diff --git a/paddle/fluid/pir/dialect/operator/ir/ops_onednn_extra.yaml b/paddle/fluid/pir/dialect/operator/ir/ops_onednn_extra.yaml
index 290b20357824eb..835926a4c377f0 100644
--- a/paddle/fluid/pir/dialect/operator/ir/ops_onednn_extra.yaml
+++ b/paddle/fluid/pir/dialect/operator/ir/ops_onednn_extra.yaml
@@ -27,7 +27,8 @@
 
 # - op : bilinear_interp
 
-# - op : cast
+- op : cast
+  dynamic_fallback : True
 
 - op : clip
   extra_args : str mkldnn_data_type="float32"
diff --git a/paddle/phi/kernels/onednn/cast_kernel.cc b/paddle/phi/kernels/onednn/cast_kernel.cc
index 9bf0a3e8a875fa..fdee92c2d4f167 100644
--- a/paddle/phi/kernels/onednn/cast_kernel.cc
+++ b/paddle/phi/kernels/onednn/cast_kernel.cc
@@ -19,6 +19,16 @@
 
 namespace phi {
 
+bool CastCheckIfOneDNNSupport(const KernelContext* ctx) {
+  if ((ctx->InputAt<phi::DenseTensor>(0).dtype() != DataType::FLOAT32 &&
+       ctx->InputAt<phi::DenseTensor>(0).dtype() != DataType::BFLOAT16) ||
+      (ctx->AttrAt<DataType>(0) != DataType::FLOAT32 &&
+       ctx->AttrAt<DataType>(0) != DataType::BFLOAT16)) {
+    return false;
+  }
+  return true;
+}
+
 template <typename T, typename Context>
 void CastKernel(const Context& dev_ctx,
                 const DenseTensor& x,
@@ -56,4 +66,6 @@ void CastKernel(const Context& dev_ctx,
 }  // namespace phi
 
 PD_REGISTER_KERNEL(
-    cast, OneDNN, ONEDNN, phi::CastKernel, float, phi::dtype::bfloat16) {}
+    cast, OneDNN, ONEDNN, phi::CastKernel, float, phi::dtype::bfloat16) {
+  kernel->check_if_onednn_kernel_support_ = phi::CastCheckIfOneDNNSupport;
+}
diff --git a/test/mkldnn/test_cast_mkldnn_op.py b/test/mkldnn/test_cast_mkldnn_op.py
index 8856fd9071391b..a5a9562cf78130 100644
--- a/test/mkldnn/test_cast_mkldnn_op.py
+++ b/test/mkldnn/test_cast_mkldnn_op.py
@@ -47,7 +47,7 @@ def setUp(self):
         self.op_type = 'cast'
 
     def test_check_output(self):
-        self.check_output(check_dygraph=False)
+        self.check_output(check_dygraph=False, check_pir_onednn=True)
 
     def test_check_grad(self):
         self.check_grad_with_place(
@@ -57,6 +57,7 @@ def test_check_grad(self):
             check_dygraph=False,
             user_defined_grads=[self.inputs['X']],
             user_defined_grad_outputs=[self.outputs['Out']],
+            check_pir_onednn=True,
         )
 
     def init_shape(self):