diff --git a/.ci/azure/linux.yml b/.ci/azure/linux.yml
index 73d0f03b5349a0..592614b677ddf7 100644
--- a/.ci/azure/linux.yml
+++ b/.ci/azure/linux.yml
@@ -85,8 +85,9 @@ jobs:
       python3 -m pip install -r $(REPO_DIR)/inference-engine/ie_bridges/python/wheel/requirements-dev.txt
       # For running Python API tests
       python3 -m pip install -r $(REPO_DIR)/inference-engine/ie_bridges/python/src/requirements-dev.txt
-      # For running nGraph unit tests dependent on Python frameworks
-      python3 -m pip install -r $(REPO_DIR)/ngraph/test/requirements_test.txt
+      # For running PaddlePaddle frontend unit tests
+      python3 -m pip install -r $(REPO_DIR)/ngraph/test/frontend/paddlepaddle/requirements_dev.txt
+      # For running ONNX frontend unit tests
       python3 -m pip install -r $(REPO_DIR)/ngraph/test/requirements_test_onnx.txt
       # For MO unit tests
       python3 -m pip install -r $(REPO_DIR)/model-optimizer/requirements.txt
@@ -171,13 +172,15 @@ jobs:
     displayName: 'Model Optimizer UT'
     continueOnError: false
 
-  - script: |
-      export FE_TEST_MODELS=$(INSTALL_DIR)/tests
-      . $(SETUPVARS) && $(INSTALL_TEST_DIR)/unit-test --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-NGraphUT.xml
+  - script: . $(SETUPVARS) && $(INSTALL_TEST_DIR)/unit-test --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-NGraphUT.xml
     workingDirectory: $(INSTALL_TEST_DIR)
     displayName: 'nGraph UT'
     continueOnError: false
 
+  - script: . $(SETUPVARS) && $(INSTALL_TEST_DIR)/paddlepaddle_tests --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-PaddlePaddle.xml
+    displayName: 'PaddlePaddle Frontend UT'
+    continueOnError: false
+
     # . $(SETUPVARS) && python3 $(WORK_DIR)/gtest-parallel/gtest_parallel.py $(INSTALL_TEST_DIR)/InferenceEngineUnitTests --workers=16 --dump_json_test_results=InferenceEngineUnitTests.json --gtest_filter=*smoke* -- --gtest_print_time=1
   - script: . $(SETUPVARS) && $(INSTALL_TEST_DIR)/InferenceEngineUnitTests --gtest_print_time=1 --gtest_output=xml:TEST-InferenceEngineUnitTests.xml
     displayName: 'IE UT old'
diff --git a/.ci/azure/windows.yml b/.ci/azure/windows.yml
index 368797841c5505..e5ec0486f9bcd2 100644
--- a/.ci/azure/windows.yml
+++ b/.ci/azure/windows.yml
@@ -86,8 +86,9 @@ jobs:
       python -m pip install --upgrade pip
       rem For running Python API tests
       python -m pip install -r $(REPO_DIR)\inference-engine\ie_bridges\python\src\requirements-dev.txt
-      rem For running nGraph unit tests dependent on Python frameworks
-      python -m pip install -r $(REPO_DIR)\ngraph\test\requirements_test.txt
+      rem For running PaddlePaddle frontend unit tests
+      python -m pip install -r $(REPO_DIR)\ngraph\test\frontend\paddlepaddle\requirements_dev.txt
+      rem For running ONNX frontend unit tests
       python -m pip install -r $(REPO_DIR)\ngraph\test\requirements_test_onnx.txt
       rem For MO unit tests
       python -m pip install -r $(REPO_DIR)\model-optimizer\requirements.txt
@@ -143,13 +144,15 @@ jobs:
     displayName: 'Clean build dir'
     continueOnError: false
 
-  - script: |
-      set FE_TEST_MODELS=$(INSTALL_DIR)\tests
-      call $(SETUPVARS) && $(INSTALL_TEST_DIR)\unit-test --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-NGraphUT.xml
+  - script: call $(SETUPVARS) && $(INSTALL_TEST_DIR)\unit-test --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-NGraphUT.xml
     workingDirectory: $(INSTALL_TEST_DIR)
     displayName: 'nGraph UT'
     continueOnError: false
 
