diff --git a/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py b/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py
index 98161402c45..cac8f98cbb7 100644
--- a/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py
+++ b/tests/test_litellm/proxy/pass_through_endpoints/test_llm_pass_through_endpoints.py
@@ -1479,10 +1479,11 @@ class TestForwardHeaders:
 
         # Create a mock request with custom headers
         mock_request = MagicMock(spec=Request)
+        mock_request.state = None  # Prevent MagicMock from returning a truthy _cached_headers
         mock_request.method = "POST"
         mock_request.url = MagicMock()
         mock_request.url.path = "/test/endpoint"
-        
+
         # User headers that should be forwarded
         user_headers = {
             "x-custom-header": "custom-value",
diff --git a/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py b/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py
index e439dfd693c..2aecc2ec2e5 100644
--- a/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py
+++ b/tests/test_litellm/proxy/spend_tracking/test_spend_management_endpoints.py
@@ -1650,58 +1650,64 @@ async def test_global_spend_keys_endpoint_limit_validation(client, monkeypatch):
     # Create a simple mock for prisma client with empty response
     mock_prisma_client = MagicMock()
     mock_db = MagicMock()
-    mock_query_raw = MagicMock()
-    mock_query_raw.return_value = asyncio.Future()
-    mock_query_raw.return_value.set_result([])
+    mock_query_raw = AsyncMock(return_value=[])
     mock_db.query_raw = mock_query_raw
     mock_prisma_client.db = mock_db
     # Apply the mock to the prisma_client module
     monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", mock_prisma_client)
 
-    # Call the endpoint without specifying a limit
-    no_limit_response = client.get("/global/spend/keys")
-    assert no_limit_response.status_code == 200
-    mock_query_raw.assert_called_once_with('SELECT * FROM "Last30dKeysBySpend";')
-    # Reset the mock for the next test
-    mock_query_raw.reset_mock()
-    # Test with valid input
-    normal_limit = "10"
-    good_input_response = client.get(f"/global/spend/keys?limit={normal_limit}")
-    assert good_input_response.status_code == 200
-    # Verify the mock was called with the correct parameters
-    mock_query_raw.assert_called_once_with(
-        'SELECT * FROM "Last30dKeysBySpend" LIMIT $1 ;', 10
+    # Override auth to bypass API key validation
+    app.dependency_overrides[ps.user_api_key_auth] = lambda: UserAPIKeyAuth(
+        user_role=LitellmUserRoles.PROXY_ADMIN, user_id="admin_user"
     )
-    # Reset the mock for the next test
-    mock_query_raw.reset_mock()
-    # Test with SQL injection payload
-    sql_injection_limit = "10; DROP TABLE spend_logs; --"
-    response = client.get(f"/global/spend/keys?limit={sql_injection_limit}")
-    # Verify the response is a validation error (422)
-    assert response.status_code == 422
-    # Verify the mock was not called with the SQL injection payload
-    # This confirms that the validation happens before the database query
-    mock_query_raw.assert_not_called()
-    # Reset the mock for the next test
-    mock_query_raw.reset_mock()
-    # Test with non-numeric input
-    non_numeric_limit = "abc"
-    response = client.get(f"/global/spend/keys?limit={non_numeric_limit}")
-    assert response.status_code == 422
-    mock_query_raw.assert_not_called()
-    mock_query_raw.reset_mock()
-    # Test with negative number
-    negative_limit = "-5"
-    response = client.get(f"/global/spend/keys?limit={negative_limit}")
-    assert response.status_code == 422
-    mock_query_raw.assert_not_called()
-    mock_query_raw.reset_mock()
-    # Test with zero
-    zero_limit = "0"
-    response = client.get(f"/global/spend/keys?limit={zero_limit}")
-    assert response.status_code == 422
-    mock_query_raw.assert_not_called()
-    mock_query_raw.reset_mock()
+
+    try:
+        # Call the endpoint without specifying a limit
+        no_limit_response = client.get("/global/spend/keys")
+        assert no_limit_response.status_code == 200
+        mock_query_raw.assert_called_once_with('SELECT * FROM "Last30dKeysBySpend";')
+        # Reset the mock for the next test
+        mock_query_raw.reset_mock()
+        # Test with valid input
+        normal_limit = "10"
+        good_input_response = client.get(f"/global/spend/keys?limit={normal_limit}")
+        assert good_input_response.status_code == 200
+        # Verify the mock was called with the correct parameters
+        mock_query_raw.assert_called_once_with(
+            'SELECT * FROM "Last30dKeysBySpend" LIMIT $1 ;', 10
+        )
+        # Reset the mock for the next test
+        mock_query_raw.reset_mock()
+        # Test with SQL injection payload
+        sql_injection_limit = "10; DROP TABLE spend_logs; --"
+        response = client.get(f"/global/spend/keys?limit={sql_injection_limit}")
+        # Verify the response is a validation error (422)
+        assert response.status_code == 422
+        # Verify the mock was not called with the SQL injection payload
+        # This confirms that the validation happens before the database query
+        mock_query_raw.assert_not_called()
+        # Reset the mock for the next test
+        mock_query_raw.reset_mock()
+        # Test with non-numeric input
+        non_numeric_limit = "abc"
+        response = client.get(f"/global/spend/keys?limit={non_numeric_limit}")
+        assert response.status_code == 422
+        mock_query_raw.assert_not_called()
+        mock_query_raw.reset_mock()
+        # Test with negative number
+        negative_limit = "-5"
+        response = client.get(f"/global/spend/keys?limit={negative_limit}")
+        assert response.status_code == 422
+        mock_query_raw.assert_not_called()
+        mock_query_raw.reset_mock()
+        # Test with zero
+        zero_limit = "0"
+        response = client.get(f"/global/spend/keys?limit={zero_limit}")
+        assert response.status_code == 422
+        mock_query_raw.assert_not_called()
+        mock_query_raw.reset_mock()
+    finally:
+        app.dependency_overrides.pop(ps.user_api_key_auth, None)
 
