Provider Overview

Prism supports 12 different LLM providers, each with unique capabilities and features. This page helps you choose the right provider for your needs.

Supported Providers

Prism integrates with the following providers:

Anthropic

Claude models with extended thinking, prompt caching, and citations

OpenAI

GPT models with reasoning, image generation, and audio processing

Google Gemini

Multimodal models with search grounding and video analysis

Ollama

Local model deployment for privacy and offline use

Mistral

Magistral reasoning models with OCR capabilities

Groq

Ultra-fast LPU inference for real-time applications

DeepSeek

Efficient models optimized for coding tasks

xAI

Grok models with extended thinking capabilities

OpenRouter

Access multiple providers through a single API

ElevenLabs

Speech-to-text with diarization and event tagging

VoyageAI

Specialized embeddings for search and retrieval

Feature Comparison

Compare capabilities across providers to find the best fit for your use case:

Core Features

Provider	Text Generation	Streaming	Structured Output	Tool Calling	Embeddings
Anthropic	✅	✅	✅ (Native + Tool)	✅	❌
OpenAI	✅	✅	✅ (Strict mode)	✅	✅
Gemini	✅	✅	✅ (Native)	✅	✅
Ollama	✅	✅	✅ (Prompt-based)	✅	✅
Mistral	✅	✅	✅	✅	✅
Groq	✅	✅	✅	✅	❌
DeepSeek	✅	✅	✅	✅	❌
xAI	✅	✅	✅ (Strict mode)	✅	❌
OpenRouter	✅	✅	✅	✅	❌
ElevenLabs	❌	❌	❌	❌	❌
VoyageAI	❌	❌	❌	❌	✅

Multimodal Support

Provider	Images	Audio (STT)	Audio (TTS)	Video	Documents	Image Generation
Anthropic	✅	❌	❌	❌	✅ (PDF)	❌
OpenAI	✅	✅ (Whisper)	✅	❌	❌	✅ (DALL-E)
Gemini	✅	✅	❌	✅ (YouTube)	✅	✅ (Imagen)
Ollama	✅	❌	❌	❌	❌	❌
Mistral	✅	✅ (Voxtral)	❌	❌	✅ (URL)	❌
Groq	✅	✅ (Whisper)	✅ (PlayAI)	❌	❌	❌
DeepSeek	❌	❌	❌	❌	❌	❌
xAI	✅	❌	❌	❌	❌	❌
OpenRouter	✅	❌	❌	✅	✅	❌
ElevenLabs	❌	✅ (Scribe)	❌	❌	❌	❌
VoyageAI	❌	❌	❌	❌	❌	❌

Advanced Capabilities

Provider	Reasoning/Thinking	Prompt Caching	Citations	Code Execution	Search Grounding	Moderation
Anthropic	✅ (Extended)	✅ (Ephemeral)	✅	✅	❌	❌
OpenAI	✅ (GPT-5)	✅ (Auto)	❌	✅	❌	✅
Gemini	✅ (Thinking)	✅	✅	❌	✅ (Google)	❌
Ollama	❌	❌	❌	❌	❌	❌
Mistral	✅ (Magistral)	❌	❌	❌	❌	❌
Groq	❌	❌	❌	❌	❌	❌
DeepSeek	❌	❌	❌	❌	❌	❌
xAI	✅ (Extended)	❌	❌	❌	❌	❌
OpenRouter	✅ (Model-dependent)	❌	❌	❌	❌	❌
ElevenLabs	❌	❌	❌	❌	❌	❌
VoyageAI	❌	❌	❌	❌	❌	❌

Configuration

All providers are configured in config/prism.php:

return [
    'providers' => [
        'openai' => [
            'url' => env('OPENAI_URL', 'https://api.openai.com/v1'),
            'api_key' => env('OPENAI_API_KEY', ''),
            'organization' => env('OPENAI_ORGANIZATION', null),
            'project' => env('OPENAI_PROJECT', null),
        ],
        'anthropic' => [
            'api_key' => env('ANTHROPIC_API_KEY', ''),
            'version' => env('ANTHROPIC_API_VERSION', '2023-06-01'),
            'url' => env('ANTHROPIC_URL', 'https://api.anthropic.com/v1'),
            'default_thinking_budget' => env('ANTHROPIC_DEFAULT_THINKING_BUDGET', 1024),
            'anthropic_beta' => env('ANTHROPIC_BETA', null),
        ],
        'gemini' => [
            'api_key' => env('GEMINI_API_KEY', ''),
            'url' => env('GEMINI_URL', 'https://generativelanguage.googleapis.com/v1beta/models'),
        ],
        // ... other providers
    ],
];

Choosing a Provider

For Production Applications

Best Overall
Multimodal
Speed
Privacy

Anthropic Claude - Industry-leading reasoning, long context windows, and reliable structured output with prompt caching for cost optimization.

Prism::text()
    ->using('anthropic', 'claude-sonnet-4-5-20250929')
    ->withPrompt('Your prompt')
    ->asText();

Google Gemini - Best for video analysis, search grounding, and multimodal applications with native structured output.

Prism::text()
    ->using('gemini', 'gemini-2.5-flash')
    ->withPrompt('Analyze this video', [Video::fromUrl($url)])
    ->asText();

Groq - Ultra-fast inference with LPU architecture, ideal for real-time applications and high-throughput workloads.

