fix: Add the Vercel AI Gateway as a Provider

packages/sdk/src/types/ai-providers.ts

@@ -0,0 +1,114 @@
+/**
+ * @file This file defines the types and enums related to AI providers and their models,
+ * including pricing structures, to be consumed by the SDK.
+ */
+
+/**
+ * Enum for available AI providers.
+ * VercelAIGateway is added to support new integrations.
+ */
+export enum AIProvider {
+  OpenAI = 'openai',
+  Anthropic = 'anthropic',
+  Gemini = 'gemini',
+  Openrouter = 'openrouter',
+  VercelAIGateway = 'vercel_ai_gateway', // New provider
+}
+
+/**
+ * Represents the specific models available through the Vercel AI Gateway.
+ * This includes various chat models and placeholders for speech/transcription
+ * models, as Vercel AI Gateway can proxy different underlying services.
+ */
+export enum VercelAIGatewayModel {
+  // Common Chat Models proxied via Vercel AI Gateway
+  OpenAI_GPT3_5_Turbo = 'openai/gpt-3.5-turbo',
+  OpenAI_GPT4 = 'openai/gpt-4',
+  OpenAI_GPT4_Turbo = 'openai/gpt-4-turbo',
+  Anthropic_ClaudeInstant = 'anthropic/claude-instant-1.2',
+  Anthropic_Claude2_1 = 'anthropic/claude-2.1',
+  Google_GeminiPro = 'google/gemini-pro',
+
+  // Speech and Transcription Models (placeholders for future support)
+  OpenAI_Whisper = 'openai/whisper-1', // For transcription
+  ElevenLabs_Speech = 'elevenlabs/speech', // Example for TTS
+  Google_SpeechToText = 'google/speech-to-text', // Another example for transcription
+}
+
+/**
+ * Defines the pricing structure for a specific AI model from any provider.
+ * This structure supports both token-based and time-based (audio) pricing.
+ */
+export interface ModelPricing {
+  provider: AIProvider;
+  model: string; // The specific model identifier (e.g., 'gpt-3.5-turbo', 'openai/whisper-1')
+  pricing: {
+    /** Cost per million input tokens in USD. */
+    inputTokenCostPerMillion?: number;
+    /** Cost per million output tokens in USD. */
+    outputTokenCostPerMillion?: number;
+    /** Cost per minute of audio input (e.g., for transcription) in USD. */
+    audioInputCostPerMinute?: number;
+    /** Cost per minute of audio output (e.g., for text-to-speech) in USD. */
+    audioOutputCostPerMinute?: number;
+    // Add other potential pricing metrics here as needed (e.g., image generation costs)
+  };
+  /** Timestamp when the pricing data was last updated. */
+  lastUpdated: string;
+}
+
+/**
+ * A registry containing pricing information for various AI models across all providers.
+ * In a real application, this data would typically be loaded from a configuration service
+ * or a database, rather than being hardcoded.
+ */
+export type PricingRegistry = ModelPricing[];
+
+/**
+ * An example default pricing registry, demonstrating how Vercel AI Gateway models
+ * and their associated pricing would be captured.
+ */
+export const defaultPricingRegistry: PricingRegistry = [
+  // --- Existing Provider Pricing Examples (for context) ---
+  {
+    provider: AIProvider.OpenAI,
+    model: 'gpt-3.5-turbo',
+    pricing: { inputTokenCostPerMillion: 0.50, outputTokenCostPerMillion: 1.50 },
+    lastUpdated: '2023-10-26T10:00:00Z',
+  },
+  {
+    provider: AIProvider.Anthropic,
+    model: 'claude-2.1',
+    pricing: { inputTokenCostPerMillion: 8.00, outputTokenCostPerMillion: 24.00 },
+    lastUpdated: '2023-10-26T10:00:00Z',
+  },
+  // --- Vercel AI Gateway Pricing ---
+  {
+    provider: AIProvider.VercelAIGateway,
+    model: VercelAIGatewayModel.OpenAI_GPT3_5_Turbo,
+    // Vercel AI Gateway often proxies, so pricing might match underlying provider or have a small markup
+    pricing: {
+      inputTokenCostPerMillion: 0.55, // Example: slightly higher than direct OpenAI
+      outputTokenCostPerMillion: 1.65,
+    },
+    lastUpdated: '2023-10-27T12:00:00Z',
+  },
+  {
+    provider: AIProvider.VercelAIGateway,
+    model: VercelAIGatewayModel.OpenAI_GPT4_Turbo,
+    pricing: {
+      inputTokenCostPerMillion: 10.50,
+      outputTokenCostPerMillion: 31.50,
+    },
+    lastUpdated: '2023-10-27T12:00:00Z',
+  },
+  {
+    provider: AIProvider.VercelAIGateway,
+    model: VercelAIGatewayModel.OpenAI_Whisper,
+    pricing: {
+      audioInputCostPerMinute: 6.50, // Example: $0.0065 per minute for transcription
+    },
+    lastUpdated: '2023-10-27T12:00:00Z',
+  },
+  // Add more Vercel AI Gateway models and their respective pricing here as they are integrated.
+];