How to Use Google Gemini AI in PHP to Solve Image CAPTCHAs

Introduction

Image CAPTCHAs are designed to block automated tools like web crawlers, but modern multimodal AI models such as Google Gemini can recognize both text and images, making it possible to automate CAPTCHA solving within PHP applications.

Create Project

composer create-project workerman/webman webman20240312

Install Dependencies

composer require google-gemini-php/client

Gemini PHP is a community‑maintained PHP API client for the Gemini AI service and requires PHP 8.1+.

Project repository: https://github.com/google-gemini-php/client

If your project does not yet use a PSR‑18 client, install the discovery plugin:

composer require php-http/discovery

composer require guzzlehttp/guzzle

Text Model Test

<?php
/**
 * @desc Use Google Gemini to generate a text response
 */
declare(strict_types=1);
require_once '../vendor/autoload.php';
$apiKey = 'AIzaSyAPxxxxxxxxxxxxxxx_uEpw';
$client = \Gemini::client($apiKey);
$result = $client->geminiPro()->generateContent('What is PHP?');
echo $result->text() . PHP_EOL;

Output example:

PHP (Hypertext Preprocessor) is a general‑purpose scripting language especially suited for web development.
**Features:**
* Open‑source and free
* Cross‑platform
* Easy to learn
* Widely used (WordPress, Facebook, Wikipedia, etc.)
* Modular extensions
* Dynamic typing

Image Model Test – Recognize CAPTCHA Text

Get original CAPTCHA image

CAPTCHA file: captcha01.jpg

Reference code

<?php
/**
 * @desc Recognize text in a CAPTCHA image using Gemini
 */
declare(strict_types=1);
require_once '../vendor/autoload.php';
use Psr\Http\Message\RequestInterface;
use Psr\Http\Message\ResponseInterface;
$apiKey = 'AIzaSyAPLiuNxxxxxxxxxxxxxxx_uEpw';
$client = \Gemini::factory()
    ->withApiKey($apiKey)
    ->withBaseUrl('https://gemini.ailard.com/v1/')
    ->withHttpClient($client = new \GuzzleHttp\Client([]))
    ->withStreamHandler(fn(RequestInterface $request): ResponseInterface => $client->send($request, ['stream' => true]))
    ->make();
$result = $client->geminiProVision()->generateContent([
    'I will provide you with an image CAPTCHA, please recognize the content inside the CAPTCHA and output the text',
    new \Gemini\Data\Blob(
        mimeType: \Gemini\Enums\MimeType::IMAGE_JPEG,
        data: base64_encode(file_get_contents('./captcha01.jpg'))
    )
]);
echo $result->text() . PHP_EOL;

Recognition output

The content inside the CAPTCHA is "AXBV".

Image Model Test – Compute CAPTCHA Result

Get calculation CAPTCHA image

CAPTCHA file: captcha02.png

Reference code

<?php
/**
 * @desc Recognize and compute the result of a calculation CAPTCHA
 */
declare(strict_types=1);
require_once '../vendor/autoload.php';
use Psr\Http\Message\RequestInterface;
use Psr\Http\Message\ResponseInterface;
$apiKey = 'AIzaSyAPLiuNxxxxxxxxxxxxxxx_uEpw';
$client = \Gemini::factory()
    ->withApiKey($apiKey)
    ->withBaseUrl('https://gemini.ailard.com/v1/')
    ->withHttpClient($client = new \GuzzleHttp\Client([]))
    ->withStreamHandler(fn(RequestInterface $request): ResponseInterface => $client->send($request, ['stream' => true]))
    ->make();
$result = $client->geminiProVision()->generateContent([
    'I will provide you with an image CAPTCHA, please recognize the content inside the CAPTCHA and output the text',
    new \Gemini\Data\Blob(
        mimeType: \Gemini\Enums\MimeType::IMAGE_PNG,
        data: base64_encode(file_get_contents('./captcha02.png'))
    )
]);
echo $result->text() . PHP_EOL;

Recognition output

7 x 8 = ?
56