README

This is a simple PHP wrapper to interact with the OpenRouter API. The wrapper allows you to send messages to the OpenRouter AI model and get responses in both streaming and non-streaming modes.

Requirements

PHP 7.4 or higher
Composer

Installation

To get started, you can install the package via Composer.

1. Clone the repository or initialize it in your project:

composer require cloud-dark/openrouter:dev-main

2. Upgrade dependencies using Composer:

composer upgrade && composer update

This will install the necessary dependencies and set up the autoloader.

Usage

Once installed, you can use the OpenRouter class to send requests to the OpenRouter API.

Example PHP Code

1. Non-Streaming (Standard Request)

<?php

// Ensure Composer autoloader is loaded
require './vendor/autoload.php';  // No need for full path when using Composer autoload

use CloudDark\OpenRouter\OpenRouter;  // Use the appropriate namespace

// Create an instance of OpenRouter for non-streaming
$openRouter = new OpenRouter(
    'sk-or-v1-XXX', // Your OpenRouter API token
    'google/gemini-2.0-flash-exp:free' // Model (optional, default is "mistralai/mixtral-8x7b-instruct")
);

// Prepare the message to send
$messages = [
    [
        'role' => 'system',
        'content' => 'You are an AI assistant.'
    ],
    [
        'role' => 'user',
        'content' => 'Hello, AI!'
    ]
];

// Send the request and get the response
$response = json_decode($openRouter->chat($messages, false));

// Output the response
print_r($response);
?>

2. Streaming (Real-time Request)

<?php

// Ensure Composer autoloader is loaded
require './vendor/autoload.php';  // No need for full path when using Composer autoload

use CloudDark\OpenRouter\OpenRouter;  // Use the appropriate namespace

// Create an instance of OpenRouter for non-streaming
$openRouter = new OpenRouter(
    'sk-or-v1-XXX', // Your OpenRouter API token
    'google/gemini-2.0-flash-exp:free' // Model (optional, default is "mistralai/mixtral-8x7b-instruct")
);

// Prepare the message to send
$messages = [
    [
        'role' => 'system',
        'content' => 'You are an AI assistant.'
    ],
    [
        'role' => 'user',
        'content' => 'Hello, AI!'
    ]
];

// Send the request and get the response
$response = json_decode($openRouter->chat($messages, true));

// Output the response
print_r($response);
?>

Expected Response

The expected response from the OpenRouter API can vary based on the mode.

Non-Streaming Response:

The response for non-streaming will be a complete JSON object, similar to:

stdClass Object
(
    [id] => gen-1738686490-prk4jkf3l7In1GdYKoAI
    [provider] => DeepInfra
    [model] => mistralai/mixtral-8x7b-instruct
    [object] => chat.completion
    [created] => 1738686490
    [choices] => Array
        (
            [0] => stdClass Object
                (
                    [logprobs] =>
                    [finish_reason] => stop
                    [native_finish_reason] => stop
                    [index] => 0
                    [message] => stdClass Object
                        (
                            [role] => assistant
                            [content] =>  Hello! How can I assist you today? I'm here to help answer your questions, provide information, and facilitate any tasks you need help with. Let me know what you need.
                            [refusal] =>
                        )

                )

        )

    [usage] => stdClass Object
        (
            [prompt_tokens] => 19
            [completion_tokens] => 38
            [total_tokens] => 57
        )

)

Streaming Response:

In the case of streaming, the data will be received and output in real-time. The response will be directly displayed as it is processed, which could look like:

: OPENROUTER PROCESSING

data: {"id":"gen-1752162887-d4GeVQ1Wdg06gaa0YgwE","provider":"Google AI Studio","model":"google/gemini-2.0-flash-exp:free","object":"chat.completion.chunk","created":1752162887,"choices":[{"index":0,"delta":{"role":"assistant","content":"Hello"},"finish_reason":null,"native_finish_reason":null,"logprobs":null}]}

data: {"id":"gen-1752162887-d4GeVQ1Wdg06gaa0YgwE","provider":"Google AI Studio","model":"google/gemini-2.0-flash-exp:free","object":"chat.completion.chunk","created":1752162887,"choices":[{"index":0,"delta":{"role":"assistant","content":" there! How"},"finish_reason":null,"native_finish_reason":null,"logprobs":null}]}

data: {"id":"gen-1752162887-d4GeVQ1Wdg06gaa0YgwE","provider":"Google AI Studio","model":"google/gemini-2.0-flash-exp:free","object":"chat.completion.chunk","created":1752162887,"choices":[{"index":0,"delta":{"role":"assistant","content":" can I help you today?\n"},"finish_reason":"stop","native_finish_reason":"STOP","logprobs":null}]}

data: {"id":"gen-1752162887-d4GeVQ1Wdg06gaa0YgwE","provider":"Google AI Studio","model":"google/gemini-2.0-flash-exp:free","object":"chat.completion.chunk","created":1752162887,"choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":null,"native_finish_reason":null,"logprobs":null}],"usage":{"prompt_tokens":10,"completion_tokens":11,"total_tokens":21}}

data: [DONE]

The output will be displayed as chunks of data in the browser as they are received from the API.

Customization

You can customize the OpenRouter class by adjusting the URL, token, or model that you're using to interact with the OpenRouter API.

Set Token: setToken('your-api-token')
Set Model: setModel('your-model-id')
Enable/Disable Streaming: chat($messages, true) for streaming or chat($messages, false) for non-streaming.

License

This project is open-source and available under the MIT License.

cloud-dark / openrouter

Maintainers

Details