LLM Providers Overview

​

Supported Providers

​

Quick Comparison

​

Best for Production

• OpenAI: Industry-leading models with extensive tooling
• Anthropic: Strong safety guarantees and reliability
• Bedrock: Enterprise compliance and AWS ecosystem

​

Best for Development

• Ollama: Free local testing with popular open-source models
• OpenRouter: Test multiple providers without managing API keys

​

Best for Cost

• DeepSeek: Competitive pricing with good quality
• OpenRouter: Route to cheapest provider for each request
• Ollama: Completely free for local deployment

​

Best for Specific Features

• Vision: Google (Gemini), OpenAI (GPT-4o)
• Long Context: Google (1M tokens), Anthropic (200K tokens)
• Reasoning: OpenAI (o-series), Anthropic (Claude Opus)
• Code Generation: OpenAI (GPT-5 Codex), DeepSeek

​

Common Usage Patterns

​

Multi-Provider Strategy

val productionExecutor = simpleOpenAIExecutor(
    apiKey = System.getenv("OPENAI_API_KEY"),
    model = OpenAIModels.Chat.GPT4o
)

val devExecutor = simpleOllamaExecutor(
    baseUrl = "http://localhost:11434",
    model = OllamaModels.Meta.LLAMA_3_2
)

val executor = if (inProduction) productionExecutor else devExecutor

​

Fallback Chain with OpenRouter

val executor = simpleOpenRouterExecutor(
    apiKey = System.getenv("OPENROUTER_API_KEY"),
    model = LLModel(
        provider = LLMProvider.OpenRouter,
        id = "anthropic/claude-3.5-sonnet"
    ),
    params = OpenRouterParams(
        // Fallback to cheaper model on failure
        models = listOf(
            "anthropic/claude-3.5-sonnet",
            "anthropic/claude-3-haiku"
        )
    )
)

​

Provider Selection Guide

​

Choose OpenAI if:

• You need the most capable models available
• You require advanced features (function calling, structured outputs)
• You’re building a production application with budget for API costs
• You need extensive documentation and community support

​

Choose Anthropic if:

• Long context windows are critical (200K tokens)
• You need strong reasoning and analysis capabilities
• Safety and refusal behavior are important
• Document processing is a primary use case

​

Choose Google if:

• You need multimodal capabilities (vision, audio)
• Very long context is required (1M tokens)
• You want to process videos or images
• You’re already in the Google Cloud ecosystem

​

Choose Ollama if:

• You’re developing locally without API costs
• Privacy and data sovereignty are critical
• You need offline capability
• You want to experiment with open-source models

​

Choose OpenRouter if:

• You want access to many providers through one API
• You need automatic failover between providers
• Cost optimization across providers is important
• You want to experiment without multiple API keys

​

Choose DeepSeek if:

• Budget is a primary constraint
• You need good performance at lower cost
• Code generation is a primary use case
• You want an alternative to major providers

​

Choose Bedrock if:

• You need enterprise compliance (HIPAA, SOC2)
• You’re already using AWS services
• You require AWS Guardrails for content filtering
• Multiple model families with consistent API appeal to you

​

Feature Support Matrix

​

Next Steps