diff --git a/lite/CMakeLists.txt b/lite/CMakeLists.txt
index b89a4de37aafdc17c10fe6cb58b7bda272cc69fb..3a9e77e35d110567f2894654aacc175552c9caf5 100644
--- a/lite/CMakeLists.txt
+++ b/lite/CMakeLists.txt
@@ -188,15 +188,17 @@ if (LITE_WITH_CUDA OR LITE_WITH_X86)
             COMMAND cp "${CMAKE_BINARY_DIR}/libpaddle_api_light_bundled.a" "${INFER_LITE_PUBLISH_ROOT}/cxx/lib"
             COMMAND cp "${CMAKE_BINARY_DIR}/lite/api/*.so" "${INFER_LITE_PUBLISH_ROOT}/cxx/lib"
             )
-        add_custom_target(publish_inference_third_party ${TARGET}
-                COMMAND mkdir -p "${INFER_LITE_PUBLISH_ROOT}/third_party"
-                COMMAND cp -r "${CMAKE_BINARY_DIR}/third_party/install/*" "${INFER_LITE_PUBLISH_ROOT}/third_party")
+        if (LITE_WITH_CUDA)
+            add_custom_target(publish_inference_third_party ${TARGET}
+                    COMMAND mkdir -p "${INFER_LITE_PUBLISH_ROOT}/third_party"
+                    COMMAND cp -r "${CMAKE_BINARY_DIR}/third_party/install/*" "${INFER_LITE_PUBLISH_ROOT}/third_party")
+            add_dependencies(publish_inference publish_inference_third_party)
+        endif()
         add_dependencies(publish_inference_cxx_lib bundle_full_api)
         add_dependencies(publish_inference_cxx_lib bundle_light_api)
         add_dependencies(publish_inference_cxx_lib paddle_full_api_shared)
         add_dependencies(publish_inference_cxx_lib paddle_light_api_shared)
         add_dependencies(publish_inference publish_inference_cxx_lib)
-        add_dependencies(publish_inference publish_inference_third_party)
     endif()
 endif()
 
@@ -238,9 +240,13 @@ if (LITE_WITH_X86)
     add_dependencies(publish_inference_x86_cxx_lib test_model_bin)
 
     add_custom_target(publish_inference_x86_cxx_demos ${TARGET}
+           COMMAND rm -rf "${INFER_LITE_PUBLISH_ROOT}/demo/cxx"
+           COMMAND mkdir -p "${INFER_LITE_PUBLISH_ROOT}/demo/cxx"
+           COMMAND cp -r "${CMAKE_SOURCE_DIR}/lite/demo/cxx/x86_mobilenetv1_light_demo" "${INFER_LITE_PUBLISH_ROOT}/demo/cxx/mobilenetv1_light"
+           COMMAND cp -r "${CMAKE_SOURCE_DIR}/lite/demo/cxx/x86_mobilenetv1_full_demo" "${INFER_LITE_PUBLISH_ROOT}/demo/cxx/mobilenetv1_full"
            COMMAND mkdir -p "${INFER_LITE_PUBLISH_ROOT}/third_party"
-           COMMAND cp -r "${CMAKE_BINARY_DIR}/third_party/eigen3" "${INFER_LITE_PUBLISH_ROOT}/third_party"
-           )
+           COMMAND cp -r "${CMAKE_BINARY_DIR}/third_party/install/mklml" "${INFER_LITE_PUBLISH_ROOT}/third_party/"
+       )
     add_dependencies(publish_inference_x86_cxx_lib publish_inference_x86_cxx_demos)
     add_dependencies(publish_inference_x86_cxx_demos paddle_full_api_shared eigen3)
     add_dependencies(publish_inference publish_inference_x86_cxx_lib)
diff --git a/lite/demo/cxx/x86_mobilenetv1_full_demo/CMakeLists.txt b/lite/demo/cxx/x86_mobilenetv1_full_demo/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..5039ef7727c089e04ef49bd3c559a0103aa767e1
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_full_demo/CMakeLists.txt
@@ -0,0 +1,21 @@
+cmake_minimum_required(VERSION 2.8)
+
+set(TARGET mobilenet_full_api)
+
+# 1. path to Paddle-Lite lib and mklml lib
