From 1c8d0eb21961465a615e196c808f6bba9a9f9ed1 Mon Sep 17 00:00:00 2001
From: Rishi-Dave <rishipdave@gmail.com>
Date: Wed, 18 Mar 2026 08:33:33 -0700
Subject: [PATCH 1/3] Add opset 21/23 CUDA kernel registrations for Flatten,
 Identity, If, Loop, Scan, ConstantOfShape, Size (#27102)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

When ONNX introduces a new version of an operator in opset 21, the kernel
registry's VerifyVersion rejects non-versioned (open-ended) CUDA kernels
because kernel_start_version != since_ver while kernel_end_version == INT_MAX.
This causes those operators to fall back from CUDA to CPU, introducing
unnecessary host↔device copies that can lead to value corruption on Windows.

PR #26075 previously fixed this for Shape, Reshape, Transpose, Squeeze, and
Unsqueeze. This commit extends the same fix to the remaining affected
operators: Flatten, Identity, If, Loop, Scan, ConstantOfShape, and Size.

For each operator:
- Cap existing non-versioned kernel to opset 20 (VERSIONED)
- Add VERSIONED(21, 22) kernel with identical type constraints
- Add non-versioned opset 23 kernel for forward compatibility
---
 .../core/providers/cuda/controlflow/if.cc     | 22 +++++++-
 .../core/providers/cuda/controlflow/loop.cc   | 26 ++++++++-
 .../core/providers/cuda/controlflow/scan.cc   | 22 +++++++-
 .../providers/cuda/cuda_execution_provider.cc | 56 ++++++++++++++-----
 .../cuda/generator/constant_of_shape.cc       | 24 +++++++-
 .../core/providers/cuda/tensor/flatten.cc     | 22 +++++++-
 .../core/providers/cuda/tensor/identity_op.cc | 22 +++++++-
 .../core/providers/cuda/tensor/size.cc        | 26 ++++++++-
 8 files changed, 199 insertions(+), 21 deletions(-)

diff --git a/onnxruntime/core/providers/cuda/controlflow/if.cc b/onnxruntime/core/providers/cuda/controlflow/if.cc
index 74cf79c619940..83c0c54f76309 100644
--- a/onnxruntime/core/providers/cuda/controlflow/if.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/if.cc
@@ -46,9 +46,29 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(If,
                                   If);
 
 // opset-19 supports float8
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(If,
+                                  kOnnxDomain,
+                                  19, 20,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  If);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(If,
+                                  kOnnxDomain,
+                                  21, 22,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  If);
+
 ONNX_OPERATOR_KERNEL_EX(If,
                         kOnnxDomain,
-                        19,
+                        23,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'cond' needs to be on CPU
diff --git a/onnxruntime/core/providers/cuda/controlflow/loop.cc b/onnxruntime/core/providers/cuda/controlflow/loop.cc
index d66de7c74e647..8d5214813527a 100644
--- a/onnxruntime/core/providers/cuda/controlflow/loop.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/loop.cc
@@ -52,9 +52,33 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(Loop,
                                   Loop);
 
 // opset-19 supports float 8 types.
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Loop,
+                                  kOnnxDomain,
+                                  19, 20,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'M' needs to be on CPU
+                                      .InputMemoryType(OrtMemTypeCPUInput, 1)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  Loop);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Loop,
+                                  kOnnxDomain,
+                                  21, 22,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'M' needs to be on CPU
+                                      .InputMemoryType(OrtMemTypeCPUInput, 1)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  Loop);
+
 ONNX_OPERATOR_KERNEL_EX(Loop,
                         kOnnxDomain,
-                        19,
+                        23,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'M' needs to be on CPU
diff --git a/onnxruntime/core/providers/cuda/controlflow/scan.cc b/onnxruntime/core/providers/cuda/controlflow/scan.cc
index ad715325e14f0..50a5b5267ff3b 100644
--- a/onnxruntime/core/providers/cuda/controlflow/scan.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/scan.cc
@@ -112,9 +112,29 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(Scan,
                                   Scan<9>);
 
