diff --git a/CMakeLists.txt b/CMakeLists.txt
index 4861e8065a8a70bcc3b859b121cf404461f47ba7..0ab80987b3ad6c4793ceeac1bf3808d2e87fbd5b 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -282,7 +282,3 @@ if(WITH_DOC)
     find_python_module(recommonmark REQUIRED)
     add_subdirectory(doc)
 endif()
-
-if (WITH_CONTRIB)
-    add_subdirectory(paddle/contrib)
-endif()
diff --git a/paddle/fluid/inference/CMakeLists.txt b/paddle/fluid/inference/CMakeLists.txt
index 86643b9aa111a9d73c184097e0c43bbec30c6b31..c9eff0fc28c5ff52f902d3d5a0ebb37fa7619e9c 100644
--- a/paddle/fluid/inference/CMakeLists.txt
+++ b/paddle/fluid/inference/CMakeLists.txt
@@ -5,7 +5,7 @@ if (TENSORRT_FOUND)
   add_subdirectory(tensorrt)
 endif()
 
-set(FLUID_CORE_MODULES proto_desc memory lod_tensor executor )
+set(FLUID_CORE_MODULES proto_desc memory lod_tensor executor)
 
 # TODO(panyx0718): Should this be called paddle_fluid_inference_api_internal?
 cc_library(paddle_fluid_api
@@ -38,3 +38,4 @@ if(WITH_TESTING)
   # both tests/book and analysis depends the models that generated by python/paddle/fluid/tests/book
   add_subdirectory(tests/book)
 endif()
+add_subdirectory(api)
diff --git a/paddle/fluid/inference/api/CMakeLists.txt b/paddle/fluid/inference/api/CMakeLists.txt
index df470a0752121e44f4305fd3609c5c41985dfaf7..9d63d08dedf6a1bcdacc51bb83d2ed261bca4117 100644
--- a/paddle/fluid/inference/api/CMakeLists.txt
+++ b/paddle/fluid/inference/api/CMakeLists.txt
@@ -43,21 +43,21 @@ function(inference_api_test TARGET_NAME)
 endfunction(inference_api_test)
 
 cc_library(paddle_inference_api
-    SRCS paddle_inference_api.cc paddle_inference_api_impl.cc
+    SRCS api.cc api_impl.cc
     DEPS ${FLUID_CORE_MODULES} ${GLOB_OP_LIB})
 if(NOT APPLE)
-  set(LINK_FLAGS "-Wl,--retain-symbols-file ${CMAKE_CURRENT_SOURCE_DIR}/paddle_inference_api.sym")
+  set(LINK_FLAGS "-Wl,--retain-symbols-file ${CMAKE_CURRENT_SOURCE_DIR}/api.sym")
   set_target_properties(paddle_inference_api PROPERTIES LINK_FLAGS "${LINK_FLAGS}")
 endif()
 
 # Here the shared library doesn't depend on other fluid libraries, or double free will occur.
 cc_library(paddle_inference_api_shared SHARED
-    SRCS paddle_inference_api.cc paddle_inference_api_impl.cc)
+    SRCS api.cc api_impl.cc)
 add_dependencies(paddle_inference_api_shared ${FLUID_CORE_MODULES} ${GLOB_OP_LIB})
 set_target_properties(paddle_inference_api_shared PROPERTIES OUTPUT_NAME paddle_inference_api)
 
 if(NOT APPLE)
-  set(LINK_FLAGS "-Wl,--version-script ${CMAKE_CURRENT_SOURCE_DIR}/paddle_inference_api.map")
+  set(LINK_FLAGS "-Wl,--version-script ${CMAKE_CURRENT_SOURCE_DIR}/api.map")
   set_target_properties(paddle_inference_api_shared PROPERTIES LINK_FLAGS "${LINK_FLAGS}")
   FILE(WRITE ${CMAKE_CURRENT_BINARY_DIR}/check_symbol.cmake
     "execute_process(COMMAND bash -c \"${CMAKE_CURRENT_SOURCE_DIR}/check_symbol.sh"
@@ -73,32 +73,32 @@ if(NOT APPLE)
 endif()
 
 cc_test(test_paddle_inference_api
-        SRCS test_paddle_inference_api.cc
+        SRCS test_api.cc
         DEPS paddle_inference_api)
 