+set(LITE_DIR "${PROJECT_SOURCE_DIR}/../../../cxx")
+set(MKLML_DIR "${PROJECT_SOURCE_DIR}/../../../third_party/mklml/")
+
+# 2. link mklml and Paddle-Lite directory
+link_directories(${LITE_DIR}/lib ${MKLML_DIR}/lib)
+include_directories(${LITE_DIR}/include/ ${MKLML_DIR}/include)
+
+# 3. compile options 
+add_definitions(-std=c++11 -g -O3 -pthread)
+set(EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR})
+
+# 4.add executable output
+add_executable(${TARGET} ${TARGET}.cc)
+target_link_libraries(${TARGET} -lpaddle_full_api_shared)
+target_link_libraries(${TARGET} -lmklml_intel)
+target_link_libraries(${TARGET} -ldl)
diff --git a/lite/demo/cxx/x86_mobilenetv1_full_demo/build.sh b/lite/demo/cxx/x86_mobilenetv1_full_demo/build.sh
new file mode 100644
index 0000000000000000000000000000000000000000..c9570e326e361d40b9a2b857dc97a1caf1450a92
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_full_demo/build.sh
@@ -0,0 +1,6 @@
+mkdir ./build
+cd ./build
+cmake ..
+make
+cd ..
+rm -rf ./build
diff --git a/lite/demo/cxx/x86_mobilenetv1_full_demo/mobilenet_full_api.cc b/lite/demo/cxx/x86_mobilenetv1_full_demo/mobilenet_full_api.cc
new file mode 100644
index 0000000000000000000000000000000000000000..c2837e0fdd9bfaa9fc146dff9daee963f707b886
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_full_demo/mobilenet_full_api.cc
@@ -0,0 +1,66 @@
+// Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include <iostream>
+#include <vector>
+#include "paddle_api.h"  // NOLINT
+
+using namespace paddle::lite_api;  // NOLINT
+
+int64_t ShapeProduction(const shape_t& shape) {
+  int64_t res = 1;
+  for (auto i : shape) res *= i;
+  return res;
+}
+
+void RunModel(std::string model_dir) {
+  // 1. Create CxxConfig
+  CxxConfig config;
+  config.set_model_dir(model_dir);
+  config.set_valid_places({Place{TARGET(kX86), PRECISION(kFloat)},
+                           Place{TARGET(kHost), PRECISION(kFloat)}});
+  // 2. Create PaddlePredictor by CxxConfig
+  std::shared_ptr<PaddlePredictor> predictor =
+      CreatePaddlePredictor<CxxConfig>(config);
+
+  // 3. Prepare input data
+  std::unique_ptr<Tensor> input_tensor(std::move(predictor->GetInput(0)));
+  input_tensor->Resize({1, 3, 224, 224});
+  auto* data = input_tensor->mutable_data<float>();
+  for (int i = 0; i < ShapeProduction(input_tensor->shape()); ++i) {
+    data[i] = 1;
+  }
+
+  // 4. Run predictor
+  predictor->Run();
+
+  // 5. Get output
+  std::unique_ptr<const Tensor> output_tensor(
+      std::move(predictor->GetOutput(0)));
+  std::cout << "Output shape " << output_tensor->shape()[1] << std::endl;
+  for (int i = 0; i < ShapeProduction(output_tensor->shape()); i += 100) {
+    std::cout << "Output[" << i << "]: " << output_tensor->data<float>()[i]
+              << std::endl;
+  }
+}
+
+int main(int argc, char** argv) {
+  if (argc < 2) {
+    std::cerr << "[ERROR] usage: ./" << argv[0] << " naive_buffer_model_dir\n";
+    exit(1);
+  }
+  std::string model_dir = argv[1];
+  RunModel(model_dir);
+  return 0;
+}
diff --git a/lite/demo/cxx/x86_mobilenetv1_light_demo/CMakeLists.txt b/lite/demo/cxx/x86_mobilenetv1_light_demo/CMakeLists.txt
new file mode 100644
