fix(vertex_ai): send enum serviceTier values to Vertex Gemini

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>

litellm/llms/vertex_ai/gemini/transformation.py

@@ -88,6 +88,17 @@ def _apply_gemini_mime_type_aliases(mime_type: str) -> str:
     return _GEMINI_MIME_TYPE_ALIASES.get(normalized, normalized)
 
 
+def _map_vertex_service_tier(service_tier: str) -> str:
+    normalized_service_tier: Final = service_tier.lower()
+    if normalized_service_tier == "flex":
+        return "SERVICE_TIER_FLEX"
+    if normalized_service_tier == "priority":
+        return "SERVICE_TIER_PRIORITY"
+    if normalized_service_tier == "default":
+        return "SERVICE_TIER_STANDARD"
+    return service_tier
+
+
 def _get_vertex_base() -> "VertexBase":
     """Lazily return the shared VertexBase instance to avoid a module-load-time cyclic import."""
     global _GCS_METADATA_VERTEX_BASE
@@ -1227,7 +1238,12 @@ def transform_request_body(
 
         if service_tier := optional_params.pop("service_tier", None):
             if isinstance(service_tier, str):
-                if service_tier.lower() == "default":
+                if custom_llm_provider in (
+                    LlmProviders.VERTEX_AI.value,
+                    LlmProviders.VERTEX_AI_BETA.value,
+                ):
+                    data["serviceTier"] = _map_vertex_service_tier(service_tier)
+                elif service_tier.lower() == "default":
                     data["serviceTier"] = "standard"
                 else:
                     data["serviceTier"] = service_tier.lower()

tests/unit/llms/vertex_ai/gemini/test_vertex_ai_gemini_transformation.py

@@ -208,13 +208,24 @@ def test_vertex_ai_includes_labels():
     assert result["labels"] == {"project": "test", "team": "ai"}
 
 
-def test_service_tier_forwarded_to_vertex_ai():
-    """Test that service_tier in optional_params is mapped to serviceTier in request body."""
-    messages = [{"role": "user", "content": "test"}]
-    optional_params = {"service_tier": "flex"}
-    litellm_params = {}
+@pytest.mark.parametrize(
+    ("service_tier", "expected_service_tier"),
+    [
+        ("flex", "SERVICE_TIER_FLEX"),
+        ("FLEX", "SERVICE_TIER_FLEX"),
+        ("priority", "SERVICE_TIER_PRIORITY"),
+        ("default", "SERVICE_TIER_STANDARD"),
+        ("SERVICE_TIER_FLEX", "SERVICE_TIER_FLEX"),
+    ],
+)
+def test_service_tier_forwarded_to_vertex_ai(
+    service_tier: str, expected_service_tier: str
+) -> None:
+    messages: Final = [{"role": "user", "content": "test"}]
+    optional_params: Final = {"service_tier": service_tier}
+    litellm_params: Final = {}
 
-    result = transform_request_body(
+    result: Final = transform_request_body(
         messages=messages,
         model="gemini-2.5-pro",
         optional_params=optional_params,
@@ -223,7 +234,19 @@ def test_service_tier_forwarded_to_vertex_ai():
         cached_content=None,
     )
 
-    assert "serviceTier" in result
+    assert result["serviceTier"] == expected_service_tier
+
+
+def test_service_tier_forwarded_to_gemini() -> None:
+    result: Final = transform_request_body(
+        messages=[{"role": "user", "content": "test"}],
+        model="gemini-2.5-pro",
+        optional_params={"service_tier": "flex"},
+        custom_llm_provider="gemini",
+        litellm_params={},
+        cached_content=None,
+    )
+
     assert result["serviceTier"] == "flex"