 // Opset 19 starts to support float 8 types for the type constraint "V"
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Scan,
+                                  kOnnxDomain,
+                                  19, 20,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      // 'I' is in the ONNX spec but is not used for any inputs or outputs
+                                      // .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorTypesIRv9()),
+                                  Scan<9>);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Scan,
+                                  kOnnxDomain,
+                                  21, 22,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      // 'I' is in the ONNX spec but is not used for any inputs or outputs
+                                      // .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorTypesIRv9()),
+                                  Scan<9>);
+
 ONNX_OPERATOR_KERNEL_EX(Scan,
                         kOnnxDomain,
-                        19,
+                        23,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             // 'I' is in the ONNX spec but is not used for any inputs or outputs
diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index c958d6d700411..7a0599ea1b608 100755
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -881,7 +881,7 @@ class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kO
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 9, int32_t, Upsample);
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 9, uint8_t, Upsample);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 2, 10, Split);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, ConstantOfShape);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 20, ConstantOfShape);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int8_t, Shrink);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int16_t, Shrink);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, Shrink);
@@ -1180,7 +1180,7 @@ class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kO
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 18, bool, Cast);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 13, Reshape);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 14, Shape);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Size);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Size);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Transpose);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 15, ScatterElements);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, int32_t, Slice);
@@ -1252,7 +1252,7 @@ class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kO
 class ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 17, uint8_t, Resize);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 18, If);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 18, Loop);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Flatten);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Flatten);
 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);
@@ -1486,9 +1486,9 @@ class ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider
 class ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Float8E5M2, MLFloat16, DequantizeLinear);
 #endif
 
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Identity);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, If);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Loop);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Identity);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, If);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Loop);
 class ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, uint8_t, float, QuantizeLinear);
 class ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, int8_t, float, QuantizeLinear);
 #if !defined(DISABLE_FLOAT8_TYPES)
@@ -1507,7 +1507,7 @@ class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain,
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, MLFloat16, Resize);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, int32_t, Resize);
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, uint8_t, Resize);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Scan);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Scan);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Shape);
 
 // Opset 20
@@ -1573,6 +1573,13 @@ class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDom
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Transpose);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Squeeze);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Unsqueeze);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, ConstantOfShape);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Identity);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, If);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Loop);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Scan);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Size);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Flatten);
 
 // Opset 22.
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 22, float, AveragePool);
@@ -1634,6 +1641,13 @@ class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, R
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Transpose);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Squeeze);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Unsqueeze);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, ConstantOfShape);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Identity);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, If);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Loop);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Scan);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Size);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Flatten);
 
 #if !defined(DISABLE_FLOAT4_TYPES)
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Float4E2M1x2, Cast);
@@ -1977,7 +1991,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 9, int32_t, Upsample)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 9, uint8_t, Upsample)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 2, 10, Split)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, ConstantOfShape)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, 20, ConstantOfShape)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int8_t, Shrink)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int16_t, Shrink)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 9, int32_t, Shrink)>,
@@ -2277,7 +2291,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 19, IsNaN)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 13, Reshape)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 14, Shape)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Size)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Size)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Transpose)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 15, ScatterElements)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, int32_t, Slice)>,
@@ -2348,7 +2362,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 17, uint8_t, Resize)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 18, If)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 18, Loop)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, Flatten)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 13, 20, Flatten)>,
       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)>,
@@ -2581,9 +2595,9 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Float8E5M2, MLFloat16, DequantizeLinear)>,
 #endif
 
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Identity)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, If)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Loop)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, If)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Loop)>,
 
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, uint8_t, float, QuantizeLinear)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_TWO_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, int8_t, float, QuantizeLinear)>,
@@ -2604,7 +2618,7 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, MLFloat16, Resize)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, int32_t, Resize)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, uint8_t, Resize)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, Scan)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Scan)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 19, 20, Shape)>,
 
