README

🦸🏽‍♀️🛠️🪚⚙️🦸🏽 YOU CAN CONTRIBUTE TO THIS PROJECT, CONTACT ME TO: jose.lopez.lara.cto@gmail.com 🦸🏽‍♀️🛠️🪚⚙️🦸🏽

Why use Gemini lite instead other open source Gemini Sdk?

Thanks to its minimalist syntax based on Facades, you can integrate AI functionalities effortlessly. Soon, it will include a ✨token and request limit management system inspired by Laravel Permission✨, perfect for streamlining monetizable projects. Moreover, its support is guaranteed until 2026, as it will be used in two projects set to be launched into production this year.Support the project and help extend its lifespan. If you'd like to contribute, feel free to contact me.

Features

Feature status:

🟢 Feature added
🟡 Feature in progress
🔴 Feature doesn't started
🟣 Feature doesn't planet (If you want contribute with this, contact me)

In progress	Progress Status
Text prompt support	🟢
Text prompt with file(Image, pdf, etc) support	🟢
Chat history support	🟢
Change model config in runtime	🟢
Change between Gemini models	🟢
JSON mode support	🟢
Easy upload file to get uri and mime type	🟢
Easy get current gemini config function	🟢
Easy get chat history of current chat instance	🟢
✨ Limit tokens and request support ✨	🟢
Tokens counter before to do prompt	🟢
Gemini flash 2.0 beta	🟢
Automatic unit testing	🟡
Embedding support	🟢
Image generator support	🟣

Gemini Lite for Laravel - Documentation

Get Started

Requeriments

You have to verify have added in your project:

php: Minimum version ^8.0
PostgreSQL
guzzlehttp/guzzle: Minimum version ^7.0
illuminate/console: Minimum version ^9.0
illuminate/database: Minimum version ^9.0
illuminate/http: Minimum version ^9.0
illuminate/support: Minimum version ^9.0

Installation

To install the Gemini Lite for Laravel package, use composer require liteopensource/gemini-lite-laravel or add the following line to your composer.json

"require": {
    "liteopensource/gemini-lite-laravel": "^0.0.3"
}

Then run: composer update

Configuration

Publish your config file geminilite.php using:

php artisan vendor:publish --tag="geminilite-config"

Add your Gemini API key to your .env file:

GEMINILITE_SECRET_API_KEY=your_api_key_here

GeminiService

GeminiService provides methods to interact with the Gemini AI model.

Creating a New Chat

Example of starting a new chat session:

use LiteOpenSource\GeminiLiteLaravel\Src\Facades\Gemini;

$chat = Gemini::newChat();

Sending Prompts

To send a prompt to the Gemini model:

$response = $chat->newPrompt("Your prompt here");

Changing Model Configuration

You can modify the model configuration at runtime:

$chat->setGeminiModelConfig($temperature, $topK, $topP, $maxOutputTokens, $returnMimeType);

Using JSON Mode

$responseSchema = [
    // Your JSON schema here
];

$chat->setGeminiModelConfig(
    temperature: 1,
    topK: 40,
    topP: 0.95,
    maxOutputTokens: 8192,
    responseMimeType: 'application/json',
    responseSchema: $responseSchema
);

Changing Gemini Model

You can switch between different Gemini models. The currently available models are:

Stable Models

gemini-1.5-flash: Fast and efficient model for everyday use.
gemini-1.5-pro: Advanced professional model with enhanced capabilities.
gemini-1.5-flash-8b: Lightweight 8B parameter version for faster processing.
gemini-2.0-flash: Improved speed and performance in the 2.0 series.
gemini-2.0-flash-lite: A lightweight version of Gemini 2.0 Flash.
gemini-2.0-flash-lite-preview-02-05: Preview of the lightweight 2.0 version.

