diff --git a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py
index cc8145bd162..3b27414e2a8 100644
--- a/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py
+++ b/litellm/proxy/pass_through_endpoints/llm_provider_handlers/vertex_passthrough_logging_handler.py
@@ -659,7 +659,14 @@ class VertexPassthroughLoggingHandler:
                 continue
             all_openai_chunks.append(parsed_chunk)
 
-        complete_streaming_response: Final = litellm.stream_chunk_builder(chunks=all_openai_chunks)
+        # With the logging object the builder leaves pricing to the logger,
+        # which honours the deployment's custom pricing; without it the
+        # builder stamps a cost priced by model name alone, which the logger
+        # then reuses as already calculated.
+        complete_streaming_response: Final = litellm.stream_chunk_builder(
+            chunks=all_openai_chunks,
+            logging_obj=litellm_logging_obj,
+        )
 
         return complete_streaming_response
 
diff --git a/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_vertex_passthrough_logging_handler.py b/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_vertex_passthrough_logging_handler.py
index dd271aab380..1f98549071d 100644
--- a/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_vertex_passthrough_logging_handler.py
+++ b/tests/unit/proxy/pass_through_endpoints/llm_provider_handlers/test_vertex_passthrough_logging_handler.py
@@ -174,3 +174,51 @@ def test_generate_content_cost_uses_the_deployments_custom_pricing(monkeypatch:
     )
 
     assert kwargs["response_cost"] == pytest.approx(1.5 + 7.5)
+
+
+def test_assembled_stream_is_priced_at_the_deployments_custom_pricing(monkeypatch: pytest.MonkeyPatch):
+    """The logger prices a streamed generateContent call from the response the
+    handler assembles; that response must not arrive carrying a cost priced
+    by model name alone, which the logger would reuse as already calculated."""
+    import litellm
+
+    deployment_id: Final = "vertex-gemini-with-custom-price"
+    custom_price: Final = {"input_cost_per_token": 1.5e-6, "output_cost_per_token": 7.5e-6}
+    monkeypatch.setitem(
+        litellm.model_cost,
+        deployment_id,
+        {**custom_price, "litellm_provider": "vertex_ai-language-models", "mode": "chat"},
+    )
+    logging_obj: Final = Logging(
+        model="gemini-2.5-flash",
+        messages=[{"role": "user", "content": "hi"}],
+        stream=True,
+        call_type="pass_through_endpoint",
+        start_time=datetime(2026, 1, 1),
+        litellm_call_id="call-1",
+        function_id="fn-1",
+    )
+    logging_obj.update_environment_variables(
+        litellm_params={"litellm_metadata": {"model_info": {"id": deployment_id, **custom_price}}},
+        optional_params={},
+        model="gemini-2.5-flash",
+        custom_llm_provider="vertex_ai",
+    )
+    last_chunk: Final = {
+        "candidates": [{"content": {"parts": [{"text": "Hi."}], "role": "model"}, "finishReason": "STOP", "index": 0}],
+        "usageMetadata": {
+            "promptTokenCount": 1_000_000,
+            "candidatesTokenCount": 1_000_000,
+            "totalTokenCount": 2_000_000,
+        },
+        "modelVersion": "gemini-2.5-flash",
+    }
+
+    response: Final = VertexPassthroughLoggingHandler._build_complete_streaming_response(
+        all_chunks=[f"data: {json.dumps(last_chunk)}"],
+        litellm_logging_obj=logging_obj,
+        model="gemini-2.5-flash",
+        url_route="/v1/projects/p/locations/global/publishers/google/models/gemini-2.5-flash:streamGenerateContent",
+    )
+
+    assert logging_obj.response_cost_calculator(result=response) == pytest.approx(1.5 + 7.5)