diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index 2a417d55d9..74ac19cb43 100644
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -328,7 +328,7 @@ class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDom
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Unsqueeze);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 8, Flatten);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Squeeze);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, Identity);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 12, Identity);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, Dropout);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Gather);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 8, float, Gemm);
@@ -1006,6 +1006,7 @@ class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, F
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, float, LRN);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, double, LRN);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, MLFloat16, LRN);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Identity);
 
 template <>
 KernelCreateInfo BuildKernelCreateInfo<void>() {
@@ -1022,7 +1023,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Unsqueeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 8, Flatten)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Squeeze)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 12, Identity)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 9, Dropout)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 1, 10, Gather)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 7, 8, float, Gemm)>,
@@ -1700,6 +1701,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, float, LRN)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, double, LRN)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, MLFloat16, LRN)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Identity)>,
   };
 
   for (auto& function_table_entry : function_table) {
diff --git a/onnxruntime/core/providers/cuda/tensor/identity_op.cc b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
index 1870e624cf..6b125aaf70 100644
--- a/onnxruntime/core/providers/cuda/tensor/identity_op.cc
+++ b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
@@ -11,8 +11,8 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
     7, 9,
     kCudaExecutionProvider,
     KernelDefBuilder()
-        .TypeConstraint("T", {DataTypeImpl::GetTensorType<MLFloat16>(), 
-                              DataTypeImpl::GetTensorType<float>(), 
+        .TypeConstraint("T", {DataTypeImpl::GetTensorType<MLFloat16>(),
+                              DataTypeImpl::GetTensorType<float>(),
                               DataTypeImpl::GetTensorType<double>()})
         .Alias(0, 0),
     IdentityOp<true>);
@@ -31,10 +31,20 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .Alias(0, 0),
     IdentityOp<true>);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Identity,
+    kOnnxDomain,
+    1, 12,
+     kCudaExecutionProvider,
+    KernelDefBuilder()
+        .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes())
+        .Alias(0, 0),
+    IdentityOp<false>);
+
 ONNX_OPERATOR_KERNEL_EX(
     Identity,
     kOnnxDomain,
-    1,
+    13,
     kCudaExecutionProvider,
     KernelDefBuilder()
         .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes())