3 files changed, 24 insertions, 23 deletions
diff --git a/packages/core/src/utils/nextSpeakerChecker.test.ts b/packages/core/src/utils/nextSpeakerChecker.test.ts
index 83ce97fd..475b5662 100644
--- a/packages/core/src/utils/nextSpeakerChecker.test.ts
+++ b/packages/core/src/utils/nextSpeakerChecker.test.ts
@@ -71,7 +71,6 @@ describe('checkNextSpeaker', () => {
     chatInstance = new GeminiChat(
       mockConfigInstance,
       mockModelsInstance, // This is the instance returned by mockGoogleGenAIInstance.getGenerativeModel
-      'gemini-pro', // model name
       {},
       [], // initial history
     );
diff --git a/packages/core/src/utils/retry.test.ts b/packages/core/src/utils/retry.test.ts
index 1988c02a..a0294c31 100644
--- a/packages/core/src/utils/retry.test.ts
+++ b/packages/core/src/utils/retry.test.ts
@@ -272,7 +272,7 @@ describe('retryWithBackoff', () => {
       expect(fallbackCallback).toHaveBeenCalledWith('oauth-personal');
 
       // Should retry again after fallback
-      expect(mockFn).toHaveBeenCalledTimes(4); // 3 initial attempts + 1 after fallback
+      expect(mockFn).toHaveBeenCalledTimes(3); // 2 initial attempts + 1 after fallback
     });
 
     it('should NOT trigger fallback for API key users', async () => {
diff --git a/packages/core/src/utils/retry.ts b/packages/core/src/utils/retry.ts
index ebe18510..372a7976 100644
--- a/packages/core/src/utils/retry.ts
+++ b/packages/core/src/utils/retry.ts
@@ -67,9 +67,9 @@ export async function retryWithBackoff<T>(
     maxAttempts,
     initialDelayMs,
     maxDelayMs,
-    shouldRetry,
     onPersistent429,
     authType,
+    shouldRetry,
   } = {
     ...DEFAULT_RETRY_OPTIONS,
     ...options,
@@ -93,28 +93,30 @@ export async function retryWithBackoff<T>(
         consecutive429Count = 0;
       }
 
-      // Check if we've exhausted retries or shouldn't retry
-      if (attempt >= maxAttempts || !shouldRetry(error as Error)) {
-        // If we have persistent 429s and a fallback callback for OAuth
-        if (
-          consecutive429Count >= 2 &&
-          onPersistent429 &&
-          authType === AuthType.LOGIN_WITH_GOOGLE_PERSONAL
-        ) {
-          try {
-            const fallbackModel = await onPersistent429(authType);
-            if (fallbackModel) {
-              // Reset attempt counter and try with new model
-              attempt = 0;
-              consecutive429Count = 0;
-              currentDelay = initialDelayMs;
-              continue;
-            }
-          } catch (fallbackError) {
-            // If fallback fails, continue with original error
-            console.warn('Fallback to Flash model failed:', fallbackError);
+      // If we have persistent 429s and a fallback callback for OAuth
+      if (
+        consecutive429Count >= 2 &&
+        onPersistent429 &&
+        authType === AuthType.LOGIN_WITH_GOOGLE_PERSONAL
+      ) {
+        try {
+          const fallbackModel = await onPersistent429(authType);
+          if (fallbackModel) {
+            // Reset attempt counter and try with new model
+            attempt = 0;
+            consecutive429Count = 0;
+            currentDelay = initialDelayMs;
+            // With the model updated, we continue to the next attempt
+            continue;
           }
+        } catch (fallbackError) {
+          // If fallback fails, continue with original error
+          console.warn('Fallback to Flash model failed:', fallbackError);
         }
+      }
+
+      // Check if we've exhausted retries or shouldn't retry
+      if (attempt >= maxAttempts || !shouldRetry(error as Error)) {
         throw error;
       }