-inference_api_test(test_paddle_inference_api_impl
+inference_api_test(test_api_impl
                     ARGS test_word2vec test_image_classification)
 
 if(WITH_GPU AND TENSORRT_FOUND)
 cc_library(paddle_inference_tensorrt_subgraph_engine
-        SRCS paddle_inference_api_tensorrt_subgraph_engine.cc
-        DEPS paddle_inference_api analysis tensorrt_engine paddle_inference_api paddle_fluid_api)
+        SRCS api_tensorrt_subgraph_engine.cc
+        DEPS paddle_inference_api analysis tensorrt_engine paddle_fluid_api)
 
-inference_api_test(test_paddle_inference_api_tensorrt_subgraph_engine ARGS test_word2vec)
+inference_api_test(test_api_tensorrt_subgraph_engine ARGS test_word2vec)
 endif()
 
 if (WITH_ANAKIN) # only needed in CI
     # Due to Anakin do not have official library releases and the versions of protobuf and cuda do not match Paddle's,
     # so anakin library will not be merged to our official inference library. To use anakin prediction API, one need to
     # compile the libinference_anakin_api.a and compile with anakin.so.
-    nv_library(inference_anakin_api SRCS paddle_inference_api.cc paddle_inference_api_anakin_engine.cc)
-    nv_library(inference_anakin_api_shared SHARED SRCS paddle_inference_api.cc paddle_inference_api_anakin_engine.cc)
+    nv_library(inference_anakin_api SRCS api.cc api_anakin_engine.cc)
+    nv_library(inference_anakin_api_shared SHARED SRCS api.cc api_anakin_engine.cc)
     target_compile_options(inference_anakin_api BEFORE PUBLIC ${ANAKIN_COMPILE_EXTRA_FLAGS})
     target_compile_options(inference_anakin_api_shared BEFORE PUBLIC ${ANAKIN_COMPILE_EXTRA_FLAGS})
     target_link_libraries(inference_anakin_api anakin anakin_saber_common)
     target_link_libraries(inference_anakin_api_shared anakin anakin_saber_common)
     if (WITH_TESTING)
-        cc_test(inference_anakin_test SRCS paddle_inference_api_anakin_engine_tester.cc
+        cc_test(inference_anakin_test SRCS api_anakin_engine_tester.cc
                                   ARGS --model=${ANAKIN_INSTALL_DIR}/mobilenet_v2.anakin.bin
                                   DEPS inference_anakin_api)
         target_compile_options(inference_anakin_test BEFORE PUBLIC ${ANAKIN_COMPILE_EXTRA_FLAGS})
diff --git a/paddle/fluid/inference/api/api.cc b/paddle/fluid/inference/api/api.cc
index 4fe198ad7d4a752882965e9e7fc460741de53d22..c6b228b3ef5e0808a51be9eb85fcc17bed10e404 100644
--- a/paddle/fluid/inference/api/api.cc
+++ b/paddle/fluid/inference/api/api.cc
@@ -12,7 +12,7 @@ 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 "paddle/contrib/inference/paddle_inference_api.h"
+#include "paddle/fluid/inference/api/api.h"
 
 namespace paddle {
 
diff --git a/paddle/fluid/inference/api/api_anakin_engine.cc b/paddle/fluid/inference/api/api_anakin_engine.cc
index 5d9208bd8a29d22debe3726fa602a708130bceda..f6f3cb335897b02905e24c229b92f3940a37dbf8 100644
--- a/paddle/fluid/inference/api/api_anakin_engine.cc
+++ b/paddle/fluid/inference/api/api_anakin_engine.cc
@@ -12,8 +12,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#include "paddle/fluid/inference/api/api_anakin_engine.h"
 #include <cuda.h>
-#include "paddle/contrib/inference/paddle_inference_api_anakin_engine.h"
+#include <vector>
 
 namespace paddle {
 
@@ -110,6 +111,6 @@ std::unique_ptr<PaddlePredictor> CreatePaddlePredictor<
   std::unique_ptr<PaddlePredictor> x(
       new PaddleInferenceAnakinPredictor(config));
   return x;
-};
+}
 
