Introduction

Prism provides a unified PHP interface for language-model APIs in Laravel applications. Use it to generate text, stream responses, call tools and work with multimodal input.

Generate text by selecting a provider and model:

use Prism\Prism\Facades\Prism;
use Prism\Prism\Enums\Provider;

$response = Prism::text()
    ->using(Provider::Anthropic, 'claude-sonnet-5-5')
    ->withSystemPrompt('Explain concepts in plain language.')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

echo $response->text;

use Prism\Prism\Facades\Prism;
use Prism\Prism\Enums\Provider;

$response = Prism::text()
    ->using(Provider::Mistral, 'mistral-medium-3-5')
    ->withSystemPrompt('Explain concepts in plain language.')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

echo $response->text;

use Prism\Prism\Facades\Prism;
use Prism\Prism\Enums\Provider;

$response = Prism::text()
    ->using(Provider::Ollama, 'llama3.2')
    ->withSystemPrompt('Explain concepts in plain language.')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

echo $response->text;

use Prism\Prism\Facades\Prism;
use Prism\Prism\Enums\Provider;

$response = Prism::text()
    ->using(Provider::OpenAI, 'gpt-4.1')
    ->withSystemPrompt('Explain concepts in plain language.')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

echo $response->text;

use Prism\Prism\Facades\Prism;
use Prism\Prism\Enums\Provider;

$response = Prism::text()
    ->using(Provider::Replicate, 'meta/meta-llama-3.1-405b-instruct')
    ->withSystemPrompt('Explain concepts in plain language.')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

echo $response->text;

Prism's API is inspired by the Vercel AI SDK.

Key Features#

• Text and conversations, structured output, and streaming.
• Custom and provider tools, including client execution and approvals on supported integrations.
• Embeddings, image generation/editing, audio processing, and moderation.
• Images, documents, audio and video as model inputs.
• Files, Batch, and Mistral fill-in-the-middle.
• Response fakes, telemetry events, and an optional OpenAI-compatible server.

Support varies by provider and model; a shared method does not imply that every integration supports it. The Ollama example requires that the named model is installed on your server.

Prism also provides a fluent prism() helper function to resolve the Prism instance from the application container.

prism()
    ->text()
    ->using(Provider::OpenAI, 'gpt-4.1')
    ->withPrompt('Explain quantum computing to a 5-year-old.')
    ->asText();

Providers#

Prism includes integrations for the following providers:

• Anthropic
• Azure OpenAI
• DeepSeek
• ElevenLabs
• Gemini
• Groq
• Mistral
• Ollama
• OpenAI
• OpenRouter
• Perplexity
• Qwen
• Replicate
• Requesty
• Vertex AI
• VoyageAI
• xAI
• Z

Provider Support#

See each provider's page for supported features, configuration and limitations. Availability also depends on the selected model.

✅ supported · 🟡 planned · ⚙️ adapted · — not supported