+  - script: call $(SETUPVARS) && $(INSTALL_TEST_DIR)\paddlepaddle_tests --gtest_print_time=1 --gtest_filter=-backend_api.config_unsupported:*IE_GPU* --gtest_output=xml:TEST-PaddlePaddle.xml
+    displayName: 'PaddlePaddle Frontend UT'
+    continueOnError: false
+
   - script: |
       set PATH=$(IB_DIR);%PATH%
       call $(SETUPVARS) && "$(IB_TESTCONSOLE)" $(INSTALL_TEST_DIR)\InferenceEngineUnitTests.exe --gtest_output=xml:TEST-InferenceEngineUnitTests-IB.xml
diff --git a/cmake/test_model_zoo.cmake b/cmake/test_model_zoo.cmake
index 06f14f2d1f5e82..c3f158626cdabe 100644
--- a/cmake/test_model_zoo.cmake
+++ b/cmake/test_model_zoo.cmake
@@ -120,6 +120,10 @@ if(ENABLE_TESTS)
         add_dependencies(test_model_zoo test_pip_prerequsites)
     endif()
 
+    if (NGRAPH_PDPD_FRONTEND_ENABLE)
+        add_dependencies(test_model_zoo paddlepaddle_test_models)
+    endif()
+
     install(DIRECTORY "${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/test_model_zoo"
             DESTINATION tests COMPONENT tests EXCLUDE_FROM_ALL)
 
diff --git a/ngraph/test/CMakeLists.txt b/ngraph/test/CMakeLists.txt
index 956a224df1ceec..37ca03d1d7da7d 100644
--- a/ngraph/test/CMakeLists.txt
+++ b/ngraph/test/CMakeLists.txt
@@ -558,31 +558,10 @@ if (NGRAPH_ONNX_IMPORT_ENABLE)
 endif()
 
 # SOURCE FOR FRONTEND TESTING
-
 file(GLOB FRONTEND_TESTS_SRC ${CMAKE_CURRENT_SOURCE_DIR}/frontend/frontend_manager.cpp)
 set(SRC ${FRONTEND_TESTS_SRC} ${SRC})
 
-file(GLOB FRONTEND_SHARED_TESTS_SRC ${CMAKE_CURRENT_SOURCE_DIR}/frontend/shared/src/*.cpp)
-file(GLOB FRONTEND_SHARED_TESTS_HDR ${CMAKE_CURRENT_SOURCE_DIR}/frontend/shared/include/*.hpp)
-set(SRC ${FRONTEND_SHARED_TESTS_SRC} ${SRC})
-
-add_definitions("-DTEST_MODEL_BUILD_DIR=\"${CMAKE_CURRENT_BINARY_DIR}\"")
-
-# ---- PaddlePaddle FrontEnd testing ------
-if (NGRAPH_PDPD_FRONTEND_ENABLE)
-    ie_check_pip_package(paddlepaddle WARNING)
-
-    if(paddlepaddle_FOUND)
-        file(GLOB FRONTEND_PDPD_TESTS_SRC ${CMAKE_CURRENT_SOURCE_DIR}/frontend/paddlepaddle/*.cpp)
-        set(SRC ${FRONTEND_PDPD_TESTS_SRC} ${SRC})
-        set(PADDLE_TEST_MODELS_DIRNAME paddle_test_models)
-        set(TEST_PADDLE_MODELS ${CMAKE_CURRENT_BINARY_DIR}/${PADDLE_TEST_MODELS_DIRNAME}/)
-        add_definitions("-DTEST_PADDLE_MODELS_DIRNAME=\"${PADDLE_TEST_MODELS_DIRNAME}/\"")
-    endif()
-endif()
-# ---- End PaddlePaddle FrontEnd testing ------
-
-add_clang_format_target(unit-test_clang FOR_SOURCES ${SRC} ${MULTI_TEST_SRC} ${FRONTEND_SHARED_TESTS_HDR})
+add_clang_format_target(unit-test_clang FOR_SOURCES ${SRC} ${MULTI_TEST_SRC})
 
 foreach(BACKEND_NAME ${ACTIVE_BACKEND_LIST})
     string(TOLOWER ${BACKEND_NAME} BACKEND_DIR)
@@ -603,7 +582,6 @@ add_executable(unit-test ${SRC})
 
 target_include_directories(unit-test PRIVATE ".")
 target_include_directories(unit-test PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/runtime)
-target_include_directories(unit-test PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/frontend/shared/include)
 
 add_definitions("-DCURDIR=\"${CMAKE_CURRENT_SOURCE_DIR}\"")
 add_definitions("-DJSON_INCLUDES=\"${JSON_INCLUDE_DIR}\"")
@@ -618,7 +596,8 @@ target_link_libraries(unit-test PRIVATE ngraph_test_util
                                         ie_backend
                                         interpreter_backend
                                         Threads::Threads
-                                        openvino::conditional_compilation)
+                                        openvino::conditional_compilation
+                                        frontend_manager)
 