       // Opset 20
@@ -2669,6 +2683,13 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Transpose)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Squeeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Unsqueeze)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, ConstantOfShape)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, If)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Loop)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Scan)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Size)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 21, 22, Flatten)>,
 
       // Opset 22
       BuildKernelCreateInfo<ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 22, float, AveragePool)>,
@@ -2733,6 +2754,13 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Transpose)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Squeeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Unsqueeze)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, ConstantOfShape)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, If)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Loop)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Scan)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Size)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Flatten)>,
 
       // Opset 24
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 24, TensorScatter)>,
diff --git a/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc b/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
index 1a7f2422a8ca6..270bee866bcdf 100644
--- a/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
+++ b/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
@@ -8,10 +8,32 @@ using namespace ONNX_NAMESPACE;
 namespace onnxruntime {
 namespace cuda {
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    ConstantOfShape,
+    kOnnxDomain,
+    9, 20,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .InputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>())
+        .TypeConstraint("T2", DataTypeImpl::AllFixedSizeTensorTypes()),
+    ConstantOfShape);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    ConstantOfShape,
+    kOnnxDomain,
+    21, 22,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .InputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>())
+        .TypeConstraint("T2", DataTypeImpl::AllFixedSizeTensorTypes()),
+    ConstantOfShape);
+
 ONNX_OPERATOR_KERNEL_EX(
     ConstantOfShape,
     kOnnxDomain,
-    9,
+    23,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .InputMemoryType(OrtMemTypeCPUInput, 0)
diff --git a/onnxruntime/core/providers/cuda/tensor/flatten.cc b/onnxruntime/core/providers/cuda/tensor/flatten.cc
index e250d711bdfcf..7568a132b409c 100644
--- a/onnxruntime/core/providers/cuda/tensor/flatten.cc
+++ b/onnxruntime/core/providers/cuda/tensor/flatten.cc
@@ -37,10 +37,30 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes()),
     Flatten);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Flatten,
+    kOnnxDomain,
+    13, 20,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .Alias(0, 0)
+        .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes()),
+    Flatten);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Flatten,
+    kOnnxDomain,
+    21, 22,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .Alias(0, 0)
+        .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes()),
+    Flatten);
+
 ONNX_OPERATOR_KERNEL_EX(
     Flatten,
     kOnnxDomain,
-    13,
+    23,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .Alias(0, 0)
diff --git a/onnxruntime/core/providers/cuda/tensor/identity_op.cc b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
index 18c02c41777a3..07b2e2af13f20 100644
--- a/onnxruntime/core/providers/cuda/tensor/identity_op.cc
+++ b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
@@ -61,10 +61,30 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .Alias(0, 0),
     IdentityOp<false>);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Identity,
+    kOnnxDomain,
+    19, 20,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorAndSequenceTensorTypesIRv9())
+        .Alias(0, 0),
+    IdentityOp<false>);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Identity,
+    kOnnxDomain,
+    21, 22,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorAndSequenceTensorTypesIRv9())
+        .Alias(0, 0),
+    IdentityOp<false>);
+
 ONNX_OPERATOR_KERNEL_EX(
     Identity,
     kOnnxDomain,
-    19,
+    23,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorAndSequenceTensorTypesIRv9())
diff --git a/onnxruntime/core/providers/cuda/tensor/size.cc b/onnxruntime/core/providers/cuda/tensor/size.cc
index 48ca32a54872e..908b51bab58ae 100644
--- a/onnxruntime/core/providers/cuda/tensor/size.cc
+++ b/onnxruntime/core/providers/cuda/tensor/size.cc
@@ -19,10 +19,34 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>()),
     Size);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Size,
+    kOnnxDomain,
+    13, 20,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        // properly force CPU/GPU synch inside the kernel
+        .OutputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T", DataTypeImpl::AllTensorTypes())
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>()),
+    Size);
+
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Size,
+    kOnnxDomain,
+    21, 22,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        // properly force CPU/GPU synch inside the kernel
+        .OutputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T", DataTypeImpl::AllTensorTypes())
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>()),
+    Size);
+
 ONNX_OPERATOR_KERNEL_EX(
     Size,
     kOnnxDomain,
-    13,
+    23,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         // properly force CPU/GPU synch inside the kernel

