diff --git a/litellm/router.py b/litellm/router.py
index 7449492e89..676a9e6e20 100644
--- a/litellm/router.py
+++ b/litellm/router.py
@@ -5278,12 +5278,12 @@ class Router:
             original_id = decoded.get("response_id", container_id)
             if original_id != container_id:
                 kwargs["container_id"] = original_id
+            decoded_provider = decoded.get("custom_llm_provider")
+            if decoded_provider and kwargs.get("custom_llm_provider") == "openai":
+                kwargs["custom_llm_provider"] = decoded_provider
             model_id = decoded.get("model_id")
             if model_id:
                 kwargs["model"] = model_id
-                decoded_provider = decoded.get("custom_llm_provider")
-                if decoded_provider and kwargs.get("custom_llm_provider") == "openai":
-                    kwargs["custom_llm_provider"] = decoded_provider
                 return await self._ageneric_api_call_with_fallbacks(
                     original_function=original_function,
                     **kwargs,
diff --git a/tests/router_unit_tests/test_router_endpoints.py b/tests/router_unit_tests/test_router_endpoints.py
index 19d7155e4a..0ce2dec9b5 100644
--- a/tests/router_unit_tests/test_router_endpoints.py
+++ b/tests/router_unit_tests/test_router_endpoints.py
@@ -1203,3 +1203,36 @@ async def test_init_containers_api_endpoints_managed_id_without_model_id_unwraps
     assert call_kw["container_id"] == "cfile_upstream_abc"
     assert call_kw["file_id"] == "cfile_xyz"
     assert call_kw["custom_llm_provider"] == "openai"
+
+
+@pytest.mark.asyncio
+async def test_init_containers_api_endpoints_managed_id_without_model_id_applies_decoded_provider():
+    """
+    A managed ``cntr_`` ID can encode a non-OpenAI provider (e.g. ``azure``) with
+    an empty ``model_id`` (streaming events without router ``model_info.id``).
+    The router must still apply the decoded provider so the request routes to
+    the correct upstream — not stay on the default ``openai``.
+    """
+    from litellm.responses.utils import ResponsesAPIRequestUtils
+
+    router = Router(model_list=[])
+    mock_original_function = AsyncMock(return_value={"ok": True})
+
+    managed_id = ResponsesAPIRequestUtils._build_container_id(
+        custom_llm_provider="azure",
+        model_id=None,
+        container_id="cfile_upstream_abc",
+    )
+
+    await router._init_containers_api_endpoints(
+        original_function=mock_original_function,
+        custom_llm_provider="openai",
+        container_id=managed_id,
+        file_id="cfile_xyz",
+    )
+
+    mock_original_function.assert_called_once()
+    call_kw = mock_original_function.call_args.kwargs
+    assert call_kw["container_id"] == "cfile_upstream_abc"
+    assert call_kw["file_id"] == "cfile_xyz"
+    assert call_kw["custom_llm_provider"] == "azure"