 # Protobuf-lite does not support parsing files from prototxt format
 # Since most of the onnx models are stored in this format it have to be disabled
@@ -655,39 +634,7 @@ install(TARGETS unit-test
         COMPONENT tests
         EXCLUDE_FROM_ALL)
 
-############ FRONTEND ############
-target_include_directories(unit-test PRIVATE ${FRONTEND_INCLUDE_PATH} frontend/shared/include)
-target_link_libraries(unit-test PRIVATE frontend_manager cnpy commonTestUtils)
-
 add_subdirectory(frontend)
-### END FRONTEND ###
-
-#PaddlePaddle - test models generator
-if (NGRAPH_PDPD_FRONTEND_ENABLE AND paddlepaddle_FOUND)
-    file(GLOB_RECURSE PDPD_GEN_SCRIPTS ${CMAKE_CURRENT_SOURCE_DIR}/files/paddlepaddle/gen_scripts/generate_*.py)
-    set(OUT_FILES "")
-    foreach(GEN_SCRIPT ${PDPD_GEN_SCRIPTS})
-        get_filename_component(FILE_WE ${GEN_SCRIPT} NAME_WE)
-        set(OUT_DONE_FILE ${TEST_PADDLE_MODELS}/${FILE_WE}_done.txt)
-        set(OUT_FILES ${OUT_DONE_FILE} ${OUT_FILES})
-        add_custom_command(OUTPUT ${OUT_DONE_FILE}
-                COMMAND ${PYTHON_EXECUTABLE}
-                        ${CMAKE_CURRENT_SOURCE_DIR}/files/paddlepaddle/gen_wrapper.py
-                        ${GEN_SCRIPT}
-                        ${TEST_PADDLE_MODELS}
-                        ${OUT_DONE_FILE}
-                DEPENDS ${GEN_SCRIPT} ${CMAKE_CURRENT_SOURCE_DIR}/files/paddlepaddle/gen_wrapper.py
-                )
-    endforeach()
-    add_custom_target(paddlepaddle_test_models DEPENDS ${OUT_FILES})
-    add_dependencies(unit-test paddlepaddle_test_models)
-    add_dependencies(unit-test paddlepaddle_ngraph_frontend)
-
-    install(DIRECTORY ${TEST_PADDLE_MODELS}
-            DESTINATION tests/${PADDLE_TEST_MODELS_DIRNAME}
-            COMPONENT tests
-            EXCLUDE_FROM_ALL)
-endif()
 
 # process models
 add_dependencies(unit-test test_model_zoo)
diff --git a/ngraph/test/frontend/CMakeLists.txt b/ngraph/test/frontend/CMakeLists.txt
index ba20be7d338f2f..67ffb5a992ac1b 100644
--- a/ngraph/test/frontend/CMakeLists.txt
+++ b/ngraph/test/frontend/CMakeLists.txt
@@ -1,6 +1,11 @@
 # Copyright (C) 2018-2021 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 #
+add_subdirectory(shared)
+
+if (NGRAPH_PDPD_FRONTEND_ENABLE)
+    add_subdirectory(paddlepaddle)
+endif()
 
