diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_token_usage.py b/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_token_usage.py
deleted file mode 100644
index c8894ca7d5..0000000000
--- a/tests/test_litellm/llms/vertex_ai/gemini/test_gemini_token_usage.py
+++ /dev/null
@@ -1,70 +0,0 @@
-
-import sys, os
-import pytest
-sys.path.insert(0, os.path.abspath('../../../../../'))
-
-from litellm.llms.vertex_ai.gemini.vertex_and_google_ai_studio_gemini import VertexGeminiConfig
-from litellm.types.llms.vertex_ai import UsageMetadata
-
-def test_gemini_3_flash_preview_token_usage_fallback():
-    """Test fallback logic when candidatesTokensDetails is missing (e.g. Gemini 3 Flash Preview)."""
-    v = VertexGeminiConfig()
-    
-    usage_metadata_dict = {
-        "promptTokenCount": 2145,
-        "candidatesTokenCount": 509,
-        "totalTokenCount": 2654,
-        # candidatesTokensDetails intentionally omitted
-    }
-    
-    completion_response = {"usageMetadata": usage_metadata_dict}
-    result = v._calculate_usage(completion_response=completion_response)
-    
-    assert result.completion_tokens == 509
-    assert result.prompt_tokens == 2145
-    assert result.total_tokens == 2654
-    
-    # Text tokens should be derived from candidatesTokenCount
-    assert result.completion_tokens_details is not None
-    assert result.completion_tokens_details.text_tokens == 509
-    assert result.completion_tokens_details.image_tokens is None
-    assert result.completion_tokens_details.audio_tokens is None
-
-def test_gemini_no_reasoning_fallback():
-    """Test fallback when reasoning_effort is absent and details are missing."""
-    v = VertexGeminiConfig()
-    
-    usage_metadata_dict = {
-        "promptTokenCount": 100,
-        "candidatesTokenCount": 264,
-        "totalTokenCount": 364,
-    }
-    
-    completion_response = {"usageMetadata": usage_metadata_dict}
-    result = v._calculate_usage(completion_response=completion_response)
-    
-    assert result.completion_tokens == 264
-    assert result.completion_tokens_details is not None
-    assert result.completion_tokens_details.text_tokens == 264
-    assert result.completion_tokens_details.reasoning_tokens is None or result.completion_tokens_details.reasoning_tokens == 0
-
-def test_gemini_token_usage_standard_response():
-    """Verify that standard responses with details are computed correctly and not overwritten."""
-    v = VertexGeminiConfig()
-    
-    usage_metadata_dict = {
-        "promptTokenCount": 100,
-        "candidatesTokenCount": 50,
-        "totalTokenCount": 150,
-        "candidatesTokensDetails": [
-            {"modality": "TEXT", "tokenCount": 40},
-            {"modality": "IMAGE", "tokenCount": 10}
-        ]
-    }
-    
-    completion_response = {"usageMetadata": usage_metadata_dict}
-    result = v._calculate_usage(completion_response=completion_response)
-    
-    assert result.completion_tokens == 50
-    assert result.completion_tokens_details.text_tokens == 40
-    assert result.completion_tokens_details.image_tokens == 10
diff --git a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
index d09de3a0f2..5d91275fa8 100644
--- a/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
+++ b/tests/test_litellm/llms/vertex_ai/gemini/test_vertex_and_google_ai_studio_gemini.py
@@ -2505,3 +2505,72 @@ def test_vertex_ai_multiple_function_declarations_grouped():
     func_names = [f["name"] for f in tools[0]["function_declarations"]]
     assert "func1" in func_names
     assert "func2" in func_names
+
+
+def test_gemini_3_flash_preview_token_usage_fallback():
+    """Test fallback logic when candidatesTokensDetails is missing (e.g. Gemini 3 Flash Preview)."""
+    v = VertexGeminiConfig()
+
+    usage_metadata_dict = {
+        "promptTokenCount": 2145,
+        "candidatesTokenCount": 509,
+        "totalTokenCount": 2654,
+        # candidatesTokensDetails intentionally omitted
+    }
+
+    completion_response = {"usageMetadata": usage_metadata_dict}
+    result = v._calculate_usage(completion_response=completion_response)
+
+    assert result.completion_tokens == 509
+    assert result.prompt_tokens == 2145
+    assert result.total_tokens == 2654
+
+    # Text tokens should be derived from candidatesTokenCount
+    assert result.completion_tokens_details is not None
+    assert result.completion_tokens_details.text_tokens == 509
+    assert result.completion_tokens_details.image_tokens is None
+    assert result.completion_tokens_details.audio_tokens is None
+
+
+def test_gemini_no_reasoning_fallback():
+    """Test fallback when reasoning_effort is absent and details are missing."""
+    v = VertexGeminiConfig()
+
+    usage_metadata_dict = {
+        "promptTokenCount": 100,
+        "candidatesTokenCount": 264,
+        "totalTokenCount": 364,
+    }
+
+    completion_response = {"usageMetadata": usage_metadata_dict}
+    result = v._calculate_usage(completion_response=completion_response)
+
+    assert result.completion_tokens == 264
+    assert result.completion_tokens_details is not None
+    assert result.completion_tokens_details.text_tokens == 264
+    assert (
+        result.completion_tokens_details.reasoning_tokens is None
+        or result.completion_tokens_details.reasoning_tokens == 0
+    )
+
+
+def test_gemini_token_usage_standard_response():
+    """Verify that standard responses with details are computed correctly and not overwritten."""
+    v = VertexGeminiConfig()
+
+    usage_metadata_dict = {
+        "promptTokenCount": 100,
+        "candidatesTokenCount": 50,
+        "totalTokenCount": 150,
+        "candidatesTokensDetails": [
+            {"modality": "TEXT", "tokenCount": 40},
+            {"modality": "IMAGE", "tokenCount": 10},
+        ],
+    }
+
+    completion_response = {"usageMetadata": usage_metadata_dict}
+    result = v._calculate_usage(completion_response=completion_response)
+
+    assert result.completion_tokens == 50
+    assert result.completion_tokens_details.text_tokens == 40
+    assert result.completion_tokens_details.image_tokens == 10