diff --git a/cmake/CMakeLists.txt b/cmake/CMakeLists.txt
index cc5b5cee4b..e9dc313861 100644
--- a/cmake/CMakeLists.txt
+++ b/cmake/CMakeLists.txt
@@ -1280,6 +1280,9 @@ if(onnxruntime_USE_OPENVINO)
   elseif (${VER} MATCHES "2021.3" OR $ENV{INTEL_OPENVINO_DIR} MATCHES "2021.3")
     set(OPENVINO_VERSION "2021.3")
     add_definitions(-DOPENVINO_2021_3=1)
+  elseif (${VER} MATCHES "2021.4" OR $ENV{INTEL_OPENVINO_DIR} MATCHES "2021.4")
+    set(OPENVINO_VERSION "2021.4")
+    add_definitions(-DOPENVINO_2021_4=1)
   else()
     message(FATAL_ERROR "Unsupported OpenVINO version: ${INTEL_OPENVINO_DIR}")
   endif()
diff --git a/onnxruntime/core/providers/openvino/backend_utils.cc b/onnxruntime/core/providers/openvino/backend_utils.cc
index ff5697a124..b245e95c8e 100644
--- a/onnxruntime/core/providers/openvino/backend_utils.cc
+++ b/onnxruntime/core/providers/openvino/backend_utils.cc
@@ -9,6 +9,12 @@
 
 #include <inference_engine.hpp>
 
+#ifdef OPENVINO_2021_4
+using Exception = InferenceEngine::Exception;
+#else
+using Exception = InferenceEngine::details::InferenceEngineException;
+#endif
+
 #include <ngraph/frontend/onnx_import/onnx.hpp>
 #include <ngraph/pass/convert_fp32_to_fp16.hpp>
 #include <ngraph/pass/constant_folding.hpp>
@@ -122,7 +128,8 @@ CreateCNNNetwork(const ONNX_NAMESPACE::ModelProto& model_proto, const GlobalCont
     ng_function->validate_nodes_and_infer_types();
   }
 
-#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || (defined OPENVINO_2021_3)
+#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || \
+    (defined OPENVINO_2021_3) || (defined OPENVINO_2021_4)
   if (!global_context.is_wholly_supported_graph) {
     std::map<std::string, std::string> result_to_output;
     for (auto& result : ng_function->get_results()) {
@@ -144,7 +151,7 @@ CreateCNNNetwork(const ONNX_NAMESPACE::ModelProto& model_proto, const GlobalCont
 
   try {
     return std::make_shared<InferenceEngine::CNNNetwork>(ng_function);
-  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+  } catch (const Exception& e) {
     ORT_THROW(log_tag + " Exception thrown while making IE::CNNNetwork: " + e.what());
   } catch (...) {
     ORT_THROW(log_tag + " Exception thrown while making IE::CNNNetwork");
@@ -203,7 +210,8 @@ void SetIODefs(const ONNX_NAMESPACE::ModelProto& model_proto,
   auto outputInfo = network->getOutputsInfo();
   for (auto iter = outputInfo.begin(); iter != outputInfo.end(); ++iter) {
     auto output_name = iter->first;
-#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || (defined OPENVINO_2021_3)
+#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || \
+    (defined OPENVINO_2021_3) || (defined OPENVINO_2021_4)
     auto it = const_outputs_map.find(output_name);
     //Output is constant and don't need to set precision
     if (it != const_outputs_map.end())
@@ -247,7 +255,8 @@ GetOutputTensor(Ort::CustomOpApi& ort, OrtKernelContext* context, size_t batch_s
   return output_tensor;
 }
 
-#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || (defined OPENVINO_2021_3)
+#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || \
+    (defined OPENVINO_2021_3) || (defined OPENVINO_2021_4)
 OrtValue*
 GetOutputTensor(Ort::CustomOpApi& ort, OrtKernelContext* context,
                 std::string output_name,
@@ -295,7 +304,8 @@ int GetFirstAvailableDevice(GlobalContext& global_context) {
   return i;
 }
 