 set(SRC ${CMAKE_CURRENT_SOURCE_DIR}/mock_frontend.cpp)
 add_library(mock1_ngraph_frontend SHARED ${SRC})
diff --git a/ngraph/test/frontend/paddlepaddle/CMakeLists.txt b/ngraph/test/frontend/paddlepaddle/CMakeLists.txt
new file mode 100644
index 00000000000000..52b898192d0beb
--- /dev/null
+++ b/ngraph/test/frontend/paddlepaddle/CMakeLists.txt
@@ -0,0 +1,70 @@
+# Copyright (C) 2018-2021 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+#
+
+set(TARGET_NAME "paddlepaddle_tests")
+
+file(GLOB_RECURSE SRC ${CMAKE_CURRENT_SOURCE_DIR}/*.cpp)
+
+add_executable(${TARGET_NAME} ${SRC})
+
+target_link_libraries(${TARGET_NAME} PRIVATE frontend_shared_test_classes)
+
+add_clang_format_target(${TARGET_NAME}_clang FOR_TARGETS ${TARGET_NAME})
+
+install(TARGETS ${TARGET_NAME}
+        RUNTIME DESTINATION tests
+        COMPONENT tests
+        EXCLUDE_FROM_ALL)
+
+# Test model generating
+ie_check_pip_package(paddlepaddle WARNING)
+
+set(TEST_PADDLE_MODELS_DIRNAME test_model_zoo/paddle_test_models)
+target_compile_definitions(${TARGET_NAME} PRIVATE -D TEST_PADDLE_MODELS_DIRNAME=\"${TEST_PADDLE_MODELS_DIRNAME}/\")
+
+# If 'paddlepaddle' is not found, code will still be compiled
+# but models will not be generated and tests will fail
+# This is done this way for 'code style' and check cases - cmake shall pass, but CI machine doesn't need to have
+# 'paddlepaddle' installed to check code style
+if (paddlepaddle_FOUND)
+    set(TEST_PADDLE_MODELS ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/${TEST_PADDLE_MODELS_DIRNAME}/)
+
+    file(GLOB_RECURSE PADDLE_GEN_SCRIPTS ${CMAKE_CURRENT_SOURCE_DIR}/test_models/gen_scripts/generate_*.py)
+    file(GLOB_RECURSE PADDLE_ALL_SCRIPTS ${CMAKE_CURRENT_SOURCE_DIR}/*.py)
+    set(OUT_FILES "")
+    foreach(GEN_SCRIPT ${PADDLE_GEN_SCRIPTS})
+        get_filename_component(FILE_WE ${GEN_SCRIPT} NAME_WE)
+        set(OUT_DONE_FILE ${TEST_PADDLE_MODELS}/${FILE_WE}_done.txt)
+        set(OUT_FILES ${OUT_DONE_FILE} ${OUT_FILES})
+        add_custom_command(OUTPUT ${OUT_DONE_FILE}
+                COMMAND ${PYTHON_EXECUTABLE}
+                ${CMAKE_CURRENT_SOURCE_DIR}/test_models/gen_wrapper.py
+                ${GEN_SCRIPT}
+                ${TEST_PADDLE_MODELS}
+                ${OUT_DONE_FILE}
+                DEPENDS ${PADDLE_ALL_SCRIPTS}
+                )
+    endforeach()
+    add_custom_target(paddlepaddle_test_models DEPENDS ${OUT_FILES})
+
+    install(DIRECTORY ${TEST_PADDLE_MODELS}
+            DESTINATION tests/${TEST_PADDLE_MODELS_DIRNAME}
+            COMPONENT tests
+            EXCLUDE_FROM_ALL)
+else()
+    # Produce warning message at build time as well
+    add_custom_command(OUTPUT unable_build_paddle_models.txt
+            COMMAND ${CMAKE_COMMAND}
+            -E cmake_echo_color --red "Warning: Unable to generate PaddlePaddle test models. Running '${TARGET_NAME}' will likely fail"
+            )
+    add_custom_target(paddlepaddle_test_models DEPENDS unable_build_paddle_models.txt)
+endif()
+
+add_dependencies(${TARGET_NAME} paddlepaddle_test_models)
+add_dependencies(${TARGET_NAME} paddlepaddle_ngraph_frontend)
+
+# Fuzzy tests for PaddlePaddle use IE_CPU engine
+if (ENABLE_MKL_DNN)
+    add_dependencies(${TARGET_NAME} MKLDNNPlugin)
+endif()
\ No newline at end of file
diff --git a/ngraph/test/frontend/paddlepaddle/incorrect_cut_model.cpp b/ngraph/test/frontend/paddlepaddle/incorrect_cut_model.cpp
index 0065dc2231f377..42cee6c6c70e25 100644
--- a/ngraph/test/frontend/paddlepaddle/incorrect_cut_model.cpp
+++ b/ngraph/test/frontend/paddlepaddle/incorrect_cut_model.cpp
@@ -21,8 +21,7 @@ TEST(FrontEndIncorrectCutModelTest, test_incorrect_cut)
     ASSERT_NO_THROW(frontEnd = fem.load_by_framework(PADDLE_FE));
     ASSERT_NE(frontEnd, nullptr);
     auto model_filename = FrontEndTestUtils::make_model_path(
-        std::string(TEST_PADDLE_MODELS_DIRNAME) +
-        std::string("2in_2out/2in_2out.pdmodel"));
+        std::string(TEST_PADDLE_MODELS_DIRNAME) + std::string("2in_2out/2in_2out.pdmodel"));
     ASSERT_NO_THROW(inputModel = frontEnd->load(model_filename));
     ASSERT_NE(inputModel, nullptr);
 
