Providers

LabsLLM lets you easily switch between different LLM providers without having to rewrite your code, following the philosophy write once, run anywhere.

Supported Providers

Currently, LabsLLM supports the following providers:

Feature	OpenAI	Google	Anthropic
Text Prompts	✅	✅	📅
System Instructions	✅	✅	📅
Chat	✅	✅	📅
Tools/Functions	✅	✅	❌
Structure Output	✅	✅	❌
Streaming	✅	✅	❌
Embeddings	📅	❌	❌
Voice	📅	❌	❌
Image Generation	📅	❌	📅

Configuring a Provider

Each provider requires specific configuration, but the interface is consistent. You just need to specify which provider you want to use when creating your instance.

OpenAI

// Using OpenAI
$execute = LabsLLM::text()
    ->using(new OpenAI('your-api-key', 'gpt-4o'))
    ->executePrompt('Hello world');

Google

// Using Google
$execute = LabsLLM::text()
    ->using(new Google('your-api-key', 'gemini-pro'))
    ->executePrompt('Hello world');

Anthropic (Planned)

// Using Anthropic (future development)
$execute = LabsLLM::text()
    ->using(new Anthropic('your-api-key', 'claude-3-opus'))
    ->executePrompt('Hello world');

Switching Between Providers

The beauty of LabsLLM is that you can easily switch between providers without changing any other code in your application. This means you can:

1. Develop with one provider and deploy with another
2. Conduct A/B testing between different providers
3. Build redundancy into your application in case of provider unavailability

Practical Example

This code works with any supported provider, allowing you to easily switch:

// Initially using OpenAI
$provider = new OpenAI('your-api-key', 'gpt-4o');

// To switch to Google, just replace the line above with:
// $provider = new Google('your-api-key', 'gemini-pro');

// The rest of the code remains identical
$execute = LabsLLM::text()
    ->using($provider)
    ->executePrompt('What is the capital of Brazil?');

$response = $execute->getResponseData();
echo $response->response; // "The capital of Brazil is Brasília."

Provider-Specific Configurations

Each provider may have its own specific configurations, such as available models, request parameters, etc.

Available Models by Provider

Provider	Popular Models
OpenAI	gpt-4o, gpt-4-turbo, gpt-3.5-turbo
Google	gemini-pro, gemini-ultra
Anthropic	claude-3-opus, claude-3-sonnet, claude-3-haiku

Model and Parameter Configuration

// Using OpenAI with specific model and temperature
$execute = LabsLLM::text()
    ->using(new OpenAI('your-api-key', 'gpt-4o'))
    ->withTemperature(0.7) // Adjusts model creativity
    ->executePrompt('Tell me a story');

// Using Google with different settings
$execute = LabsLLM::text()
    ->using(new Google('your-api-key', 'gemini-pro'))
    ->withTopP(0.9) // Controls response diversity
    ->executePrompt('Tell me a story');

Dependency Injection

In real applications, consider using dependency injection to facilitate provider switching:

class AIService
{
    private $llm;
    
    public function __construct(ProviderInterface $provider)
    {
        $this->llm = LabsLLM::text()->using($provider);
    }
    
    public function ask($question)
    {
        $execute = $this->llm->executePrompt($question);
        return $execute->getResponseData()->response;
    }
}

// In your configuration file or service provider
$service = new AIService(new OpenAI('your-api-key', 'gpt-4o'));

// To switch to another provider, just change this line
// $service = new AIService(new Google('your-api-key', 'gemini-pro'));

Recommendations

• Start with OpenAI: The OpenAI implementation is the most complete and tested.
• Centralize Configuration: Keep your provider configuration in a centralized location for easy switching.
• Test with Multiple Providers: If possible, test your application with different providers to ensure compatibility.

Next Steps

Now that you understand how to configure and switch between providers, you can explore the different features of LabsLLM, such as Prompts, Chat, and Tools.