From 619b74899478479be94c9b231e64aa0569bf4ed5 Mon Sep 17 00:00:00 2001
From: Rishi-Dave <rishipdave@gmail.com>
Date: Wed, 18 Mar 2026 08:58:33 -0700
Subject: [PATCH 2/3] Extend CUDA kernel registrations through opset 25

Bound the opset-23 registrations to 23-24 and add unbounded opset-25
registrations for Flatten, Identity, If, Loop, Scan, ConstantOfShape,
and Size, matching the CPU provider's version ranges. This prevents
opset 24/25 models from falling back to CPU due to VerifyVersion
rejecting open-ended opset-23 kernels.
---
 .../core/providers/cuda/controlflow/if.cc     | 12 +++++-
 .../core/providers/cuda/controlflow/loop.cc   | 14 ++++++-
 .../core/providers/cuda/controlflow/scan.cc   | 12 +++++-
 .../providers/cuda/cuda_execution_provider.cc | 42 ++++++++++++-------
 .../cuda/generator/constant_of_shape.cc       | 13 +++++-
 .../core/providers/cuda/tensor/flatten.cc     | 12 +++++-
 .../core/providers/cuda/tensor/identity_op.cc | 12 +++++-
 .../core/providers/cuda/tensor/size.cc        | 14 ++++++-
 8 files changed, 110 insertions(+), 21 deletions(-)

diff --git a/onnxruntime/core/providers/cuda/controlflow/if.cc b/onnxruntime/core/providers/cuda/controlflow/if.cc
index 83c0c54f76309..4e96fc760dabe 100644
--- a/onnxruntime/core/providers/cuda/controlflow/if.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/if.cc
@@ -66,9 +66,19 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(If,
                                       .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
                                   If);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(If,
+                                  kOnnxDomain,
+                                  23, 24,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  If);
+
 ONNX_OPERATOR_KERNEL_EX(If,
                         kOnnxDomain,
-                        23,
+                        25,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'cond' needs to be on CPU
diff --git a/onnxruntime/core/providers/cuda/controlflow/loop.cc b/onnxruntime/core/providers/cuda/controlflow/loop.cc
index 8d5214813527a..39030092675fb 100644
--- a/onnxruntime/core/providers/cuda/controlflow/loop.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/loop.cc
@@ -76,9 +76,21 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(Loop,
                                       .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
                                   Loop);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Loop,
+                                  kOnnxDomain,
+                                  23, 24,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'M' needs to be on CPU
+                                      .InputMemoryType(OrtMemTypeCPUInput, 1)  // 'cond' needs to be on CPU
+                                      .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("B", DataTypeImpl::GetTensorType<bool>())
+                                      .TypeConstraint("V", DataTypeImpl::AllTensorAndSequenceTensorTypesIRv9()),
+                                  Loop);
+
 ONNX_OPERATOR_KERNEL_EX(Loop,
                         kOnnxDomain,
-                        23,
+                        25,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             .InputMemoryType(OrtMemTypeCPUInput, 0)  // 'M' needs to be on CPU
diff --git a/onnxruntime/core/providers/cuda/controlflow/scan.cc b/onnxruntime/core/providers/cuda/controlflow/scan.cc
index 50a5b5267ff3b..a4930fc105ce0 100644
--- a/onnxruntime/core/providers/cuda/controlflow/scan.cc
+++ b/onnxruntime/core/providers/cuda/controlflow/scan.cc
@@ -132,9 +132,19 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(Scan,
                                       .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorTypesIRv9()),
                                   Scan<9>);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(Scan,
+                                  kOnnxDomain,
+                                  23, 24,
+                                  kCudaExecutionProvider,
+                                  (*KernelDefBuilder::Create())
+                                      // 'I' is in the ONNX spec but is not used for any inputs or outputs
+                                      // .TypeConstraint("I", DataTypeImpl::GetTensorType<int64_t>())
+                                      .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorTypesIRv9()),
+                                  Scan<9>);
+
 ONNX_OPERATOR_KERNEL_EX(Scan,
                         kOnnxDomain,
-                        23,
+                        25,
                         kCudaExecutionProvider,
                         (*KernelDefBuilder::Create())
                             // 'I' is in the ONNX spec but is not used for any inputs or outputs
diff --git a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
index 7a0599ea1b608..b3da2c8c07b74 100755
--- a/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
+++ b/onnxruntime/core/providers/cuda/cuda_execution_provider.cc
@@ -1641,13 +1641,13 @@ class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, R
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Transpose);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Squeeze);
 class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Unsqueeze);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, ConstantOfShape);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Identity);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, If);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Loop);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Scan);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Size);
-class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Flatten);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, ConstantOfShape);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Identity);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, If);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Loop);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Scan);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Size);
+class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Flatten);
 