Prism::text()
    ->using('groq', 'llama-3.3-70b-versatile')
    ->withPrompt('Your prompt')
    ->asText();

Ollama - Run models locally for complete data privacy and offline operation.

Prism::text()
    ->using('ollama', 'llama3.2')
    ->withPrompt('Your prompt')
    ->asText();

For Specific Use Cases

Complex Reasoning Tasks

Use providers with reasoning/thinking capabilities:

Anthropic Claude 3.7 - Extended thinking with budget control
OpenAI GPT-5 - Reasoning effort levels (low/medium/high)
Mistral Magistral - Efficient reasoning for complex problems
xAI Grok - Extended thinking mode

Prism::text()
    ->using('anthropic', 'claude-3-7-sonnet-latest')
    ->withProviderOptions(['thinking' => ['enabled' => true]])
    ->withPrompt('Solve this complex problem...')
    ->asText();

Audio Processing

Speech-to-Text:

OpenAI Whisper (via OpenAI/Groq) - Industry standard
Mistral Voxtral - Multilingual with 30min support
ElevenLabs Scribe - Diarization and event tagging

Text-to-Speech:

OpenAI TTS - Natural voices with HD quality
Groq PlayAI - Fast TTS with Arabic support

Image Generation

OpenAI DALL-E 3 - Highest quality with prompt revision
OpenAI GPT-Image-1 - Advanced editing with masks
Gemini Imagen 4 - HD generation with aspect ratio control

Embeddings

VoyageAI - Specialized for search/retrieval tasks
OpenAI - General purpose embeddings
Gemini - Task-specific embeddings

Cost Optimization

Anthropic - Prompt caching (5m/1h TTL)
OpenAI - Automatic caching for structured output
Gemini - Content caching with custom TTL

Multiple Providers

OpenRouter - Access multiple providers through a single API with automatic fallback routing.

Prism::text()
    ->using('openrouter', 'openai/gpt-4o')
    ->withProviderOptions([
        'models' => ['anthropic/claude-sonnet-4.5', 'openai/gpt-4o-mini']
    ])
    ->asText();

Provider Reliability

Structured Output Reliability

Native Support (Most Reliable):

Anthropic (Claude Sonnet 4.5+, Opus 4.1+)
OpenAI (with strict mode)
Gemini (native structured output)
xAI (strict mode)

Tool Calling Mode (Reliable):

Anthropic (non-native models)
OpenRouter

Prompt-based (Less Reliable):

Ollama
DeepSeek

Message Order Requirements

Anthropic is strict about message order: must alternate User → Assistant → User. Other providers are more flexible.

Quick Start by Provider

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

// Anthropic Claude
$response = Prism::text()
    ->using(Provider::Anthropic, 'claude-sonnet-4-5-20250929')
    ->withPrompt('Hello!')
    ->asText();

// OpenAI GPT
$response = Prism::text()
    ->using(Provider::OpenAI, 'gpt-4o')
    ->withPrompt('Hello!')
    ->asText();

// Google Gemini
$response = Prism::text()
    ->using(Provider::Gemini, 'gemini-2.5-flash')
    ->withPrompt('Hello!')
    ->asText();

// Ollama (Local)
$response = Prism::text()
    ->using(Provider::Ollama, 'llama3.2')
    ->withClientOptions(['timeout' => 60])
    ->withPrompt('Hello!')
    ->asText();

// Mistral
$response = Prism::text()
    ->using(Provider::Mistral, 'mistral-large-latest')
    ->withPrompt('Hello!')
    ->asText();

// Groq (Fast)
$response = Prism::text()
    ->using(Provider::Groq, 'llama-3.3-70b-versatile')
    ->withPrompt('Hello!')
    ->asText();

// DeepSeek
$response = Prism::text()
    ->using(Provider::DeepSeek, 'deepseek-chat')
    ->withPrompt('Hello!')
    ->asText();

// xAI Grok
$response = Prism::text()
    ->using(Provider::XAI, 'grok-4')
    ->withPrompt('Hello!')
    ->asText();

// OpenRouter (Multi-provider)
$response = Prism::text()
    ->using(Provider::OpenRouter, 'openai/gpt-4o')
    ->withPrompt('Hello!')
    ->asText();

Next Steps

Provider Details

Explore detailed documentation for each provider

Core Concepts

Learn about text generation, streaming, and tools

Structured Output

Generate type-safe structured data

Multimodal Input

Work with images, audio, video, and documents

Supported Providers

​Supported Providers

Anthropic

OpenAI

Google Gemini

Ollama

Mistral

Groq

DeepSeek

xAI

OpenRouter

ElevenLabs

VoyageAI

​Feature Comparison

​Core Features

​Multimodal Support

​Advanced Capabilities

​Configuration

​Choosing a Provider

​For Production Applications

​For Specific Use Cases

​Provider Reliability

​Structured Output Reliability

​Message Order Requirements

​Quick Start by Provider

​Next Steps

Provider Details

Core Concepts

Structured Output

Multimodal Input

Build docs developers (and LLMs) love

Supported Providers

Feature Comparison

Core Features

Multimodal Support

Advanced Capabilities

Configuration

Choosing a Provider

For Production Applications

For Specific Use Cases

Provider Reliability

Structured Output Reliability

Message Order Requirements

Quick Start by Provider

Next Steps