-#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || (defined OPENVINO_2021_3)
+#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || \
+    (defined OPENVINO_2021_3) || (defined OPENVINO_2021_4)
 void FillOutputsWithConstantData(Ort::CustomOpApi& ort, std::shared_ptr<ngraph::Node> node, OrtValue* out_tensor) {
   switch (node->get_element_type()) {
     case ngraph::element::Type_t::f32: {
@@ -320,7 +330,8 @@ void FillOutputsWithConstantData(Ort::CustomOpApi& ort, std::shared_ptr<ngraph::
 }
 #endif
 
-#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || (defined OPENVINO_2021_3)
+#if (defined OPENVINO_2020_4) || (defined OPENVINO_2021_1) || (defined OPENVINO_2021_2) || \
+    (defined OPENVINO_2021_3) || (defined OPENVINO_2021_4)
 template <typename T>
 void FillOutputHelper(Ort::CustomOpApi& ort, OrtValue* out_tensor, std::shared_ptr<ngraph::Node> node) {
   auto const_node = std::dynamic_pointer_cast<ngraph::op::Constant>(node);
diff --git a/onnxruntime/core/providers/openvino/backend_utils.h b/onnxruntime/core/providers/openvino/backend_utils.h
index 1c3cd10059..585c4b8edc 100644
--- a/onnxruntime/core/providers/openvino/backend_utils.h
+++ b/onnxruntime/core/providers/openvino/backend_utils.h
@@ -48,7 +48,8 @@ CreateCNNNetwork(const ONNX_NAMESPACE::ModelProto& model_proto, const GlobalCont
 
 int GetFirstAvailableDevice(GlobalContext& global_context);
 
-#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
 void FillOutputsWithConstantData(Ort::CustomOpApi& ort, std::shared_ptr<ngraph::Node> node, OrtValue* out_tensor);
 
 template <typename T>
diff --git a/onnxruntime/core/providers/openvino/backends/basic_backend.cc b/onnxruntime/core/providers/openvino/backends/basic_backend.cc
index 5995f57e5f..a3981ad08a 100644
--- a/onnxruntime/core/providers/openvino/backends/basic_backend.cc
+++ b/onnxruntime/core/providers/openvino/backends/basic_backend.cc
@@ -9,6 +9,14 @@
 
 #include <inference_engine.hpp>
 
+#ifdef OPENVINO_2021_4
+using Exception = InferenceEngine::Exception;
+using WaitMode = InferenceEngine::InferRequest::WaitMode;
+#else
+using Exception = InferenceEngine::details::InferenceEngineException;
+using WaitMode = InferenceEngine::IInferRequest::WaitMode;
+#endif
+
 #include "core/providers/shared_library/provider_api.h"
 
 #include "../backend_utils.h"
@@ -70,7 +78,8 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
   if(!openvino_ep::BackendManager::GetGlobalContext().is_wholly_supported_graph) {
     ie_cnn_network_ = CreateCNNNetwork(model_proto, global_context_, subgraph_context_, const_outputs_map_);
     SetIODefs(model_proto, ie_cnn_network_, subgraph_context_.output_names, const_outputs_map_, global_context_.device_type);
-  #if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+  #if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+      defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
     if (const_outputs_map_.size() == subgraph_context_.output_names.size())
       subgraph_context_.is_constant = true;
   #endif
@@ -95,7 +104,7 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
           throw std::runtime_error("The compiled blob path is not set");
         exe_network_ = global_context_.ie_core.ImportNetwork(compiled_blob_path, hw_target, {});
       }
-    } catch (InferenceEngine::details::InferenceEngineException &e) {
+    } catch (Exception &e) {
       ORT_THROW(log_tag + " Exception while Importing Network for graph: " + subgraph_context_.subgraph_name + ": " + e.what());
     } catch(...) {
       ORT_THROW(log_tag + " Exception while Importing Network for graph: " + subgraph_context_.subgraph_name);
@@ -108,7 +117,8 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
     if(!openvino_ep::backend_utils::UseCompiledNetwork()) {
       ie_cnn_network_ = CreateCNNNetwork(model_proto, global_context_, subgraph_context_, const_outputs_map_);
       SetIODefs(model_proto, ie_cnn_network_, subgraph_context_.output_names, const_outputs_map_, global_context_.device_type);
-    #if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+    #if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+        defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
       if (const_outputs_map_.size() == subgraph_context_.output_names.size())
         subgraph_context_.is_constant = true;
     #endif
@@ -123,7 +133,8 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
       }
     #endif
       if (global_context_.device_type.find("MYRIAD") != std::string::npos) {
-    #if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+    #if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+        defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
         if (subgraph_context_.set_vpu_config) {
           config["MYRIAD_DETECT_NETWORK_BATCH"] = CONFIG_VALUE(NO);
         }
@@ -146,7 +157,7 @@ BasicBackend::BasicBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
       }
       try {
         exe_network_ = global_context_.ie_core.LoadNetwork(*ie_cnn_network_, hw_target, config);
-      } catch (const InferenceEngine::details::InferenceEngineException& e) {
+      } catch (const Exception& e) {
         ORT_THROW(log_tag + " Exception while Loading Network for graph: " + subgraph_context_.subgraph_name + ": " + e.what());
       } catch (...) {
         ORT_THROW(log_tag + " Exception while Loading Network for graph " + subgraph_context_.subgraph_name);
@@ -185,7 +196,7 @@ void BasicBackend::StartAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext*
     try {
       graph_input_blob = infer_request->GetBlob(input_name);
 