 #if !defined(DISABLE_FLOAT4_TYPES)
 class ONNX_OPERATOR_TYPED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Float4E2M1x2, Cast);
@@ -1664,6 +1664,13 @@ class ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDom
 // Opset 25.
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Squeeze);
 class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Unsqueeze);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, ConstantOfShape);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Identity);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, If);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Loop);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Scan);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Size);
+class ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Flatten);
 
 #endif
 
@@ -2754,13 +2761,13 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Transpose)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Squeeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 23, Unsqueeze)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, ConstantOfShape)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Identity)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, If)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Loop)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Scan)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Size)>,
-      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, Flatten)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, ConstantOfShape)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, If)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Loop)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Scan)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Size)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_VERSIONED_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 23, 24, Flatten)>,
 
       // Opset 24
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 24, TensorScatter)>,
@@ -2773,6 +2780,13 @@ static Status RegisterCudaKernels(KernelRegistry& kernel_registry) {
       // Opset 25
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Squeeze)>,
       BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Unsqueeze)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, ConstantOfShape)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Identity)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, If)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Loop)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Scan)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Size)>,
+      BuildKernelCreateInfo<ONNX_OPERATOR_KERNEL_CLASS_NAME(kCudaExecutionProvider, kOnnxDomain, 25, Flatten)>,
 #endif
   };
 
diff --git a/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc b/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
index 270bee866bcdf..3ff52fe10ec06 100644
--- a/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
+++ b/onnxruntime/core/providers/cuda/generator/constant_of_shape.cc
@@ -30,10 +30,21 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .TypeConstraint("T2", DataTypeImpl::AllFixedSizeTensorTypes()),
     ConstantOfShape);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    ConstantOfShape,
+    kOnnxDomain,
+    23, 24,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .InputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>())
+        .TypeConstraint("T2", DataTypeImpl::AllFixedSizeTensorTypes()),
+    ConstantOfShape);
+
 ONNX_OPERATOR_KERNEL_EX(
     ConstantOfShape,
     kOnnxDomain,
-    23,
+    25,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .InputMemoryType(OrtMemTypeCPUInput, 0)
diff --git a/onnxruntime/core/providers/cuda/tensor/flatten.cc b/onnxruntime/core/providers/cuda/tensor/flatten.cc
index 7568a132b409c..7d5be7b09929a 100644
--- a/onnxruntime/core/providers/cuda/tensor/flatten.cc
+++ b/onnxruntime/core/providers/cuda/tensor/flatten.cc
@@ -57,10 +57,20 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes()),
     Flatten);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Flatten,
+    kOnnxDomain,
+    23, 24,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .Alias(0, 0)
+        .TypeConstraint("T", DataTypeImpl::AllFixedSizeTensorTypes()),
+    Flatten);
+
 ONNX_OPERATOR_KERNEL_EX(
     Flatten,
     kOnnxDomain,
-    23,
+    25,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .Alias(0, 0)
diff --git a/onnxruntime/core/providers/cuda/tensor/identity_op.cc b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
index 07b2e2af13f20..5120a661ef971 100644
--- a/onnxruntime/core/providers/cuda/tensor/identity_op.cc
+++ b/onnxruntime/core/providers/cuda/tensor/identity_op.cc
@@ -81,10 +81,20 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .Alias(0, 0),
     IdentityOp<false>);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Identity,
+    kOnnxDomain,
+    23, 24,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorAndSequenceTensorTypesIRv9())
+        .Alias(0, 0),
+    IdentityOp<false>);
+
 ONNX_OPERATOR_KERNEL_EX(
     Identity,
     kOnnxDomain,
-    23,
+    25,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         .TypeConstraint("V", DataTypeImpl::AllFixedSizeTensorAndSequenceTensorTypesIRv9())
diff --git a/onnxruntime/core/providers/cuda/tensor/size.cc b/onnxruntime/core/providers/cuda/tensor/size.cc
index 908b51bab58ae..15cd2d2e719bb 100644
--- a/onnxruntime/core/providers/cuda/tensor/size.cc
+++ b/onnxruntime/core/providers/cuda/tensor/size.cc
@@ -43,10 +43,22 @@ ONNX_OPERATOR_VERSIONED_KERNEL_EX(
         .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>()),
     Size);
 