diff --git a/ngraph/test/frontend/paddlepaddle/main.cpp b/ngraph/test/frontend/paddlepaddle/main.cpp
new file mode 100644
index 00000000000000..f3c89e3ae80314
--- /dev/null
+++ b/ngraph/test/frontend/paddlepaddle/main.cpp
@@ -0,0 +1,10 @@
+// Copyright (C) 2018-2021 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+//
+
+#include "utils.hpp"
+
+int main(int argc, char** argv)
+{
+    return FrontEndTestUtils::run_tests(argc, argv);
+}
diff --git a/ngraph/test/frontend/paddlepaddle/paddle_utils.hpp b/ngraph/test/frontend/paddlepaddle/paddle_utils.hpp
index 78960af5a05598..270a56e2e9da7a 100644
--- a/ngraph/test/frontend/paddlepaddle/paddle_utils.hpp
+++ b/ngraph/test/frontend/paddlepaddle/paddle_utils.hpp
@@ -7,3 +7,4 @@
 #include <string>
 
 static const std::string PADDLE_FE = "paddle";
+
diff --git a/ngraph/test/frontend/paddlepaddle/places.cpp b/ngraph/test/frontend/paddlepaddle/places.cpp
index 2da7d59c93b37e..60800085950baa 100644
--- a/ngraph/test/frontend/paddlepaddle/places.cpp
+++ b/ngraph/test/frontend/paddlepaddle/places.cpp
@@ -56,9 +56,11 @@ std::vector<std::string> tensor_names = {
 TEST(PDPD_Places, check_tensor_names)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -70,9 +72,11 @@ TEST(PDPD_Places, check_tensor_names)
 TEST(PDPD_Places, check_input_outputs)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     auto inputs = input_model->get_inputs();
     auto outputs = input_model->get_outputs();
@@ -105,9 +109,11 @@ TEST(PDPD_Places, check_input_outputs)
 TEST(PDPD_Places, check_out_port_of_all_ops)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -128,9 +134,11 @@ TEST(PDPD_Places, check_out_port_of_all_ops)
 TEST(PDPD_Places, check_in_out_ports_of_model_outputs)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     auto outputs = input_model->get_outputs();
     for (const auto& output : outputs)
@@ -163,9 +171,11 @@ TEST(PDPD_Places, check_in_out_ports_of_model_outputs)
 TEST(PDPD_Places, check_source_target_tensors_of_model_outputs)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     auto outputs = input_model->get_outputs();
     for (const auto& output : outputs)
@@ -198,9 +208,11 @@ TEST(PDPD_Places, check_source_target_tensors_of_model_outputs)
 TEST(PDPD_Places, check_producing_consuming_ops_of_model_outputs)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     auto outputs = input_model->get_outputs();
     for (const auto& output : outputs)
@@ -234,9 +246,11 @@ TEST(PDPD_Places, check_producing_consuming_ops_of_model_outputs)
 TEST(PDPD_Places, check_data_flow)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -275,9 +289,11 @@ TEST(PDPD_Places, check_data_flow)
 TEST(PDPD_Places, check_tensor_to_multiple_ports)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -307,9 +323,11 @@ TEST(PDPD_Places, check_tensor_to_multiple_ports)
 TEST(PDPD_Places, check_consuming_ops)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -352,9 +370,11 @@ TEST(PDPD_Places, check_consuming_ops)
 TEST(PDPD_Places, check_consuming_ops_2)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     auto it = find(tensor_names.begin(), tensor_names.end(), "lstm_0.tmp_2");
     EXPECT_NE(it, tensor_names.end());
@@ -394,9 +414,11 @@ TEST(PDPD_Places, check_consuming_ops_2)
 TEST(PDPD_Places, check_producing_ops)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     for (const auto& tensor_name : tensor_names)
     {
@@ -418,9 +440,11 @@ TEST(PDPD_Places, check_producing_ops)
 TEST(PDPD_Places, check_input_output_ports_dy_idx)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     std::vector<std::string> output_names = {"save_infer_model/scale_0.tmp_1",
                                              "save_infer_model/scale_1.tmp_1",
@@ -445,9 +469,11 @@ TEST(PDPD_Places, check_input_output_ports_dy_idx)
 TEST(PDPD_Places, check_ops_tensors_by_idx)
 {
     FrontEndTestUtils::setupTestEnv();
-    auto m_fem = FrontEndManager();
-    auto frontend = m_fem.load_by_framework(PADDLE_FE);
-    auto input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file));
+    auto fem = FrontEndManager();
+    FrontEnd::Ptr frontend;
+    ASSERT_NO_THROW(frontend = fem.load_by_framework(PADDLE_FE));
+    InputModel::Ptr input_model;
+    ASSERT_NO_THROW(input_model = frontend->load(FrontEndTestUtils::make_model_path(model_file)));
 
