fix(proxy): extract model from vertex ai passthrough url pattern (#18097)

extract model id from vertex ai passthrough routes that follow the pattern:
/vertex_ai/*/models/{model_id}:*

the model extraction now handles vertex ai routes by regex matching the model
segment from the url path, which allows proper model identification for
authentication and authorization in proxy pass-through endpoints.

adds comprehensive test coverage for vertex ai model extraction including:
- various vertex api versions (v1, v1beta1)
- different locations (us-central1, asia-southeast1)
- model names with special suffixes (gemini-1.5-pro, gemini-2.0-flash)
- precedence verification (request body model over url)
- non-vertex route isolation

litellm/proxy/auth/auth_utils.py

@@ -616,6 +616,14 @@ def get_model_from_request(
         if match:
             model = match.group(1)
 
+    # If still not found, extract from Vertex AI passthrough route
+    # Pattern: /vertex_ai/.../models/{model_id}:*
+    # Example: /vertex_ai/v1/.../models/gemini-1.5-pro:generateContent
+    if model is None and "/vertex" in route.lower():
+        vertex_match = re.search(r"/models/([^/:]+)", route)
+        if vertex_match:
+            model = vertex_match.group(1)
+
     return model
 
 

tests/local_testing/test_auth_utils.py

@@ -311,3 +311,56 @@ def test_get_internal_user_header_from_mapping_no_internal_returns_none():
     single_mapping = {"header_name": "X-Only-Customer", "litellm_user_role": "customer"}
     result = LiteLLMProxyRequestSetup.get_internal_user_header_from_mapping(single_mapping)
     assert result is None
+
+
+@pytest.mark.parametrize(
+    "request_data, route, expected_model",
+    [
+        # Vertex AI passthrough URL patterns
+        (
+            {},
+            "/vertex_ai/v1/projects/my-project/locations/us-central1/publishers/google/models/gemini-1.5-pro:generateContent",
+            "gemini-1.5-pro"
+        ),
+        (
+            {},
+            "/vertex_ai/v1beta1/projects/my-project/locations/us-central1/publishers/google/models/gemini-1.0-pro:streamGenerateContent",
+            "gemini-1.0-pro"
+        ),
+        (
+            {},
+            "/vertex_ai/v1/projects/my-project/locations/asia-southeast1/publishers/google/models/gemini-2.0-flash:generateContent",
+            "gemini-2.0-flash"
+        ),
+        # Model without method suffix (no colon) - should still extract
+        (
+            {},
+            "/vertex_ai/v1/projects/my-project/locations/us-central1/publishers/google/models/gemini-pro",
+            "gemini-pro"  # Should match even without colon
+        ),
+        # Request body model takes precedence over URL
+        (
+            {"model": "gpt-4o"},
+            "/vertex_ai/v1/projects/my-project/locations/us-central1/publishers/google/models/gemini-1.5-pro:generateContent",
+            "gpt-4o"
+        ),
+        # Non-vertex route should not extract from vertex pattern
+        (
+            {},
+            "/openai/v1/chat/completions",
+            None
+        ),
+        # Azure deployment pattern should still work
+        (
+            {},
+            "/openai/deployments/my-deployment/chat/completions",
+            "my-deployment"
+        ),
+    ],
+)
+def test_get_model_from_request_vertex_ai_passthrough(request_data, route, expected_model):
+    """Test that get_model_from_request correctly extracts Vertex AI model from URL"""
+    from litellm.proxy.auth.auth_utils import get_model_from_request
+
+    model = get_model_from_request(request_data, route)
+    assert model == expected_model