diff --git a/onnxruntime/core/providers/openvino/backend_utils.cc b/onnxruntime/core/providers/openvino/backend_utils.cc
index 4bfdc7e7aa9a7..cf7e9b2446bcb 100644
--- a/onnxruntime/core/providers/openvino/backend_utils.cc
+++ b/onnxruntime/core/providers/openvino/backend_utils.cc
@@ -47,11 +47,6 @@ struct static_cast_int64 {
 
 std::shared_ptr<InferenceEngine::CNNNetwork>
 CreateCNNNetwork(const Provider_ModelProto& model_proto, const GlobalContext& global_context, const SubGraphContext& subgraph_context, std::map<std::string, std::shared_ptr<ngraph::Node>>& const_outputs_map) {
-#if (defined OPENVINO_2020_2) || (defined OPENVINO_2020_3)
-  ORT_UNUSED_PARAMETER(const_outputs_map);
-#endif
-
-  std::istringstream model_stream{model_proto.SerializeAsString()};
   std::shared_ptr<ngraph::Function> ng_function;
 
 #ifndef NDEBUG
@@ -60,6 +55,9 @@ CreateCNNNetwork(const Provider_ModelProto& model_proto, const GlobalContext& gl
   }
 #endif
 
+#if (defined OPENVINO_2020_2) || (defined OPENVINO_2020_3)
+  ORT_UNUSED_PARAMETER(const_outputs_map);
+  std::istringstream model_stream{model_proto.SerializeAsString()};
   try {
     ng_function = ngraph::onnx_import::import_onnx_model(model_stream);
     LOGS_DEFAULT(INFO) << "ONNX Import Done";
@@ -68,6 +66,20 @@ CreateCNNNetwork(const Provider_ModelProto& model_proto, const GlobalContext& gl
   } catch (...) {
     ORT_THROW(log_tag + "[OpenVINO-EP] Unknown exception while importing model to nGraph Func");
   }
+#else
+  InferenceEngine::CNNNetwork cnn_network;
+  const std::string model = model_proto.SerializeAsString();
+  InferenceEngine::Blob::Ptr blob = {nullptr};
+  try {
+    cnn_network = global_context.ie_core.ReadNetwork(model, blob);
+    LOGS_DEFAULT(INFO) << "Read network Done";
+  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    ORT_THROW(log_tag + "[OpenVINO-EP] Exception while Reading network: " + std::string(e.what()));
+  } catch (...) {
+    ORT_THROW(log_tag + "[OpenVINO-EP] Unknown exception while Reading network");
+  }
+  ng_function = cnn_network.getFunction();
+#endif
 
   if (global_context.device_type.find("GPU") != std::string::npos &&
       subgraph_context.precision == InferenceEngine::Precision::FP16) {
diff --git a/onnxruntime/core/providers/openvino/ov_versions/capability_2021_1.cc b/onnxruntime/core/providers/openvino/ov_versions/capability_2021_1.cc
index d6459c4c3a79e..a31b4b4f9a062 100644
--- a/onnxruntime/core/providers/openvino/ov_versions/capability_2021_1.cc
+++ b/onnxruntime/core/providers/openvino/ov_versions/capability_2021_1.cc
@@ -272,16 +272,17 @@ static bool IsUnsupportedOpMode(const Node* node, const GraphViewer& graph_viewe
         return true;
     }
   } else if (optype == "Max" || optype == "Min" || optype == "Mean" || optype == "Sum") {
-    if (GetInputCount(node, initializers) == 1)
+    if (GetInputCount(node, initializers) == 1) {
       return true;
-      if (optype == "Max" || optype == "Min") {
-        for (size_t i = 0; i < node->InputDefs().size(); i++) {
-          auto dtype = node->InputDefs()[i]->TypeAsProto()->tensor_type().elem_type();
-          if (dtype == ONNX_NAMESPACE::TensorProto_DataType::TensorProto_DataType_UINT8 ||
-              dtype == ONNX_NAMESPACE::TensorProto_DataType::TensorProto_DataType_INT16)
-            return true;
-       }
+    }
+    if (optype == "Max" || optype == "Min") {
+      for (size_t i = 0; i < node->InputDefs().size(); i++) {
+        auto dtype = node->InputDefs()[i]->TypeAsProto()->tensor_type().elem_type();
+        if (dtype == ONNX_NAMESPACE::TensorProto_DataType::TensorProto_DataType_UINT8 ||
+            dtype == ONNX_NAMESPACE::TensorProto_DataType::TensorProto_DataType_INT16)
+          return true;
       }
+    }
   } else if (optype == "Clip") {
     //Only float 16, float and double data types are supported
     const bool data_is_float = node->InputDefs()[0]->Type()->find("float") != std::string::npos;