     std::vector<std::string> output_names = {"save_infer_model/scale_0.tmp_1",
                                              "save_infer_model/scale_1.tmp_1",
diff --git a/ngraph/test/requirements_test.txt b/ngraph/test/frontend/paddlepaddle/requirements_dev.txt
similarity index 63%
rename from ngraph/test/requirements_test.txt
rename to ngraph/test/frontend/paddlepaddle/requirements_dev.txt
index 9615699d8bd920..f9f5ae3d96c92b 100644
--- a/ngraph/test/requirements_test.txt
+++ b/ngraph/test/frontend/paddlepaddle/requirements_dev.txt
@@ -3,6 +3,4 @@ paddlepaddle==2.1.0
 gast==0.3.3
 numpy~=1.19.2
 six~=1.15.0
-# ONNX - generate test models
-docopt~=0.6.2
-onnx~=1.9.0
\ No newline at end of file
+
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_2in_2out.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_2in_2out.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_2in_2out.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_2in_2out.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_2in_2out_dynbatch.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_2in_2out_dynbatch.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_2in_2out_dynbatch.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_2in_2out_dynbatch.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_argmax.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_argmax.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_argmax.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_argmax.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_assign_value.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_assign_value.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_assign_value.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_assign_value.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_batch_norm.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_batch_norm.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_batch_norm.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_batch_norm.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_clip.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_clip.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_clip.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_clip.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d_relu.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d_relu.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d_relu.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d_relu.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d_s.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d_s.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_conv2d_s.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_conv2d_s.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_multi_tensor_split.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_multi_tensor_split.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_multi_tensor_split.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_multi_tensor_split.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_place_test_model.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_place_test_model.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_place_test_model.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_place_test_model.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_relu.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_relu.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_relu.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_relu.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/generate_unsupported_relu.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_unsupported_relu.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/generate_unsupported_relu.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/generate_unsupported_relu.py
diff --git a/ngraph/test/files/paddlepaddle/gen_scripts/save_model.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/save_model.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_scripts/save_model.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_scripts/save_model.py
diff --git a/ngraph/test/files/paddlepaddle/gen_wrapper.py b/ngraph/test/frontend/paddlepaddle/test_models/gen_wrapper.py
similarity index 100%
rename from ngraph/test/files/paddlepaddle/gen_wrapper.py
rename to ngraph/test/frontend/paddlepaddle/test_models/gen_wrapper.py
diff --git a/ngraph/test/frontend/shared/CMakeLists.txt b/ngraph/test/frontend/shared/CMakeLists.txt
new file mode 100644
index 00000000000000..c4dda394fb1a05
--- /dev/null
+++ b/ngraph/test/frontend/shared/CMakeLists.txt
@@ -0,0 +1,17 @@
+# Copyright (C) 2018-2021 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+#
+
+set(TARGET_NAME "frontend_shared_test_classes")
+
+file(GLOB_RECURSE LIBRARY_SRC ${CMAKE_CURRENT_SOURCE_DIR}/src/*.cpp)
+file(GLOB_RECURSE LIBRARY_HEADERS ${CMAKE_CURRENT_SOURCE_DIR}/include/*.hpp)
+
+add_library(${TARGET_NAME} STATIC ${LIBRARY_SRC} ${LIBRARY_HEADERS})
+
+target_include_directories(${TARGET_NAME} PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/include)
+target_include_directories(${TARGET_NAME} PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../..)
+target_link_libraries(${TARGET_NAME} PUBLIC frontend_manager
+        ngraph::ngraph cnpy ie_backend ngraph_test_util commonTestUtils)
+
+add_clang_format_target(${TARGET_NAME}_clang FOR_TARGETS ${TARGET_NAME})
diff --git a/ngraph/test/frontend/shared/include/utils.hpp b/ngraph/test/frontend/shared/include/utils.hpp
index 1b373d747d34fd..37828f1a09a919 100644
--- a/ngraph/test/frontend/shared/include/utils.hpp
+++ b/ngraph/test/frontend/shared/include/utils.hpp
@@ -4,15 +4,32 @@
 
