diff --git a/deploy/cpp/demo/classifier.cpp b/deploy/cpp/demo/classifier.cpp
index f8b47cba7a6d2257c41fd0a64784a083c96544d9..548eaff411a737ea0ffcfca63d36a7f18cd9d994 100644
--- a/deploy/cpp/demo/classifier.cpp
+++ b/deploy/cpp/demo/classifier.cpp
@@ -61,9 +61,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
 
   // Predict
   int imgs = 1;
diff --git a/deploy/cpp/demo/detector.cpp b/deploy/cpp/demo/detector.cpp
index 02e4912c2459d590563963199dca0bac2ef3996b..f5fefc05d0bbc4bbd482c23f0db8c066b7d1013b 100644
--- a/deploy/cpp/demo/detector.cpp
+++ b/deploy/cpp/demo/detector.cpp
@@ -66,9 +66,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
   int imgs = 1;
   std::string save_dir = "output";
   // Predict
diff --git a/deploy/cpp/demo/segmenter.cpp b/deploy/cpp/demo/segmenter.cpp
index b38105e68ffbba79896e7d99375ea258f5d37c84..0d888001490759f65790d51837e2e69a6f448c4b 100644
--- a/deploy/cpp/demo/segmenter.cpp
+++ b/deploy/cpp/demo/segmenter.cpp
@@ -63,9 +63,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
   int imgs = 1;
   // Predict
   if (FLAGS_image_list != "") {
diff --git a/deploy/cpp/demo/video_classifier.cpp b/deploy/cpp/demo/video_classifier.cpp
index fd29dedb0849ed0b46de334eddcaa3a12f9411e3..c0485791ccb42fc880ab384ae2cf5e1d9d48b1ae 100644
--- a/deploy/cpp/demo/video_classifier.cpp
+++ b/deploy/cpp/demo/video_classifier.cpp
@@ -67,9 +67,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
 
   // Open video
   cv::VideoCapture capture;
diff --git a/deploy/cpp/demo/video_detector.cpp b/deploy/cpp/demo/video_detector.cpp
index fb8b4a78c87e27b33206572a06fa580d8c5e1521..e617dbd1339b73676225a65a667a42a06abfa63e 100644
--- a/deploy/cpp/demo/video_detector.cpp
+++ b/deploy/cpp/demo/video_detector.cpp
@@ -69,9 +69,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
   // Open video
   cv::VideoCapture capture;
   if (FLAGS_use_camera) {
diff --git a/deploy/cpp/demo/video_segmenter.cpp b/deploy/cpp/demo/video_segmenter.cpp
index 54e650e402a1059021e161acc5c07dd591712754..35af64f4b00ea5983653bb135394da9389539604 100644
--- a/deploy/cpp/demo/video_segmenter.cpp
+++ b/deploy/cpp/demo/video_segmenter.cpp
@@ -67,9 +67,9 @@ int main(int argc, char** argv) {
              FLAGS_use_gpu,
              FLAGS_use_trt,
              FLAGS_use_mkl,
+             FLAGS_mkl_thread_num,
              FLAGS_gpu_id,
-             FLAGS_key,
-             FLAGS_mkl_thread_num);
+             FLAGS_key);
   // Open video
   cv::VideoCapture capture;
   if (FLAGS_use_camera) {
diff --git a/deploy/cpp/include/paddlex/paddlex.h b/deploy/cpp/include/paddlex/paddlex.h
index 7a8e60021cd74b402561527d9b40c3eb72a6d672..3af3dd85e290dd18fec3d7bcfb68207a0cf2d5b5 100644
--- a/deploy/cpp/include/paddlex/paddlex.h
+++ b/deploy/cpp/include/paddlex/paddlex.h
@@ -70,6 +70,8 @@ class Model {
    * @param model_dir: the directory which contains model.yml
    * @param use_gpu: use gpu or not when infering
    * @param use_trt: use Tensor RT or not when infering
+   * @param use_trt: use mkl or not when infering
+   * @param mkl_thread_num: the threads of mkl when infering
    * @param gpu_id: the id of gpu when infering with using gpu
    * @param key: the key of encryption when using encrypted model
    * @param use_ir_optim: use ir optimization when infering
@@ -78,28 +80,27 @@ class Model {
             bool use_gpu = false,
             bool use_trt = false,
             bool use_mkl = true,
+            int mkl_thread_num = 4,
             int gpu_id = 0,
             std::string key = "",
-            int mkl_thread_num = 4,
             bool use_ir_optim = true) {
     create_predictor(
                      model_dir,
                      use_gpu,
                      use_trt,
                      use_mkl,
+                     mkl_thread_num,
                      gpu_id,
                      key,
-                     mkl_thread_num,
                      use_ir_optim);
   }
-
   void create_predictor(const std::string& model_dir,
                         bool use_gpu = false,
                         bool use_trt = false,
                         bool use_mkl = true,
+                        int mkl_thread_num = 4,
                         int gpu_id = 0,
                         std::string key = "",
-                        int mkl_thread_num = 4,
                         bool use_ir_optim = true);
 
   /*
diff --git a/deploy/cpp/src/paddlex.cpp b/deploy/cpp/src/paddlex.cpp
index 68038b6bc4bbe12de24c04dac7a186c69f42ea86..57d35b89f638173aa1fdc46600d059e16f183c14 100644
--- a/deploy/cpp/src/paddlex.cpp
+++ b/deploy/cpp/src/paddlex.cpp
@@ -29,9 +29,9 @@ void Model::create_predictor(const std::string& model_dir,
                              bool use_gpu,
                              bool use_trt,
                              bool use_mkl,
+                             int mkl_thread_num,
                              int gpu_id,
                              std::string key,
-                             int mkl_thread_num,
                              bool use_ir_optim) {
   paddle::AnalysisConfig config;
   std::string model_file = model_dir + OS_PATH_SEP + "__model__";