 }  // namespace paddle
diff --git a/paddle/fluid/inference/api/api_anakin_engine.h b/paddle/fluid/inference/api/api_anakin_engine.h
index 159699b4f80d9a9257ad350c3863ec252ce4d97f..0b7de8877ae12d7a081e0b02668b6e047e1a5ba6 100644
--- a/paddle/fluid/inference/api/api_anakin_engine.h
+++ b/paddle/fluid/inference/api/api_anakin_engine.h
@@ -19,7 +19,8 @@ limitations under the License. */
 
 #pragma once
 
-#include "paddle/contrib/inference/paddle_inference_api.h"
+#include <vector>
+#include "paddle/fluid/inference/api/api.h"
 
 // from anakin
 #include "framework/core/net/net.h"
@@ -31,7 +32,7 @@ class PaddleInferenceAnakinPredictor : public PaddlePredictor {
  public:
   PaddleInferenceAnakinPredictor() {}
 
-  PaddleInferenceAnakinPredictor(const AnakinConfig& config);
+  explicit PaddleInferenceAnakinPredictor(const AnakinConfig& config);
 
   // NOTE Unlike the native engine, the buffers of anakin engine's output_data
   // should be allocated first.
diff --git a/paddle/fluid/inference/api/api_anakin_engine_tester.cc b/paddle/fluid/inference/api/api_anakin_engine_tester.cc
index f92e9d4190412f5847e353ef1dc0324cad668c9a..94c828a9d874c46131b7d9933f2136f4b9c5d48a 100644
--- a/paddle/fluid/inference/api/api_anakin_engine_tester.cc
+++ b/paddle/fluid/inference/api/api_anakin_engine_tester.cc
@@ -16,7 +16,7 @@ limitations under the License. */
 #include <glog/logging.h>
 #include <gtest/gtest.h>
 
-#include "paddle/contrib/inference/paddle_inference_api.h"
+#include "paddle/fluid/inference/api/api.h"
 
 DEFINE_string(model, "", "Directory of the inference model.");
 
diff --git a/paddle/fluid/inference/api/api_impl.cc b/paddle/fluid/inference/api/api_impl.cc
index b65f82b1c5c1531d70699c2eb39c989b89e6e971..9d9e126e134deafc18f10f7299e81d92702e9ca0 100644
--- a/paddle/fluid/inference/api/api_impl.cc
+++ b/paddle/fluid/inference/api/api_impl.cc
@@ -21,7 +21,7 @@ limitations under the License. */
 #include <utility>
 #include <vector>
 
-#include "paddle/contrib/inference/paddle_inference_api_impl.h"
+#include "paddle/fluid/inference/api/api_impl.h"
 
 namespace paddle {
 namespace {
@@ -105,7 +105,7 @@ NativePaddlePredictor::~NativePaddlePredictor() {
     PADDLE_ENFORCE_NOT_NULL(scope_, "Should have parent scope!");
     scope_->DeleteScope(sub_scope_);
   }
-};
+}
 
 bool NativePaddlePredictor::Run(const std::vector<PaddleTensor> &inputs,
                                 std::vector<PaddleTensor> *output_data) {
diff --git a/paddle/fluid/inference/api/api_impl.h b/paddle/fluid/inference/api/api_impl.h
index f9ec6f55449fc46b4a44b9563980cb5f8e80a951..5dbcb26d21b675740e80f00f1feb2642d11f3a3a 100644
--- a/paddle/fluid/inference/api/api_impl.h
+++ b/paddle/fluid/inference/api/api_impl.h
@@ -19,7 +19,7 @@
 #include <string>
 #include <vector>
 