 #pragma once
 
+#include <frontend_manager/frontend_manager.hpp>
 #include <fstream>
 #include <string>
 #include "backend.hpp"
+#include "common_test_utils/file_utils.hpp"
 #include "ngraph/env_util.hpp"
 #include "ngraph/file_util.hpp"
 
 // Helper functions
 namespace FrontEndTestUtils
 {
+    int run_tests(int argc, char** argv);
+
+    std::string get_current_executable_path();
+
+    inline std::tuple<ngraph::frontend::FrontEnd::Ptr, ngraph::frontend::InputModel::Ptr>
+        load_from_file(ngraph::frontend::FrontEndManager& fem,
+                       const std::string& frontend_name,
+                       const std::string& model_file)
+    {
+        auto frontend = fem.load_by_framework(frontend_name);
+        auto inputModel = frontend->load(model_file);
+        return std::tuple<ngraph::frontend::FrontEnd::Ptr, ngraph::frontend::InputModel::Ptr>{
+            frontend, inputModel};
+    }
+
     inline std::string fileToTestName(const std::string& fileName)
     {
         // TODO: GCC 4.8 has limited support of regex
@@ -57,16 +74,6 @@ namespace FrontEndTestUtils
 
     inline std::string make_model_path(const std::string& modelsRelativePath)
     {
-        // First try build path
-        auto res = std::string(TEST_MODEL_BUILD_DIR) + "/" + modelsRelativePath;
-        if (exists(res))
-        {
-            return res;
-        }
-        else
-        {
-            // Install case: if model file does not exist, use base path from env variable
-            return std::string(ngraph::getenv_string("FE_TEST_MODELS")) + "/" + modelsRelativePath;
-        }
+        return CommonTestUtils::getModelFromTestModelZoo(modelsRelativePath);
     }
 } // namespace FrontEndTestUtils
\ No newline at end of file
diff --git a/ngraph/test/frontend/shared/src/basic_api.cpp b/ngraph/test/frontend/shared/src/basic_api.cpp
index 8fc3c57ab2d3f5..92a7254c76ec2f 100644
--- a/ngraph/test/frontend/shared/src/basic_api.cpp
+++ b/ngraph/test/frontend/shared/src/basic_api.cpp
@@ -30,12 +30,8 @@ void FrontEndBasicTest::initParamTest()
 
 void FrontEndBasicTest::doLoadFromFile()
 {
-    std::vector<std::string> frontends;
-    ASSERT_NO_THROW(frontends = m_fem.get_available_front_ends());
-    ASSERT_NO_THROW(m_frontEnd = m_fem.load_by_framework(m_feName));
-    ASSERT_NE(m_frontEnd, nullptr);
-    ASSERT_NO_THROW(m_inputModel = m_frontEnd->load(m_modelFile));
-    ASSERT_NE(m_inputModel, nullptr);
+    std::tie(m_frontEnd, m_inputModel) =
+        FrontEndTestUtils::load_from_file(m_fem, m_feName, m_modelFile);
 }
 
 TEST_P(FrontEndBasicTest, testLoadFromFile)
diff --git a/ngraph/test/frontend/shared/src/cut_specific_model.cpp b/ngraph/test/frontend/shared/src/cut_specific_model.cpp
index 5edf87fae6571f..cc9527bf8f95c0 100644
--- a/ngraph/test/frontend/shared/src/cut_specific_model.cpp
+++ b/ngraph/test/frontend/shared/src/cut_specific_model.cpp
@@ -40,13 +40,8 @@ void FrontEndCutModelTest::initParamTest()
 
 void FrontEndCutModelTest::doLoadFromFile()
 {
-    std::vector<std::string> frontends;
-    FrontEnd::Ptr fe;
-    ASSERT_NO_THROW(frontends = m_fem.get_available_front_ends());
-    ASSERT_NO_THROW(m_frontEnd = m_fem.load_by_framework(m_param.m_frontEndName));
-    ASSERT_NE(m_frontEnd, nullptr);
-    ASSERT_NO_THROW(m_inputModel = m_frontEnd->load(m_param.m_modelName));
-    ASSERT_NE(m_inputModel, nullptr);
+    std::tie(m_frontEnd, m_inputModel) =
+        FrontEndTestUtils::load_from_file(m_fem, m_param.m_frontEndName, m_param.m_modelName);
 }
 
