diff --git a/paddle/fluid/pybind/op_function_generator.cc b/paddle/fluid/pybind/op_function_generator.cc
index da30a12c1849dc0c8f7a1bdc782c7a1264cda2e7..62f1de35ae602b7c800a88de323e22d812bfeb4f 100644
--- a/paddle/fluid/pybind/op_function_generator.cc
+++ b/paddle/fluid/pybind/op_function_generator.cc
@@ -37,7 +37,8 @@ std::map<std::string, std::set<std::string>> op_passing_out_map = {
     {"momentum", {"ParamOut", "VelocityOut"}},
     {"batch_norm", {"MeanOut", "VarianceOut"}},
     {"accuracy", {"Correct", "Total"}},
-    {"fill_constant", {"Out"}}};
+    {"fill_constant", {"Out"}},
+    {"matmul", {"Out"}}};
 
 // clang-format off
 const char* OUT_INITIALIZER_TEMPLATE =
diff --git a/python/paddle/fluid/dygraph/nn.py b/python/paddle/fluid/dygraph/nn.py
index 6b617f04681c6be0ea1f7fadfd9cb2306456d15e..8f6b0a7d5a30ce84daf21fc8122f7e7439349a8f 100644
--- a/python/paddle/fluid/dygraph/nn.py
+++ b/python/paddle/fluid/dygraph/nn.py
@@ -936,8 +936,9 @@ class Linear(layers.Layer):
 
     def forward(self, input):
         if in_dygraph_mode():
-            pre_bias = core.ops.matmul(input, self.weight, 'transpose_X', False,
-                                       'transpose_Y', False, "alpha", 1)
+            pre_bias = _varbase_creator(dtype=input.dtype)
+            core.ops.matmul(input, self.weight, pre_bias, 'transpose_X', False,
+                            'transpose_Y', False, "alpha", 1)
             pre_act = dygraph_utils._append_bias_in_dygraph(
                 pre_bias, self.bias, axis=len(input.shape) - 1)
 
diff --git a/python/paddle/fluid/tests/unittests/test_matmul_op.py b/python/paddle/fluid/tests/unittests/test_matmul_op.py
index 76199abb310a0f08db9d9c46107b4db180cd5e88..018b5c62862f075c4ce4243bf3c008bddcf416be 100644
--- a/python/paddle/fluid/tests/unittests/test_matmul_op.py
+++ b/python/paddle/fluid/tests/unittests/test_matmul_op.py
@@ -280,6 +280,42 @@ class API_TestMm(unittest.TestCase):
             "two value is\
             {}\n{}, check diff!".format(np_res, expected_result))
 
+    def test_dygraph_with_out(self):
+        device = fluid.CPUPlace()
+        with fluid.dygraph.guard(device):
+            input_array1 = np.random.rand(3, 4).astype("float64")
+            input_array2 = np.random.rand(4, 3).astype("float64")
+            out_array = np.random.rand(3, 3).astype("float64")
+            data1 = fluid.dygraph.to_variable(input_array1)
+            data2 = fluid.dygraph.to_variable(input_array2)
+            paddle_out_holder = fluid.dygraph.to_variable(out_array)
+            out = paddle.mm(data1, data2, out=paddle_out_holder)
+        self.assertTrue(np.allclose(paddle_out_holder.numpy(), out.numpy()))
+
+    def test_dygraph_without_out(self):
+        device = fluid.CPUPlace()
+        with fluid.dygraph.guard(device):
+            input_array1 = np.random.rand(3, 4).astype("float64")
+            input_array2 = np.random.rand(4, 3).astype("float64")
+            data1 = fluid.dygraph.to_variable(input_array1)
+            data2 = fluid.dygraph.to_variable(input_array2)
+            out = paddle.mm(data1, data2)
+            expected_result = np.matmul(input_array1, input_array2)
+        self.assertTrue(np.allclose(expected_result, out.numpy()))
+
+
+class Test_API_Matmul(unittest.TestCase):
+    def test_dygraph_without_out(self):
+        device = fluid.CPUPlace()
+        with fluid.dygraph.guard(device):
+            input_array1 = np.random.rand(3, 4).astype("float64")
+            input_array2 = np.random.rand(4, 3).astype("float64")
+            data1 = fluid.dygraph.to_variable(input_array1)
+            data2 = fluid.dygraph.to_variable(input_array2)
+            out = paddle.matmul(data1, data2)
+            expected_result = np.matmul(input_array1, input_array2)
+        self.assertTrue(np.allclose(expected_result, out.numpy()))
+
 
 class API_TestMmError(unittest.TestCase):
     def test_errors(self):
diff --git a/python/paddle/tensor/linalg.py b/python/paddle/tensor/linalg.py
index 05ab142a902d11741abb2d030492aef10df6979c..aedc88db156e34ee0a444cd9f950e78b3ba82125 100644
--- a/python/paddle/tensor/linalg.py
+++ b/python/paddle/tensor/linalg.py
@@ -14,7 +14,7 @@
 from paddle.common_ops_import import *
 from ..fluid.layer_helper import LayerHelper
 from ..fluid.data_feeder import check_variable_and_dtype, check_type
-from ..fluid.framework import in_dygraph_mode
+from ..fluid.framework import in_dygraph_mode, _varbase_creator
 
 __all__ = [
     'matmul',
@@ -109,8 +109,10 @@ def matmul(x, y, transpose_x=False, transpose_y=False, alpha=1.0, name=None):
     }
 
     if in_dygraph_mode():
-        return core.ops.matmul(x, y, 'transpose_X', transpose_x, 'transpose_Y',
-                               transpose_y, 'alpha', float(alpha))
+        out = _varbase_creator(dtype=x.dtype)
+        core.ops.matmul(x, y, out, 'transpose_X', transpose_x, 'transpose_Y',
+                        transpose_y, 'alpha', float(alpha))
+        return out
 
     def __check_input(x, y):
         var_names = {'x': x, 'y': y}
diff --git a/python/paddle/tensor/math.py b/python/paddle/tensor/math.py
index 4c9bb0f0a8273ea147152c3a99b3c6bc2321a2dc..ac9a56a90fef1a1facda4980ce6d8230315da4f8 100644
--- a/python/paddle/tensor/math.py
+++ b/python/paddle/tensor/math.py
@@ -19,7 +19,7 @@ from __future__ import print_function
 
 from paddle.common_ops_import import *
 from ..fluid import layers
-from ..fluid.framework import core
+from ..fluid.framework import core, _varbase_creator
 from ..fluid.layers.layer_function_generator import _generate_doc_string_
 
 # TODO: define math functions
@@ -902,7 +902,10 @@ def mm(input, mat2, out=None, name=None):
             out = paddle.mm(x, mat2) # out shape is [2, 2]
     """
     if in_dygraph_mode():
-        return core.ops.matmul(input, mat2)
+        if out is None:
+            out = _varbase_creator(dtype=input.dtype)
+        core.ops.matmul(input, mat2, out)
+        return out
 
     def __check_input(x, y):
         var_names = {'x': x, 'y': y}