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + " Cannot access IE Blob for input: " + input_name + e.what());
     } catch (...) {
       ORT_THROW(log_tag + " Cannot access IE Blob for input: " + input_name);
@@ -197,7 +208,7 @@ void BasicBackend::StartAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext*
   // Start Async inference
   try {
     infer_request->StartAsync();
-  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+  } catch (const Exception& e) {
     ORT_THROW(log_tag + " Couldn't start Inference: " + e.what());
   } catch (...) {
     ORT_THROW(log_tag + " Couldn't start Inference");
@@ -209,8 +220,8 @@ void BasicBackend::StartAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext*
 void BasicBackend::CompleteAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext* context, std::shared_ptr<InferenceEngine::InferRequest> infer_request) {
   // Wait for Async inference completion
   try {
-    infer_request->Wait(InferenceEngine::IInferRequest::WaitMode::RESULT_READY);
-  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    infer_request->Wait(WaitMode::RESULT_READY);
+  } catch (const Exception& e) {
     ORT_THROW(log_tag + " Exception with completing Inference" + e.what());
   } catch (...) {
     ORT_THROW(log_tag + " Exception with completing Inference");
@@ -224,7 +235,7 @@ void BasicBackend::CompleteAsyncInference(Ort::CustomOpApi& ort, OrtKernelContex
     auto output_name = output_info_iter->first;
     try {
       graph_output_blob = infer_request->GetBlob(output_name);
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + " Cannot access IE Blob for output: " + output_name + e.what());
     } catch (...) {
       ORT_THROW(log_tag + " Cannot access IE Blob for output: " + output_name);
@@ -236,7 +247,8 @@ void BasicBackend::CompleteAsyncInference(Ort::CustomOpApi& ort, OrtKernelContex
     size_t batch_slice = 0;
     FillOutputBlob(graph_output_blob, output_tensor, ort, precision, batch_slice);
   }
-#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
   if (!const_outputs_map_.empty()) {
     for (auto item : const_outputs_map_) {
       auto out_name = item.first;
@@ -256,7 +268,8 @@ void BasicBackend::Infer(Ort::CustomOpApi& ort, OrtKernelContext* context) {
   LOGS_DEFAULT(INFO) << log_tag << "In Infer";
 
   if (subgraph_context_.is_constant) {
-#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1)  || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1)  || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
     for (auto item : const_outputs_map_) {
       auto out_name = item.first;
       auto node = item.second;
diff --git a/onnxruntime/core/providers/openvino/backends/basic_backend.h b/onnxruntime/core/providers/openvino/backends/basic_backend.h
index 1083a2ce6b..7ca037710a 100644
--- a/onnxruntime/core/providers/openvino/backends/basic_backend.h
+++ b/onnxruntime/core/providers/openvino/backends/basic_backend.h
@@ -12,6 +12,7 @@
 #include "core/providers/openvino/ibackend.h"
 
