🧠 AI ProvidersOpenRouter

OpenRouter 🔀

Access 100+ AI models through a single unified API.

🌐

What you can do: Route requests to any AI model (OpenAI, Anthropic, Google, Meta, etc.), automatic fallbacks, provider-specific optimizations, and cost tracking - all through one API.

Setup

Add your OpenRouter API key in the ProtectMyAPI Dashboard.

Chat Completions

Basic Chat

let openRouter = ProtectMyAPI.openRouterService()
 
let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "anthropic/claude-3.5-sonnet",
        messages: [
            .system("You are a helpful assistant."),
            .user("Explain the difference between TCP and UDP")
        ]
    )
)
 
print(response.choices.first?.message.content ?? "")

Switch Between Models

Use the same code with any model:

// OpenAI
let gpt = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "openai/gpt-4-turbo",
        messages: [.user("Hello!")]
    )
)
 
// Google
let gemini = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "google/gemini-pro-1.5",
        messages: [.user("Hello!")]
    )
)
 
// Meta
let llama = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "meta-llama/llama-3.1-70b-instruct",
        messages: [.user("Hello!")]
    )
)

Streaming

for try await chunk in openRouter.createChatCompletionStream(
    request: OpenRouterChatRequest(
        model: "anthropic/claude-3.5-sonnet",
        messages: [.user("Write a detailed guide on Swift concurrency")]
    )
) {
    print(chunk.choices.first?.delta?.content ?? "", terminator: "")
}

Vision (Image Analysis)

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "anthropic/claude-3.5-sonnet", // or "openai/gpt-4-vision-preview"
        messages: [
            .user(content: [
                .text("What's in this image?"),
                .imageUrl(url: "https://example.com/image.jpg")
            ])
        ]
    )
)
 
print(response.choices.first?.message.content ?? "")

With Base64 Image

let imageData = // ... your image data
let base64 = imageData.base64EncodedString()
 
let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "openai/gpt-4-vision-preview",
        messages: [
            .user(content: [
                .text("Describe this image in detail"),
                .imageUrl(url: "data:image/jpeg;base64,\(base64)")
            ])
        ]
    )
)

Fallbacks & Routing

Automatic Fallbacks

Route to backup models if primary fails:

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "anthropic/claude-3.5-sonnet",
        messages: [.user("Hello!")],
        route: "fallback",
        models: [
            "anthropic/claude-3.5-sonnet",
            "openai/gpt-4-turbo",
            "google/gemini-pro-1.5"
        ]
    )
)

Provider Preferences

Specify preferred providers:

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "openai/gpt-4",
        messages: [.user("Hello!")],
        providerPreferences: OpenRouterProviderPreferences(
            order: ["OpenAI", "Azure"],
            allowFallbacks: true
        )
    )
)

Function Calling

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "openai/gpt-4-turbo",
        messages: [.user("What's the weather in Tokyo?")],
        tools: [
            OpenRouterTool(
                type: "function",
                function: OpenRouterFunction(
                    name: "get_weather",
                    description: "Get weather for a location",
                    parameters: [
                        "type": "object",
                        "properties": [
                            "location": ["type": "string"]
                        ],
                        "required": ["location"]
                    ]
                )
            )
        ]
    )
)
 
if let toolCall = response.choices.first?.message.toolCalls?.first {
    print("Function: \(toolCall.function.name)")
    print("Args: \(toolCall.function.arguments)")
}

JSON Mode

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "openai/gpt-4-turbo",
        messages: [
            .system("Output valid JSON only."),
            .user("List 3 fruits with colors")
        ],
        responseFormat: OpenRouterResponseFormat(type: "json_object")
    )
)

List Available Models

let models = try await openRouter.listModels()
 
for model in models.data {
    print("\(model.id): \(model.name)")
    print("  Context: \(model.contextLength)")
    print("  Pricing: \(model.pricing.prompt)/\(model.pricing.completion)")
}

Popular Models

Premium Models

ModelProviderContextBest For
anthropic/claude-3.5-sonnetAnthropic200KCoding, analysis
openai/gpt-4-turboOpenAI128KGeneral purpose
openai/gpt-4oOpenAI128KFast, multimodal
google/gemini-pro-1.5Google1MLong context

Open Source Models

ModelProviderContextBest For
meta-llama/llama-3.1-405b-instructMeta128KHighest open quality
meta-llama/llama-3.1-70b-instructMeta128KBalance
mistralai/mixtral-8x22b-instructMistral65KMoE efficiency

Vision Models

ModelDescription
openai/gpt-4-vision-previewGPT-4 with vision
anthropic/claude-3.5-sonnetClaude with vision
google/gemini-pro-visionGemini with vision

Free Models

ModelDescription
meta-llama/llama-3.1-8b-instruct:freeFree Llama 8B
google/gemma-7b-it:freeFree Gemma
mistralai/mistral-7b-instruct:freeFree Mistral

App Information

Include app info for analytics:

let response = try await openRouter.createChatCompletion(
    request: OpenRouterChatRequest(
        model: "anthropic/claude-3.5-sonnet",
        messages: [.user("Hello!")],
        httpReferer: "https://myapp.com",
        xTitle: "My Mobile App"
    )
)

Pricing

OpenRouter passes through provider pricing with a small markup (~5%). Benefits:

  • Single bill for all providers
  • Automatic currency handling
  • Credit-based system
  • No separate accounts needed

Check their models page for current pricing.

Fireworks AIBrave Search