 std::vector<ngraph::frontend::Place::Ptr> FrontEndCutModelTest::constructNewInputs() const
diff --git a/ngraph/test/frontend/shared/src/main.cpp b/ngraph/test/frontend/shared/src/main.cpp
new file mode 100644
index 00000000000000..3828e04f333aed
--- /dev/null
+++ b/ngraph/test/frontend/shared/src/main.cpp
@@ -0,0 +1,17 @@
+// Copyright (C) 2018-2021 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+//
+
+#include "gtest/gtest.h"
+
+using namespace std;
+
+namespace FrontEndTestUtils
+{
+    int run_tests(int argc, char** argv)
+    {
+        ::testing::InitGoogleTest(&argc, argv);
+        int rc = RUN_ALL_TESTS();
+        return rc;
+    }
+} // namespace FrontEndTestUtils
\ No newline at end of file
diff --git a/ngraph/test/frontend/shared/src/op_fuzzy.cpp b/ngraph/test/frontend/shared/src/op_fuzzy.cpp
index 50a973cd9eee2d..a668de217fefb1 100644
--- a/ngraph/test/frontend/shared/src/op_fuzzy.cpp
+++ b/ngraph/test/frontend/shared/src/op_fuzzy.cpp
@@ -40,12 +40,8 @@ void FrontEndFuzzyOpTest::initParamTest()
 
 void FrontEndFuzzyOpTest::doLoadFromFile()
 {
-    std::vector<std::string> frontends;
-    ASSERT_NO_THROW(frontends = m_fem.get_available_front_ends());
-    ASSERT_NO_THROW(m_frontEnd = m_fem.load_by_framework(m_feName));
-    ASSERT_NE(m_frontEnd, nullptr);
-    ASSERT_NO_THROW(m_inputModel = m_frontEnd->load(m_modelFile));
-    ASSERT_NE(m_inputModel, nullptr);
+    std::tie(m_frontEnd, m_inputModel) =
+        FrontEndTestUtils::load_from_file(m_fem, m_feName, m_modelFile);
 }
 
 template <typename T>
diff --git a/ngraph/test/frontend/shared/src/partial_shape.cpp b/ngraph/test/frontend/shared/src/partial_shape.cpp
index a35c54680e5294..2d74fc270975d8 100644
--- a/ngraph/test/frontend/shared/src/partial_shape.cpp
+++ b/ngraph/test/frontend/shared/src/partial_shape.cpp
@@ -38,13 +38,8 @@ void FrontEndPartialShapeTest::initParamTest()
 
 void FrontEndPartialShapeTest::doLoadFromFile()
 {
-    std::vector<std::string> frontends;
-    FrontEnd::Ptr fe;
-    ASSERT_NO_THROW(frontends = m_fem.get_available_front_ends());
-    ASSERT_NO_THROW(m_frontEnd = m_fem.load_by_framework(m_baseParam.m_frontEndName));
-    ASSERT_NE(m_frontEnd, nullptr);
-    ASSERT_NO_THROW(m_inputModel = m_frontEnd->load(m_partShape.m_modelName));
-    ASSERT_NE(m_inputModel, nullptr);
+    std::tie(m_frontEnd, m_inputModel) = FrontEndTestUtils::load_from_file(
+        m_fem, m_baseParam.m_frontEndName, m_partShape.m_modelName);
 }
 
 ///////////////////////////////////////////////////////////////////
diff --git a/ngraph/test/frontend/shared/src/set_element_type.cpp b/ngraph/test/frontend/shared/src/set_element_type.cpp
index 5cfbdce990b4d9..9b29a308298b63 100644
--- a/ngraph/test/frontend/shared/src/set_element_type.cpp
+++ b/ngraph/test/frontend/shared/src/set_element_type.cpp
@@ -31,13 +31,8 @@ void FrontEndElementTypeTest::initParamTest()
 
 void FrontEndElementTypeTest::doLoadFromFile()
 {
-    std::vector<std::string> frontends;
-    FrontEnd::Ptr fe;
-    ASSERT_NO_THROW(frontends = m_fem.get_available_front_ends());
-    ASSERT_NO_THROW(m_frontEnd = m_fem.load_by_framework(m_param.m_frontEndName));
-    ASSERT_NE(m_frontEnd, nullptr);
-    ASSERT_NO_THROW(m_inputModel = m_frontEnd->load(m_param.m_modelName));
-    ASSERT_NE(m_inputModel, nullptr);
+    std::tie(m_frontEnd, m_inputModel) =
+        FrontEndTestUtils::load_from_file(m_fem, m_param.m_frontEndName, m_param.m_modelName);
 }
 
 ///////////////////////////////////////////////////////////////////