-#include "paddle/contrib/inference/paddle_inference_api.h"
+#include "paddle/fluid/inference/api/api.h"
 
 #include "paddle/fluid/framework/ddim.h"
 #include "paddle/fluid/framework/lod_tensor.h"
diff --git a/paddle/fluid/inference/api/api_tensorrt_subgraph_engine.cc b/paddle/fluid/inference/api/api_tensorrt_subgraph_engine.cc
index 8ed8c1551f65c5b70f2976958430e6776ba3d941..6d5de1a98bb61801725fd1b7b5f7e965d8e225de 100644
--- a/paddle/fluid/inference/api/api_tensorrt_subgraph_engine.cc
+++ b/paddle/fluid/inference/api/api_tensorrt_subgraph_engine.cc
@@ -12,9 +12,9 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "paddle/contrib/inference/paddle_inference_api.h"
-#include "paddle/contrib/inference/paddle_inference_api_impl.h"
+#include "paddle/fluid/inference/api/api.h"
 #include "paddle/fluid/inference/analysis/analyzer.h"
+#include "paddle/fluid/inference/api/api_impl.h"
 #include "paddle/fluid/inference/utils/singleton.h"
 
 namespace paddle {
diff --git a/paddle/fluid/inference/api/test_api.cc b/paddle/fluid/inference/api/test_api.cc
index bc7faab6e208a66d7a56e41a56bd743c7644eea2..03b5e8dacdca35ad4c237f89b112c52ba25fb335 100644
--- a/paddle/fluid/inference/api/test_api.cc
+++ b/paddle/fluid/inference/api/test_api.cc
@@ -12,10 +12,9 @@ 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 "paddle/contrib/inference/paddle_inference_api.h"
-
 #include <glog/logging.h>
 #include <gtest/gtest.h>
+#include "paddle/fluid/inference/api/api.h"
 
 namespace paddle {
 
diff --git a/paddle/fluid/inference/api/test_api_impl.cc b/paddle/fluid/inference/api/test_api_impl.cc
index 5eef53c0f1901f78023aa9641f85f6eb768c4383..fc1364b80ac1ee2d304eb2fe429eae5f56967516 100644
--- a/paddle/fluid/inference/api/test_api_impl.cc
+++ b/paddle/fluid/inference/api/test_api_impl.cc
@@ -15,10 +15,10 @@ limitations under the License. */
 #include <glog/logging.h>
 #include <gtest/gtest.h>
 
-#include <thread>
+#include <thread>  // NOLINT
 
 #include "gflags/gflags.h"
-#include "paddle/contrib/inference/paddle_inference_api_impl.h"
+#include "paddle/fluid/inference/api/api_impl.h"
 #include "paddle/fluid/inference/tests/test_helper.h"
 
 DEFINE_string(dirname, "", "Directory of the inference model.");
diff --git a/paddle/fluid/inference/api/test_api_tensorrt_subgraph_engine.cc b/paddle/fluid/inference/api/test_api_tensorrt_subgraph_engine.cc
index b100630dbe412ca811f1a8f2b8191356f5ebec2f..2e26781a7bce111830c99c2cd44c93a596e253a2 100644
--- a/paddle/fluid/inference/api/test_api_tensorrt_subgraph_engine.cc
+++ b/paddle/fluid/inference/api/test_api_tensorrt_subgraph_engine.cc
@@ -15,7 +15,7 @@
 #include <gflags/gflags.h>
 #include <glog/logging.h>
 #include <gtest/gtest.h>
-#include "paddle/contrib/inference/paddle_inference_api.h"
+#include "paddle/fluid/inference/api/api.h"
 
 namespace paddle {
 
@@ -61,4 +61,4 @@ void Main(bool use_gpu) {
 
 TEST(paddle_inference_api_tensorrt_subgraph_engine, main) { Main(true); }
 
-}  // namespace paddle
\ No newline at end of file
+}  // namespace paddle