Experimental Models

gemini-2.0-flash-exp: Cutting-edge features in the experimental Flash model.
gemini-2.0-pro-exp-02-05: Professional experimental model in the 2.0 series.
gemini-2.0-flash-thinking-exp-01-21: Enhanced reasoning capabilities.
gemini-2.0-flash-exp-image-generation: Specialized model for image generation tasks.
learnlm-1.5-pro-experimental: Learning-focused experimental model.

Gemini 2.5 Models

gemini-2.5-pro-preview-03-25: Preview version of Gemini 2.5 Pro (note: no free quota tier).
gemini-2.5-pro-exp-03-25: Experimental version of Gemini 2.5 Pro with free quota tier access.

Note: The gemini-2.5-pro-preview-03-25 model doesn't have a free quota tier. Google recommends using the experimental version (gemini-2.5-pro-exp-03-25) for free tier access.

$chat->changeGeminiModel("gemini-2.0-flash"); // Example using new model

Getting Current Model Configuration

You can retrieve the current configuration of the Gemini model:

$currentConfig = $chat->getGeminiModelConfig();

Getting Chat History

You can retrieve the complete chat history of your conversation with Gemini:

$gemini = Gemini::newChat();

// Make some prompts
$response1 = $gemini->newPrompt('Hello');
$response2 = $gemini->newPrompt('How are you?');

// Get the full chat history
$history = $gemini->getHistory();

// The history contains all messages exchanged, including both user prompts and model responses
// Each message contains:
// - role: "user" or "model"
// - parts: array containing the text or file data

UploadFileToGeminiService

This service allows you to upload files for processing by Gemini.

Processing Files from a Path

To process a file from a local path:

use LiteOpenSource\GeminiLiteLaravel\Src\Facades\UploadFileToGemini;

$filePath = storage_path('your/file/path.file');
$fileProcessed = UploadFileToGemini::processFileFromPath($filePath);

$uri = $fileProcessed->getUri();
$mimeType = $fileProcessed->getMimeType();

Processing Uploaded Files

This is really useful when you work with Livewire and using use WithFileUploads;

To process an uploaded file:

use LiteOpenSource\GeminiLiteLaravel\Src\Facades\UploadFileToGemini;

$file = $request->file($yourFile);
$fileProcessed = UploadFileToGemini::processFileFromUpload($file);

$uri = $fileProcessed->getUri();
$mimeType = $fileProcessed->getMimeType();

GeminiTokenCountService

This service is responsible for counting the tokens consumed by a text prompt. It returns an integer indicating the number of tokens consumed by the text prompt.

$tokens = GeminiTokenCount::coutTextTokens("Input text");

Token Limit

Token limit is independent of GeminiChat so the use of this feature is not integrated into the use of Gemini

This feature is intended for controlling user token consumption. It uses tables in the database to control usage. Therefore, it is important to publish the respective seeders for its operation.

Command to publish seeder and tables:

php artisan vendor:publish --tag="geminilite-limit-tokes"

To use this feature you need to add the HasGeminiRoles trait to the model user

use LiteOpenSource\GeminiLiteLaravel\Src\Traits\HasGeminiRoles;

class User extends Authenticatable
{
    use  HasGeminiRoles;

}

AssignRole

$testUser = User::create([
    'name' => 'John',
    'email' => 'test@example.com',
    'password'=> Hash::make("password"),
]);
/*
You can assign the role by the 
index or name of the role in the database 

$testUser2->assignGeminiRole('limited_user');

*/
$testUser->assignGeminiRole(1);

Can make request?

This function is for check if the user can make request to Gemini according to the established limits. The function returs a boolean value

// $user MUST BE an instance of model User
$user = User::find(1); 
$canMakeRequest = $user->canMakeRequestToGemini();

Is active

This function is for check if the user is active. The function returs a boolean value

// $user MUST BE an instance of model User
$user = User::find(1); 
$canMakeRequest = $user->isActiveInGemini();

The difference between isActive and canMakeRequest is that canMakeRequest works based on the limit of tokens and daily and monthly requests and isActive is a more manual and controlled way of controlling the use.

Update usage

This method updates gemini usage, increments the request counter, and increments the token counter. As a parameter it receives the number of tokens used by a prompt as an integer.