 
 @pytest.mark.asyncio
diff --git a/tests/test_litellm/proxy/test_health_check_functions.py b/tests/test_litellm/proxy/test_health_check_functions.py
index 4c91d0ae91e..354698b02fe 100644
--- a/tests/test_litellm/proxy/test_health_check_functions.py
+++ b/tests/test_litellm/proxy/test_health_check_functions.py
@@ -480,7 +480,7 @@ async def test_perform_health_check_and_save_passes_model_id_to_perform_health_c
     healthy = [{"model": "gpt-4"}]
     unhealthy = []
 
-    async def mock_perform_health_check(model_list, model=None, cli_model=None, details=True, model_id=None):
+    async def mock_perform_health_check(model_list, model=None, cli_model=None, details=True, model_id=None, max_concurrency=None):
         return healthy, unhealthy
 
     with patch(
diff --git a/tests/test_litellm/proxy/test_model_dump_with_preserved_fields.py b/tests/test_litellm/proxy/test_model_dump_with_preserved_fields.py
index 3001c87ebed..316c1e879cc 100644
--- a/tests/test_litellm/proxy/test_model_dump_with_preserved_fields.py
+++ b/tests/test_litellm/proxy/test_model_dump_with_preserved_fields.py
@@ -242,7 +242,7 @@ def test_full_output_structure_non_streaming():
     )
     result = model_dump_with_preserved_fields(response, exclude_unset=True)
 
-    # Top-level keys
+    # Top-level keys (usage is None when not explicitly set and excluded by exclude_unset=True)
     assert set(result.keys()) == {
         "id",
         "choices",
@@ -250,7 +250,6 @@ def test_full_output_structure_non_streaming():
         "model",
         "object",
         "system_fingerprint",
-        "usage",
     }
     assert result["object"] == "chat.completion"
     assert result["model"] == "gpt-4.1"
@@ -270,12 +269,6 @@ def test_full_output_structure_non_streaming():
     assert msg["content"] == "Hello!"
     assert msg["role"] == "assistant"
 