 #include <vector>
+#include <iostream>
 #include <string>
 #include <condition_variable>
 #include <mutex>
@@ -47,7 +48,7 @@ class InferRequestsQueue {
   InferRequestsQueue(InferenceEngine::ExecutableNetwork& net, size_t nireq) {
     InferenceEngine::InferRequest::Ptr infer_request;
     for (size_t id = 0; id < nireq; id++) {
-      infer_request = net.CreateInferRequestPtr();
+      infer_request = std::make_shared<InferenceEngine::InferRequest>(net.CreateInferRequest());
       infer_requests_.push_back(infer_request);
     }
   }
diff --git a/onnxruntime/core/providers/openvino/backends/vadm_backend.cc b/onnxruntime/core/providers/openvino/backends/vadm_backend.cc
index 22dbe2384c..eb970fd749 100644
--- a/onnxruntime/core/providers/openvino/backends/vadm_backend.cc
+++ b/onnxruntime/core/providers/openvino/backends/vadm_backend.cc
@@ -9,12 +9,21 @@
 
 #include <inference_engine.hpp>
 
+#ifdef OPENVINO_2021_4
+using Exception = InferenceEngine::Exception;
+using WaitMode = InferenceEngine::InferRequest::WaitMode;
+#else
+using Exception = InferenceEngine::details::InferenceEngineException;
+using WaitMode = InferenceEngine::IInferRequest::WaitMode;
+#endif
+
 #include "core/providers/shared_library/provider_api.h"
 
 #include "../contexts.h"
 #include "../backend_utils.h"
 #include "vadm_backend.h"
-#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2)  || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
 #include <vpu/hddl_config.hpp>
 #else
 #include <vpu/hddl_plugin_config.hpp>
@@ -68,14 +77,15 @@ VADMBackend::VADMBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
   if (global_context_.is_wholly_supported_graph && subgraph_context_.enable_batching) {
     for (int j = 0; j < 8; j++) {
       InferenceEngine::ExecutableNetwork exe_network;
-#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
       config[InferenceEngine::HDDL_DEVICE_TAG] = global_context_.deviceTags[j];
 #else
       config[VPU_HDDL_CONFIG_KEY(DEVICE_TAG)] = global_context_.deviceTags[j];
 #endif
       try {
         exe_network = global_context_.ie_core.LoadNetwork(*ie_cnn_network_, hw_target, config);
-      } catch (const InferenceEngine::details::InferenceEngineException& e) {
+      } catch (const Exception& e) {
         ORT_THROW(log_tag + " Exception while Loading Network for graph: " + subgraph_context_.subgraph_name + e.what());
       } catch (...) {
         ORT_THROW(log_tag + " Exception while Loading Network for graph " + subgraph_context_.subgraph_name);
@@ -86,8 +96,8 @@ VADMBackend::VADMBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
     for (size_t j = 0; j < num_inf_reqs_; j++) {
       InferenceEngine::InferRequest::Ptr infRequest;
       try {
-        infRequest = exe_networks[j].CreateInferRequestPtr();
-      } catch (const InferenceEngine::details::InferenceEngineException& e) {
+        infRequest = std::make_shared<InferenceEngine::InferRequest>(exe_networks[j].CreateInferRequest());
+      } catch (const Exception& e) {
         ORT_THROW(log_tag + "Exception while creating InferRequest object: " + e.what());
       } catch (...) {
         ORT_THROW(log_tag + "Exception while creating InferRequest object.");
@@ -101,7 +111,8 @@ VADMBackend::VADMBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
   else {
     i = GetFirstAvailableDevice(global_context);
     LOGS_DEFAULT(INFO) << log_tag << "Device Tag is: " << i;
-#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
     config[InferenceEngine::HDDL_DEVICE_TAG] = global_context_.deviceTags[i];
 #else
     config[VPU_HDDL_CONFIG_KEY(DEVICE_TAG)] = global_context_.deviceTags[i];
@@ -109,7 +120,7 @@ VADMBackend::VADMBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
     InferenceEngine::ExecutableNetwork exe_network;
     try {
       exe_network = global_context_.ie_core.LoadNetwork(*ie_cnn_network_, hw_target, config);
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + " Exception while Loading Network for graph: " + subgraph_context_.subgraph_name + e.what());
     } catch (...) {
       ORT_THROW(log_tag + " Exception while Loading Network for graph " + subgraph_context_.subgraph_name);
@@ -117,8 +128,8 @@ VADMBackend::VADMBackend(const ONNX_NAMESPACE::ModelProto& model_proto,
     LOGS_DEFAULT(INFO) << log_tag << "Loaded model to the plugin";
     InferenceEngine::InferRequest::Ptr infRequest;
     try {
-      infRequest = exe_network.CreateInferRequestPtr();
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+      infRequest = std::make_shared<InferenceEngine::InferRequest>(exe_network.CreateInferRequest());
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + "Exception while creating InferRequest object: " + e.what());
     } catch (...) {
       ORT_THROW(log_tag + "Exception while creating InferRequest object.");
@@ -143,7 +154,7 @@ void VADMBackend::StartAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext* c
     std::string input_name = input_info_iter->first;
     try {
       graph_input_blob = infer_request->GetBlob(input_name);
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + " Cannot access IE Blob for input: " + input_name + e.what());
     } catch (...) {
       ORT_THROW(log_tag + " Cannot access IE Blob for input: " + input_name);
@@ -155,7 +166,7 @@ void VADMBackend::StartAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext* c
   // Start Async inference
   try {
     infer_request->StartAsync();
-  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+  } catch (const Exception& e) {
     ORT_THROW(log_tag + " Couldn't start Inference: " + e.what());
   } catch (...) {
     ORT_THROW(log_tag + " Couldn't start Inference");
@@ -171,8 +182,8 @@ void VADMBackend::CompleteAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext
 
