AI Providers

SmartRAG AI provider configuration - OpenAI, Anthropic, Google Gemini, Azure OpenAI and custom providers

AI Provider Configuration

SmartRAG supports various AI providers:

OpenAI

OpenAI provides advanced language models and embeddings for production-ready applications:

{
  "AI": {
    "OpenAI": {
      "ApiKey": "sk-proj-YOUR_KEY",
      "Endpoint": "https://api.openai.com/v1",
      "Model": "gpt-5.1",
      "EmbeddingModel": "text-embedding-3-small",
      "MaxTokens": 4096,
      "Temperature": 0.7
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.AIProvider = AIProvider.OpenAI;
});

Models:

  • gpt-5.1 - Latest advanced reasoning model (recommended)
  • gpt-5 - Advanced reasoning capabilities
  • gpt-4o - Previous generation advanced model
  • gpt-4o-mini - Cost-effective and fast
  • text-embedding-3-small, text-embedding-3-large - Embeddings (recommended)
  • text-embedding-ada-002 - Legacy embeddings

Anthropic (Claude)

Important: VoyageAI Required

Anthropic Claude models require a separate VoyageAI API key for embeddings since Anthropic doesn't provide embedding models. 

{
  "AI": {
    "Anthropic": {
      "ApiKey": "sk-ant-ANTHROPIC_KEY",
      "Model": "claude-sonnet-4-5",
      "MaxTokens": 4096,
      "Temperature": 0.3,
      "EmbeddingApiKey": "pa-VOYAGE_KEY",
      "EmbeddingModel": "voyage-3.5"
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.AIProvider = AIProvider.Anthropic;
});

Claude Models:

  • claude-sonnet-4-5 - Latest and most intelligent (recommended)
  • claude-3.5-sonnet - Previous generation
  • claude-3-opus-20240229 - Highest capability
  • claude-3-haiku-20240307 - Fastest

VoyageAI Embedding Models:

  • voyage-3.5 - High quality (recommended)
  • voyage-code-2 - Optimized for code
  • voyage-2 - General purpose

Google Gemini

Google Gemini offers cost-effective AI models with multimodal capabilities:

{
  "AI": {
    "Gemini": {
      "ApiKey": "GEMINI_KEY",
      "Model": "gemini-2.5-pro",
      "EmbeddingModel": "embedding-001",
      "MaxTokens": 4096,
      "Temperature": 0.7
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.AIProvider = AIProvider.Gemini;
});

Models:

  • gemini-2.5-pro - Advanced reasoning capabilities (recommended)
  • gemini-2.5-flash - Fast and cost-effective
  • gemini-2.0-flash - Previous generation workhorse
  • gemini-1.5-pro - Legacy advanced model
  • embedding-001 - Text embeddings

Azure OpenAI

Azure OpenAI provides enterprise-grade AI services with enhanced security and compliance:

{
  "AI": {
    "AzureOpenAI": {
      "ApiKey": "AZURE_KEY",
      "Endpoint": "https://your-resource.openai.azure.com/",
      "Model": "gpt-5.1",
      "EmbeddingModel": "text-embedding-3-small",
      "DeploymentName": "gpt-5.1-deployment",
      "MaxTokens": 4096,
      "Temperature": 0.7
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.AIProvider = AIProvider.AzureOpenAI;
});

Custom Provider (Ollama / LM Studio)

Ollama / LM Studio with 100% On-Premise AI

Run AI models completely on-premise for complete data privacy - perfect for enterprise deployments, GDPR/HIPAA compliance.

Ollama (On-premise Models)

{
  "AI": {
    "Custom": {
      "ApiKey": "not-required",
      "Endpoint": "http://localhost:11434/v1/chat/completions",
      "Model": "llama2",
      "EmbeddingModel": "nomic-embed-text"
    }
  }
}

LM Studio (On-premise Models)

{
  "AI": {
    "Custom": {
      "ApiKey": "not-required",
      "Endpoint": "http://localhost:1234/v1/chat/completions",
      "Model": "local-model",
      "EmbeddingModel": "local-embedding"
    }
  }
}

builder.Services.AddSmartRag(configuration, options =>
{
    options.AIProvider = AIProvider.Custom;
});

Supported Custom APIs:

  • 🦙 Ollama - On-premise models
  • 🏠 LM Studio - On-premise AI environment
  • 🔗 OpenRouter - Access to 100+ models
  • ⚡ Groq - Lightning-fast inference
  • 🌐 Together AI - Open source models
  • Any OpenAI-compatible API

Provider Comparison

Compare AI providers to choose the best option for your use case:

Provider Strengths Weaknesses Best Use Case
OpenAI Advanced models, reliable Expensive, data privacy concerns Production, critical applications
Anthropic Security-focused, quality output VoyageAI required, limited access Security-critical applications
Google Gemini Cost-effective, multimodal Limited production support Prototyping, development
Azure OpenAI Enterprise security, SLA Complex setup Enterprise applications
Ollama/LM Studio 100% on-premise, free Performance limitations Data privacy critical

Next Steps

Storage Providers

Qdrant, Redis, SQLite and other storage options

Storage Providers

Database Configuration

Multi-database connections and schema analysis

Database Configuration