+ONNX_OPERATOR_VERSIONED_KERNEL_EX(
+    Size,
+    kOnnxDomain,
+    23, 24,
+    kCudaExecutionProvider,
+    (*KernelDefBuilder::Create())
+        // properly force CPU/GPU synch inside the kernel
+        .OutputMemoryType(OrtMemTypeCPUInput, 0)
+        .TypeConstraint("T", DataTypeImpl::AllTensorTypes())
+        .TypeConstraint("T1", DataTypeImpl::GetTensorType<int64_t>()),
+    Size);
+
 ONNX_OPERATOR_KERNEL_EX(
     Size,
     kOnnxDomain,
-    23,
+    25,
     kCudaExecutionProvider,
     (*KernelDefBuilder::Create())
         // properly force CPU/GPU synch inside the kernel

From 10138d50d0385c52c23fc4609ed9cac748bc4a17 Mon Sep 17 00:00:00 2001
From: Rishi-Dave <rishipdave@gmail.com>
Date: Fri, 20 Mar 2026 13:23:03 -0700
Subject: [PATCH 3/3] Update docs/OperatorKernels.md for new CUDA kernel
 version ranges

Add version-bounded rows for ConstantOfShape, Flatten, Identity, If,
Loop, Scan, and Size to reflect the new CUDA EP kernel registrations
(21-22, 23-24 bounded, 25+ unbounded).
---
 docs/OperatorKernels.md | 35 ++++++++++++++++++++++++++++-------
 1 file changed, 28 insertions(+), 7 deletions(-)