index 0000000000000000000000000000000000000000..6f917b5353b066b86547e6c0b31ab643e876ead1
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_light_demo/CMakeLists.txt
@@ -0,0 +1,21 @@
+cmake_minimum_required(VERSION 2.8)
+
+set(TARGET mobilenet_light_api)
+
+# 1. path to Paddle-Lite lib and mklml lib
+set(LITE_DIR "${PROJECT_SOURCE_DIR}/../../../cxx")
+set(MKLML_DIR "${PROJECT_SOURCE_DIR}/../../../third_party/mklml/")
+
+# 2. link mklml and Paddle-Lite directory
+link_directories(${LITE_DIR}/lib ${MKLML_DIR}/lib)
+include_directories(${LITE_DIR}/include/ ${MKLML_DIR}/include)
+
+# 3. compile options 
+add_definitions(-std=c++11 -g -O3 -pthread)
+set(EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR})
+
+# 4.add executable output
+add_executable(${TARGET} ${TARGET}.cc)
+target_link_libraries(${TARGET} -lpaddle_light_api_shared)
+target_link_libraries(${TARGET} -lmklml_intel)
+target_link_libraries(${TARGET} -ldl)
diff --git a/lite/demo/cxx/x86_mobilenetv1_light_demo/build.sh b/lite/demo/cxx/x86_mobilenetv1_light_demo/build.sh
new file mode 100644
index 0000000000000000000000000000000000000000..c9570e326e361d40b9a2b857dc97a1caf1450a92
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_light_demo/build.sh
@@ -0,0 +1,6 @@
+mkdir ./build
+cd ./build
+cmake ..
+make
+cd ..
+rm -rf ./build
diff --git a/lite/demo/cxx/x86_mobilenetv1_light_demo/mobilenet_light_api.cc b/lite/demo/cxx/x86_mobilenetv1_light_demo/mobilenet_light_api.cc
new file mode 100644
index 0000000000000000000000000000000000000000..17997a228f03d84ab23638df9b64ae30ef844861
--- /dev/null
+++ b/lite/demo/cxx/x86_mobilenetv1_light_demo/mobilenet_light_api.cc
@@ -0,0 +1,64 @@
+// Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include <iostream>
+#include <vector>
+#include "paddle_api.h"  // NOLINT
+
+using namespace paddle::lite_api;  // NOLINT
+
+int64_t ShapeProduction(const shape_t& shape) {
+  int64_t res = 1;
+  for (auto i : shape) res *= i;
+  return res;
+}
+
+void RunModel(std::string model_dir) {
+  // 1. Create MobileConfig
+  MobileConfig config;
+  config.set_model_dir(model_dir);
+  // 2. Create PaddlePredictor by CxxConfig
+  std::shared_ptr<PaddlePredictor> predictor =
+      CreatePaddlePredictor<MobileConfig>(config);
+
+  // 3. Prepare input data
+  std::unique_ptr<Tensor> input_tensor(std::move(predictor->GetInput(0)));
+  input_tensor->Resize({1, 3, 224, 224});
+  auto* data = input_tensor->mutable_data<float>();
+  for (int i = 0; i < ShapeProduction(input_tensor->shape()); ++i) {
+    data[i] = 1;
+  }
+
+  // 4. Run predictor
+  predictor->Run();
+
+  // 5. Get output
+  std::unique_ptr<const Tensor> output_tensor(
+      std::move(predictor->GetOutput(0)));
+  std::cout << "Output shape " << output_tensor->shape()[1] << std::endl;
+  for (int i = 0; i < ShapeProduction(output_tensor->shape()); i += 100) {
+    std::cout << "Output[" << i << "]: " << output_tensor->data<float>()[i]
+              << std::endl;
+  }
+}
+
+int main(int argc, char** argv) {
+  if (argc < 2) {
+    std::cerr << "[ERROR] usage: ./" << argv[0] << " naive_buffer_model_dir\n";
+    exit(1);
+  }
+  std::string model_dir = argv[1];
+  RunModel(model_dir);
+  return 0;
+}