   // Wait for Async inference completion
   try {
-    infer_request->Wait(InferenceEngine::IInferRequest::WaitMode::RESULT_READY);
-  } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    infer_request->Wait(WaitMode::RESULT_READY);
+  } catch (const Exception& e) {
     ORT_THROW(log_tag + " Exception with completing Inference: " + e.what());
   } catch (...) {
     ORT_THROW(log_tag + " Exception with completing Inference");
@@ -186,7 +197,7 @@ void VADMBackend::CompleteAsyncInference(Ort::CustomOpApi& ort, OrtKernelContext
     auto output_name = output_info_iter->first;
     try {
       graph_output_blob = infer_request->GetBlob(output_name);
-    } catch (const InferenceEngine::details::InferenceEngineException& e) {
+    } catch (const Exception& e) {
       ORT_THROW(log_tag + " Cannot access IE Blob for output: " + output_name + e.what());
     } catch (...) {
       ORT_THROW(log_tag + " Cannot access IE Blob for output: " + output_name);
@@ -246,7 +257,8 @@ void VADMBackend::Infer(Ort::CustomOpApi& ort, OrtKernelContext* context) {
   size_t remainder_parallel_runs = batch_size % num_inf_reqs_;
 
   if (subgraph_context_.is_constant) {
-#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || defined(OPENVINO_2021_3)
+#if defined(OPENVINO_2020_4) || defined(OPENVINO_2021_1) || defined(OPENVINO_2021_2) || \
+    defined(OPENVINO_2021_3) || defined(OPENVINO_2021_4)
     for (auto item : const_outputs_map_) {
       auto out_name = item.first;
       auto node = item.second;
diff --git a/onnxruntime/core/providers/openvino/openvino_execution_provider.cc b/onnxruntime/core/providers/openvino/openvino_execution_provider.cc
index b25a1f5ea2..1de082c440 100644
--- a/onnxruntime/core/providers/openvino/openvino_execution_provider.cc
+++ b/onnxruntime/core/providers/openvino/openvino_execution_provider.cc
@@ -86,6 +86,10 @@ OpenVINOExecutionProvider::GetCapability(const GraphViewer& graph_viewer, const
   openvino_ep::GetCapability obj(graph_viewer,
                                  openvino_ep::BackendManager::GetGlobalContext().device_type, "V_2021_3");
   result = obj.Execute();
+#elif defined (OPENVINO_2021_4)
+  openvino_ep::GetCapability obj(graph_viewer,
+                                 openvino_ep::BackendManager::GetGlobalContext().device_type, "V_2021_3");
+  result = obj.Execute();
 #endif
 
   return result;
diff --git a/onnxruntime/core/providers/openvino/openvino_execution_provider.h b/onnxruntime/core/providers/openvino/openvino_execution_provider.h
index 3db730e29c..469ce4e9c0 100644
--- a/onnxruntime/core/providers/openvino/openvino_execution_provider.h
+++ b/onnxruntime/core/providers/openvino/openvino_execution_provider.h
@@ -6,6 +6,7 @@
 #include "backend_manager.h"
 #include <map>
 #include <algorithm>
+#include <iostream>
 
 namespace onnxruntime {