diff --git a/tests/test_litellm/proxy/auth/test_auth_checks.py b/tests/test_litellm/proxy/auth/test_auth_checks.py
index 6b8260e7f08..f480e096081 100644
--- a/tests/test_litellm/proxy/auth/test_auth_checks.py
+++ b/tests/test_litellm/proxy/auth/test_auth_checks.py
@@ -51,6 +51,8 @@ from litellm.proxy.auth.auth_checks import (
     get_key_object,
     get_user_object,
     invalidate_team_member_spend_state,
+    request_skips_budget_checks,
+    route_skips_budget_checks,
     vector_store_access_check,
 )
 from litellm.caching.in_memory_cache import InMemoryCache
@@ -8447,3 +8449,15 @@ async def test_access_group_model_fallback_uses_the_injected_database(channel: s
                 llm_router=None, prisma_client=client,
             ) is True
     reader.assert_awaited_once_with(where={"access_group_id": "group-a"})
+
+
+def test_route_skips_budget_checks_marks_only_spend_free_routes() -> None:
+    assert route_skips_budget_checks(route="/v1/models") is True
+    assert route_skips_budget_checks(route="/spend/logs") is True
+    assert route_skips_budget_checks(route="/health") is False
+    assert route_skips_budget_checks(route="/v1/chat/completions") is False
+
+
+def test_request_skips_budget_checks_extends_route_rule_with_zero_cost_models() -> None:
+    assert request_skips_budget_checks(route="/v1/models", model=None, llm_router=None) is True
+    assert request_skips_budget_checks(route="/v1/chat/completions", model=None, llm_router=None) is False
diff --git a/tests/test_litellm/proxy/test_common_request_processing.py b/tests/test_litellm/proxy/test_common_request_processing.py
index 5f1e4736373..4ac687625c2 100644
--- a/tests/test_litellm/proxy/test_common_request_processing.py
+++ b/tests/test_litellm/proxy/test_common_request_processing.py
@@ -673,6 +673,26 @@ class TestProxyBaseLLMRequestProcessing:
         assert exc_info.value.code == "429"
         assert "guardrail-tag" in exc_info.value.message
 
+    @pytest.mark.asyncio
+    async def test_enforce_guardrail_added_tag_budgets_still_checks_when_model_is_unparseable(self, monkeypatch):
+        from litellm.proxy.common_request_processing import _enforce_guardrail_added_tag_budgets
+
+        tag_budget_check = AsyncMock()
+        monkeypatch.setattr(litellm.proxy.common_request_processing, "tag_max_budget_check_for_tags", tag_budget_check)
+        monkeypatch.setattr("litellm.proxy.proxy_server.prisma_client", MagicMock())
+        monkeypatch.setattr("litellm.proxy.proxy_server.user_api_key_cache", MagicMock())
+
+        await _enforce_guardrail_added_tag_budgets(
+            data={"model": 5, "metadata": {"tags": ["guardrail-tag"]}},
+            tags_before_guardrails=frozenset(),
+            route="/v1/chat/completions",
+            llm_router=None,
+            user_api_key_dict=ProxyUserAPIKeyAuth(api_key="sk-test"),
+            proxy_logging_obj=MagicMock(spec=ProxyLogging),
+        )
+
+        tag_budget_check.assert_awaited_once()
+
     @pytest.mark.asyncio
     async def test_common_processing_pre_call_logic_arms_auto_router_compression_before_guardrails(
         self, monkeypatch