-    # Usage structure
-    usage = result["usage"]
-    assert "prompt_tokens" in usage
-    assert "completion_tokens" in usage
-    assert "total_tokens" in usage
-
 
 def test_full_output_structure_tool_calls():
     """
diff --git a/tests/test_litellm/proxy/test_proxy_cli.py b/tests/test_litellm/proxy/test_proxy_cli.py
index c839c22de5f..c6b2015984e 100644
--- a/tests/test_litellm/proxy/test_proxy_cli.py
+++ b/tests/test_litellm/proxy/test_proxy_cli.py
@@ -331,7 +331,8 @@ class TestProxyInitializationHelpers:
 
     @patch("uvicorn.run")
     @patch("builtins.print")
-    def test_max_requests_before_restart_flag(self, mock_print, mock_uvicorn_run):
+    @patch("litellm.proxy.db.prisma_client.PrismaManager.setup_database")
+    def test_max_requests_before_restart_flag(self, mock_setup_db, mock_print, mock_uvicorn_run):
         """Test that the max_requests_before_restart flag is passed to uvicorn as limit_max_requests"""
         from click.testing import CliRunner
 
@@ -344,7 +345,10 @@ class TestProxyInitializationHelpers:
         mock_key_mgmt = MagicMock()
         mock_save_worker_config = MagicMock()
 
+        clean_env = {k: v for k, v in os.environ.items() if k not in ("DATABASE_URL", "DIRECT_URL")}
         with patch.dict(
+            os.environ, clean_env, clear=True,
+        ), patch.dict(
             "sys.modules",
             {
                 "proxy_server": MagicMock(
@@ -367,7 +371,7 @@ class TestProxyInitializationHelpers:
                 run_server, ["--local", "--max_requests_before_restart", "123"]
             )
 
-            assert result.exit_code == 0
+            assert result.exit_code == 0, f"exit_code={result.exit_code}, output={result.output}"
             mock_uvicorn_run.assert_called_once()
 
             # Check that uvicorn.run was called with limit_max_requests parameter
diff --git a/tests/test_litellm/proxy/test_shared_health_check.py b/tests/test_litellm/proxy/test_shared_health_check.py
index 82deebc424a..0212d87baab 100644
--- a/tests/test_litellm/proxy/test_shared_health_check.py
+++ b/tests/test_litellm/proxy/test_shared_health_check.py
@@ -1,10 +1,13 @@
 import asyncio
 import json
-import pytest
 import time
 from unittest.mock import AsyncMock, MagicMock, patch
 
-from litellm.proxy.health_check_utils.shared_health_check_manager import SharedHealthCheckManager
+import pytest
+
+from litellm.proxy.health_check_utils.shared_health_check_manager import (
+    SharedHealthCheckManager,
+)
 
 
 class TestSharedHealthCheckManager:
@@ -272,7 +275,7 @@ class TestSharedHealthCheckManager:
             )
         
         # Should call perform_health_check and cache results
-        mock_perform.assert_called_once_with(model_list=model_list, details=True)
+        mock_perform.assert_called_once_with(model_list=model_list, details=True, max_concurrency=None)
         assert healthy == expected_healthy
         assert unhealthy == expected_unhealthy
         
@@ -329,7 +332,7 @@ class TestSharedHealthCheckManager:
         
         # Should fall back to local health check
         mock_sleep.assert_called_once_with(2)
-        mock_perform.assert_called_once_with(model_list=model_list, details=True)
+        mock_perform.assert_called_once_with(model_list=model_list, details=True, max_concurrency=None)
         assert healthy == expected_healthy
         assert unhealthy == expected_unhealthy