextual Conversations: Maintain conversation context

Highlights

• ⚡ Fast Responses: More than 2x faster than googlebard
• 🪶 Lightweight: Just 2.5kb minzipped
• 📚 Typesafe: Types included out-of-the-box

Warning

Google is trying to stop these 3rd party APIs, so bard-ai may not work for some of you. Try following the steps on the FAQ. However, if that still does not work, it may be time to migrate to the palm-api package, for the official PaLM v2 API.

Comparison

Compared to alternative Bard API googlebard:

Learn more...

bard-ai surpasses leading competition GoogleBard by PawanOsman for three key reasons:

Size

bard-ai is a tiny 2.5kb minzipped, nearly 98% smaller than googlebard (112.8kb minzipped). This leads to faster downloads, and maximum efficiency.

Speed

bard-ai performs up to a whopping 2.25x faster than googlebard, when subjected to 4 consecutive requests, as measured with hyperfine.¹

Features

bard-ai is updated with the latest and most advanced features available in Google Bard itself, including powerful Image Recognition with Google Lens.

Quick Start

Copy the cookie labled __Secure-1PSID on bard.google.com, and use it in the code here:

import Bard from "bard-ai";

let myBard = new Bard(COOKIE);

console.log(await myBard.ask("Hello, world!"));

Continue to explore full features, including Google Lens integration, with the docs.

Contributors

A special shoutout to these amazing indviduals:

@acheong08 Original Python Bard API

@ThatXliner Maintainer

@Aldhanekaa Typescript and Feature Developer, Maintainer

@RFS-ADRENO TypeScript Developer, Maintainer

However, we thank every person that helps in the development process of this library, no matter that be in code, ideas, or anything else.

Read the docs to get started.

Footnotes

1. Run with hyperfine --runs 3 between bard-ai v2.0 and googlebard. Equivalent code was used in the runs:

   bard-ai:

   import Bard from "bard-ai";
   
   let bot = new Bard(COOKIE);
   
   console.log(await bot.ask("Hello world!"));
   console.log(await bot.ask("What is 1+1?"));
   console.log(await bot.ask("What about 2+2?"));
   console.log(await bot.ask("Goodbye."));
   

   googlebard:

   import { Bard } from "googlebard";
   
   let bot = new Bard(`__Secure-1PSID=${COOKIE}`);
   
   console.log(await bot.ask("Hello world!"));
   console.log(await bot.ask("What is 1+1?"));
   console.log(await bot.ask("What about 2+2?"));
   console.log(await bot.ask("Goodbye."));
   

   Result:

   Benchmark 1: node bard-ai.js
   Time (mean ± σ):     18.346 s ±  0.768 s    [User: 0.307 s, System: 0.067 s]
   Range (min … max):   17.531 s … 19.057 s    3 runs
   
   Benchmark 2: node googlebard.js
     Time (mean ± σ):     37.179 s ±  3.714 s    [User: 1.416 s, System: 0.191 s]
     Range (min … max):   34.230 s … 41.349 s    3 runs
   
   Summary
   'node bard-ai.js' ran
   2.03 ± 0.22 times faster than 'node googlebard.js'
   

   ↩