diff --git a/onnxruntime/core/providers/cpu/cpu_execution_provider.cc b/onnxruntime/core/providers/cpu/cpu_execution_provider.cc
index 625b6a9c40..496d96bba7 100644
--- a/onnxruntime/core/providers/cpu/cpu_execution_provider.cc
+++ b/onnxruntime/core/providers/cpu/cpu_execution_provider.cc
@@ -270,6 +270,7 @@ class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain,
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, float, Where);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, int32_t, Where);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, int64_t, Where);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, uint8_t, Where);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, 10, Flatten);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, BatchNormalization);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, 10, Gemm);
@@ -890,6 +891,8 @@ Status RegisterOnnxOperatorKernels(KernelRegistry& kernel_registry) {
                                                                   Where)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, int64_t,
                                                                   Where)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, uint8_t,
+                                                                  Where)>,                                                                  
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9, 10,
                                                                       Flatten)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCpuExecutionProvider, kOnnxDomain, 9,
diff --git a/onnxruntime/core/providers/cpu/tensor/where_op.cc b/onnxruntime/core/providers/cpu/tensor/where_op.cc
index d518352af8..7585943c00 100644
--- a/onnxruntime/core/providers/cpu/tensor/where_op.cc
+++ b/onnxruntime/core/providers/cpu/tensor/where_op.cc
@@ -22,7 +22,7 @@ namespace onnxruntime {
   WHERE_TYPED_KERNEL_WITH_TYPE_NAME(type, type)
 
 // start with a subset of types, enable more as needed...
-//WHERE_TYPED_KERNEL(uint8_t)
+WHERE_TYPED_KERNEL(uint8_t)
 //WHERE_TYPED_KERNEL(uint16_t)
 //WHERE_TYPED_KERNEL(uint32_t)
 //WHERE_TYPED_KERNEL(uint64_t)
diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index 375d714612..92f7ba87e6 100644
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -539,6 +539,7 @@ class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain,
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, float, Where);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, Where);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int64_t, Where);
+class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, uint8_t, Where);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, bool, NonZero);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, uint8_t, NonZero);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, NonZero);
@@ -1004,6 +1005,7 @@ static void RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, float, Where)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, Where)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int64_t, Where)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, uint8_t, Where)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, bool, NonZero)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, uint8_t, NonZero)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, NonZero)>,
diff --git a/onnxruntime/core/providers/cuda/tensor/where.cc b/onnxruntime/core/providers/cuda/tensor/where.cc
index 142dc92944..1e341f3f07 100644
--- a/onnxruntime/core/providers/cuda/tensor/where.cc
+++ b/onnxruntime/core/providers/cuda/tensor/where.cc
@@ -183,6 +183,7 @@ Status Where<T>::ComputeInternal(OpKernelContext* context) const {
 #define SPECIALIZED_COMPUTE(T) \
   SPECIALIZED_COMPUTE_WITH_NAME(T, T)
 
+SPECIALIZED_COMPUTE(uint8_t)
 SPECIALIZED_COMPUTE(int32_t)
 SPECIALIZED_COMPUTE(int64_t)
 SPECIALIZED_COMPUTE(float)
diff --git a/onnxruntime/core/providers/cuda/tensor/where_impl.cu b/onnxruntime/core/providers/cuda/tensor/where_impl.cu
index 7bde79a276..6c7f92dcf7 100644
--- a/onnxruntime/core/providers/cuda/tensor/where_impl.cu
+++ b/onnxruntime/core/providers/cuda/tensor/where_impl.cu
@@ -120,6 +120,7 @@ void WhereImpl(
                              T* output_data,                         \
                              size_t count);
 
+SPECIALIZED_IMPL(uint8_t)
 SPECIALIZED_IMPL(int32_t)
 SPECIALIZED_IMPL(int64_t)
 SPECIALIZED_IMPL(float)