// $user MUST BE an instance of model User
$user = User::find(1); 
$user->updateUsageTracking($tokens);

Store Gemini Request

This function is intended to store data from the request to gemini, this function does not affect the tracking of user usage, it only stores data.

$testUser->storeGeminiRequest(requestType: "Test",
 consumedTokens: $tokens, 
 requestSuccessful: true, 
 requestData: ["request"=> $prompt],
 responseData: ["response"=> $response]);

EmbeddingService

This service provides text embedding capabilities using Gemini's text-embedding-004 model.

Basic Usage

use LiteOpenSource\GeminiLiteLaravel\Src\Facades\Embedding;

// Generate embedding for a single text
$embedding = Embedding::embedText("Hello world");

// Generate batch embeddings
$embeddings = Embedding::embedBatch([
    "First text",
    "Second text",
    "Third text"
]);

// With additional options
$embedding = Embedding::embedText($text, [
    'taskType' => 'SEMANTIC_SIMILARITY',
    'title' => 'Document Title'
]);

Examples

Text-Based Chat

You can keep chat history easily:

$gemini = Gemini::newChat();
$response = $gemini->newPrompt('How much is 1 + 1?');
$followUp = $gemini->newPrompt('Add 8 to the previous result');

Image-Based Chat

You can combine image prompt and text prompt in the only one prompt

$testImagePath = storage_path('app/public/test_image.jpeg');
$uploadedFile = UploadFileToGemini::processFileFromPath($testImagePath);

$gemini = Gemini::newChat();
$response = $gemini->newPrompt(
    "What do you see in this image?",
    $uploadedFile->getUri(),
    $uploadedFile->getMimeType()
);

Changing Configuration at Runtime

$gemini = Gemini::newChat();
$gemini->setGeminiModelConfig(1, 40, 0.95, 8192, 'text/plain');
$response = $gemini->newPrompt('Generate a creative story');

JSON Mode Chat

Here's an example of using JSON mode to generate cookie recipes:

$responseSchema = [
    "responseSchema" => [
        "type" => "object",
        "description" => "Return some of the most popular cookie recipes",
        "properties" => [
            "recipes" => [
                "type" => "array",
                "items" => [
                    "type" => "object",
                    "properties" => [
                        "recipe_name" => [
                            "type" => "string",
                            "description" => "name of recipe using upper case"
                        ],
                        "ingredients_number" => [
                            "type" => "number"
                        ]
                    ],
                    "required" => [
                        "recipe_name",
                        "ingredients_number"
                    ]
                ]
            ],
            "status_response" => [
                "type" => "array",
                "items" => [
                    "type" => "object",
                    "properties" => [
                        "sucess" => [
                            "type" => "string",
                            "description" => "Short message in uppercase about request"
                        ],
                        "code" => [
                            "type" => "string",
                            "description" => "Status code",
                            "enum" => [
                                "200",
                                "400"
                            ]
                        ]
                    ],
                    "required" => [
                        "sucess",
                        "code"
                    ]
                ]
            ]
        ],
        "required" => [
            "recipes",
            "status_response"
        ]
    ]
];

$geminiChat = Gemini::newChat();
$geminiChat->setGeminiModelConfig(
    temperature: 1,
    topK: 40,
    topP: 0.95,
    maxOutputTokens: 8192,
    responseMimeType: 'application/json',
    responseSchema: $responseSchema
);

$response = $geminiChat->newPrompt("Generate a list of cookie recipes. Make the outputs in JSON format.");

$responseObject = json_decode($response);
$recipes = $responseObject->recipes;
$firstRecipeName = $recipes[0]->recipe_name;

// You can now work with the structured JSON response

Changing Gemini Model Example

You can compare responses from different Gemini models:

$geminiChat1 = Gemini::newChat();
$geminiChat1->changeGeminiModel("gemini-1.5-flash-8b");
$response1 = $geminiChat1->newPrompt("Your prompt here");

$geminiChat2 = Gemini::newChat();
$geminiChat2->changeGeminiModel("gemini-1.5-pro-002");
$response2 = $geminiChat2->newPrompt("Your prompt here");

// Compare $response1 and $response2

Getting Current Model Configuration Example

You can compare responses from different Gemini models:

use LiteOpenSource\GeminiLiteLaravel\Src\Facades\Gemini;

$geminiChat = Gemini::newChat();

// Get initial configuration
$initialModelConfig = $geminiChat->getGeminiModelConfig();

// Change configuration
$geminiChat->setGeminiModelConfig(2, 64, 1, 8192, 'text/plain');

// Get updated configuration
$updatedModelConfig = $geminiChat->getGeminiModelConfig();

// Now you can compare or use these configurations

Count the number of tokens in a given text.

$tokens = GeminiTokenCount::coutTextTokens("Hello Gemini can you write a funny story");

Check if the user can make a request to Gemini.

$user = User::find(1);
$canMakeRequest = $user->canMakeRequestToGemini();

return response()->json([
    'success'=> true,
    'message'=> 'Everything okay',
    'User can make request ' => $canMakeRequest
],200);

Check if the user is active in Gemini.

$user = User::find(1);
$isActive = $user->isActiveInGemini();

return response()->json([
    'success'=> true,
    'message'=> 'Everything okay',
    'User is active' => $isActive
],200);

Update the usage tracking for the user.

$tokens = GeminiTokenCount::coutTextTokens("Write a teror story for my class");
$user = User::find(1);

if(!$user->canMakeRequestToGemini()){
    return response()->json([
        "success"=> false,
        "message"=> "User unauthorized",
    ],403);
}

$user->updateUsageTracking($tokens);

Assign roles to test users.

$faker = Faker::create();
$testUser = User::create([
    'name' => $faker->name,
    'email' => $faker->unique()->safeEmail,
    'password'=> Hash::make("1234"),
]);

$role = $testUser->assignGeminiRole(1);

$testUser2 = User::create([
    'name' => $faker->name,
    'email' => $faker->unique()->safeEmail,
    'password'=> Hash::make("1234"),
]);

$role = $testUser2->assignGeminiRole('limited_user');

Log the requests made by the user.

$prompt = "Write my tesis please";
$gemini = Gemini::newChat();

if(!$testUser->canMakeRequestToGemini()){
    return response()->json([
        'code' => 403,
        "success"=> true,
        "message"=> "The user cannot make requests to gemini",
    ],403);
}

$response = $gemini->newPrompt($prompt);
$tokens = GeminiTokenCount::coutTextTokens($prompt);
$testUser->updateUsageTracking($tokens);
$testUser->storeGeminiRequest(requestType: "Test", consumedTokens: $tokens, requestSuccessful: true, requestData: ["request"=> $prompt],responseData: ["response"=> $response]);

License

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “liteopensource/gemini-lite-laravel”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

liteopensource / gemini-lite-laravel

Maintainers

Details

README

Why use Gemini lite instead other open source Gemini Sdk?

Features

Table of Contents

Get Started

Requeriments

Installation

Configuration

GeminiService

Creating a New Chat

Sending Prompts

Changing Model Configuration

Using JSON Mode

Changing Gemini Model

Stable Models

Experimental Models

Gemini 2.5 Models

Getting Current Model Configuration

Getting Chat History

UploadFileToGeminiService

Processing Files from a Path

Processing Uploaded Files

GeminiTokenCountService

Token Limit

AssignRole

Can make request?

Is active

Update usage

Store Gemini Request

EmbeddingService

Basic Usage

Examples

Text-Based Chat

Image-Based Chat

Changing Configuration at Runtime

JSON Mode Chat

Changing Gemini Model Example

Getting Current Model Configuration Example

Count the number of tokens in a given text.

Check if the user can make a request to Gemini.

Check if the user is active in Gemini.

Update the usage tracking for the user.

Assign roles to test users.

Log the requests made by the user.

License