fix: Support previous v4 of Vercel AI SDK (#962)

fix: Fix metric tracking for v5 responses

Author: jsonbailey

packages/ai-providers/server-ai-vercel/__tests__/VercelProvider.test.ts

@@ -68,6 +68,52 @@ describe('VercelProvider', () => {
         },
       });
     });
+
+    it('supports v5 field names (inputTokens, outputTokens)', () => {
+      const mockResponse = {
+        usage: {
+          inputTokens: 40,
+          outputTokens: 60,
+          totalTokens: 100,
+        },
+      };
+
+      const result = VercelProvider.createAIMetrics(mockResponse);
+
+      expect(result).toEqual({
+        success: true,
+        usage: {
+          total: 100,
+          input: 40,
+          output: 60,
+        },
+      });
+    });
+
+    it('prefers v5 field names over v4 when both are present', () => {
+      const mockResponse = {
+        usage: {
+          // v4 field names
+          promptTokens: 10,
+          completionTokens: 20,
+          // v5 field names (should be preferred)
+          inputTokens: 40,
+          outputTokens: 60,
+          totalTokens: 100,
+        },
+      };
+
+      const result = VercelProvider.createAIMetrics(mockResponse);
+
+      expect(result).toEqual({
+        success: true,
+        usage: {
+          total: 100,
+          input: 40, // inputTokens preferred over promptTokens
+          output: 60, // outputTokens preferred over completionTokens
+        },
+      });
+    });
   });
 
   describe('invokeModel', () => {

packages/ai-providers/server-ai-vercel/package.json

@@ -32,7 +32,6 @@
     "@ai-sdk/google": "^2.0.0",
     "@ai-sdk/mistral": "^2.0.0",
     "@ai-sdk/openai": "^2.0.0",
-    "@ai-sdk/provider": "^2.0.0",
     "@launchdarkly/server-sdk-ai": "^0.12.2",
     "@trivago/prettier-plugin-sort-imports": "^4.1.1",
     "@types/jest": "^29.5.3",
@@ -57,7 +56,6 @@
     "@ai-sdk/google": "^2.0.0",
     "@ai-sdk/mistral": "^2.0.0",
     "@ai-sdk/openai": "^2.0.0",
-    "@ai-sdk/provider": "^2.0.0",
     "@launchdarkly/server-sdk-ai": "^0.12.2",
     "ai": "^4.0.0 || ^5.0.0"
   },

packages/ai-providers/server-ai-vercel/src/VercelProvider.ts

@@ -1,5 +1,4 @@
-import { LanguageModelV2 } from '@ai-sdk/provider';
-import { generateText } from 'ai';
+import { generateText, LanguageModel } from 'ai';
 
 import { AIProvider } from '@launchdarkly/server-sdk-ai';
 import type {
@@ -16,10 +15,10 @@ import type {
  * This provider integrates Vercel AI SDK with LaunchDarkly's tracking capabilities.
  */
 export class VercelProvider extends AIProvider {
-  private _model: LanguageModelV2;
+  private _model: LanguageModel;
   private _parameters: Record<string, unknown>;
 
-  constructor(model: LanguageModelV2, parameters: Record<string, unknown>, logger?: LDLogger) {
+  constructor(model: LanguageModel, parameters: Record<string, unknown>, logger?: LDLogger) {
     super(logger);
     this._model = model;
     this._parameters = parameters;
@@ -47,6 +46,8 @@ export class VercelProvider extends AIProvider {
    */
   async invokeModel(messages: LDMessage[]): Promise<ChatResponse> {
     // Call Vercel AI generateText
+    // Type assertion: our MinLanguageModel is compatible with the expected LanguageModel interface
+    // The generateText function will work with any object that has the required properties
     const result = await generateText({
       model: this._model,
       messages,
@@ -71,7 +72,7 @@ export class VercelProvider extends AIProvider {
   /**
    * Get the underlying Vercel AI model instance.
    */
-  getModel(): LanguageModelV2 {
+  getModel(): LanguageModel {
     return this._model;
   }
 
@@ -98,16 +99,18 @@ export class VercelProvider extends AIProvider {
    * Create AI metrics information from a Vercel AI response.
    * This method extracts token usage information and success status from Vercel AI responses
    * and returns a LaunchDarkly AIMetrics object.
+   * Supports both v4 and v5 field names for backward compatibility.
    */
   static createAIMetrics(vercelResponse: any): LDAIMetrics {
     // Extract token usage if available
     let usage: LDTokenUsage | undefined;
     if (vercelResponse?.usage) {
-      const { promptTokens, completionTokens, totalTokens } = vercelResponse.usage;
+      const { totalTokens, inputTokens, promptTokens, outputTokens, completionTokens } =
+        vercelResponse.usage;
       usage = {
-        total: totalTokens || 0,
-        input: promptTokens || 0,
-        output: completionTokens || 0,
+        total: totalTokens ?? 0,
+        input: inputTokens ?? promptTokens ?? 0,
+        output: outputTokens ?? completionTokens ?? 0,
       };
     }
 
@@ -125,7 +128,7 @@ export class VercelProvider extends AIProvider {
    * @param aiConfig The LaunchDarkly AI configuration
    * @returns A Promise that resolves to a configured Vercel AI model
    */
-  static async createVercelModel(aiConfig: LDAIConfig): Promise<LanguageModelV2> {
+  static async createVercelModel(aiConfig: LDAIConfig): Promise<LanguageModel> {
     const providerName = VercelProvider.mapProvider(aiConfig.provider?.name || '');
     const modelName = aiConfig.model?.name || '';
     // Parameters are not used in model creation but kept for future use
@@ -143,28 +146,28 @@ export class VercelProvider extends AIProvider {
         }
       case 'anthropic':
         try {
-          const { anthropic } = await import('@ai-sdk/anthropic' as any);
+          const { anthropic } = await import('@ai-sdk/anthropic');
           return anthropic(modelName);
         } catch (error) {
           throw new Error(`Failed to load @ai-sdk/anthropic: ${error}`);
         }
       case 'google':
         try {
-          const { google } = await import('@ai-sdk/google' as any);
+          const { google } = await import('@ai-sdk/google');
           return google(modelName);
         } catch (error) {
           throw new Error(`Failed to load @ai-sdk/google: ${error}`);
         }
       case 'cohere':
         try {
-          const { cohere } = await import('@ai-sdk/cohere' as any);
+          const { cohere } = await import('@ai-sdk/cohere');
           return cohere(modelName);
         } catch (error) {
           throw new Error(`Failed to load @ai-sdk/cohere: ${error}`);
         }
       case 'mistral':
         try {
-          const { mistral } = await import('@ai-sdk/mistral' as any);
+          const { mistral } = await import('@ai-sdk/mistral');
           return mistral(modelName);
         } catch (error) {
           throw new Error(`Failed to load @ai-sdk/mistral: ${error}`);