diff --git a/docs/OperatorKernels.md b/docs/OperatorKernels.md
index f59108e98e556..49d771d2500bb 100644
--- a/docs/OperatorKernels.md
+++ b/docs/OperatorKernels.md
@@ -684,7 +684,10 @@ Do not modify directly.*
 |||[11, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[4, 10]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |ConcatFromSequence|*in* input_sequence:**S**<br> *out* concat_result:**T**|11+|**S** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8))|
-|ConstantOfShape|*in* input:**T1**<br> *out* output:**T2**|9+|**T1** = tensor(int64)<br/> **T2** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|ConstantOfShape|*in* input:**T1**<br> *out* output:**T2**|25+|**T1** = tensor(int64)<br/> **T2** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**T1** = tensor(int64)<br/> **T2** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**T1** = tensor(int64)<br/> **T2** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[9, 20]|**T1** = tensor(int64)<br/> **T2** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |Conv|*in* X:**T**<br> *in* W:**T**<br> *in* B:**T**<br> *out* Y:**T**|22+|**T** = tensor(bfloat16), tensor(double), tensor(float), tensor(float16)|
 |||[11, 21]|**T** = tensor(double), tensor(float), tensor(float16)|
 |||[1, 10]|**T** = tensor(double), tensor(float), tensor(float16)|
@@ -721,7 +724,10 @@ Do not modify directly.*
 |Expand|*in* input:**T**<br> *in* shape:**tensor(int64)**<br> *out* output:**T**|13+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[8, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |EyeLike|*in* input:**T1**<br> *out* output:**T2**|9+|**T1** = tensor(double), tensor(float), tensor(int32), tensor(int64), tensor(uint64)<br/> **T2** = tensor(double), tensor(float), tensor(int32), tensor(int64), tensor(uint64)|
-|Flatten|*in* input:**T**<br> *out* output:**T**|13+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|Flatten|*in* input:**T**<br> *out* output:**T**|25+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[13, 20]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[11, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[9, 10]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[1, 8]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
@@ -758,11 +764,17 @@ Do not modify directly.*
 |||[6, 21]|**T** = tensor(double), tensor(float), tensor(float16)|
 |HardSwish|*in* X:**T**<br> *out* Y:**T**|22+|**T** = tensor(bfloat16), tensor(double), tensor(float), tensor(float16)|
 |||[14, 21]|**T** = tensor(double), tensor(float), tensor(float16)|
-|Identity|*in* input:**T**<br> *out* output:**T**<br><br>or<br><br>*in* input:**V**<br> *out* output:**V**|19+|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|Identity|*in* input:**T**<br> *out* output:**T**<br><br>or<br><br>*in* input:**V**<br> *out* output:**V**|25+|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[19, 20]|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[14, 18]|**V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||13|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[1, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
-|If|*in* cond:**B**<br> *out* outputs:**V**|19+|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|If|*in* cond:**B**<br> *out* outputs:**V**|25+|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[19, 20]|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[13, 18]|**B** = tensor(bool)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[11, 12]|**B** = tensor(bool)<br/> **V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[1, 10]|**B** = tensor(bool)<br/> **V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
@@ -791,7 +803,10 @@ Do not modify directly.*
 |LogSoftmax|*in* input:**T**<br> *out* output:**T**|13+|**T** = tensor(double), tensor(float), tensor(float16)|
 |||[11, 12]|**T** = tensor(double), tensor(float), tensor(float16)|
 |||[1, 10]|**T** = tensor(double), tensor(float), tensor(float16)|
-|Loop|*in* M:**I**<br> *in* cond:**B**<br> *in* v_initial:**V**<br> *out* v_final_and_scan_outputs:**V**|19+|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|Loop|*in* M:**I**<br> *in* cond:**B**<br> *in* v_initial:**V**<br> *out* v_final_and_scan_outputs:**V**|25+|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[19, 20]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(float8e4m3fn)), seq(tensor(float8e4m3fnuz)), seq(tensor(float8e5m2)), seq(tensor(float8e5m2fnuz)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[13, 18]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = seq(tensor(bfloat16)), seq(tensor(bool)), seq(tensor(double)), seq(tensor(float)), seq(tensor(float16)), seq(tensor(int16)), seq(tensor(int32)), seq(tensor(int64)), seq(tensor(int8)), seq(tensor(string)), seq(tensor(uint16)), seq(tensor(uint32)), seq(tensor(uint64)), seq(tensor(uint8)), tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[11, 12]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[1, 10]|**B** = tensor(bool)<br/> **I** = tensor(int64)<br/> **V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
@@ -890,7 +905,10 @@ Do not modify directly.*
 |RotaryEmbedding|*in* X:**T**<br> *in* cos_cache:**T**<br> *in* sin_cache:**T**<br> *in* position_ids:**M**<br> *out* Y:**T**|23+|**M** = tensor(int64)<br/> **T** = tensor(bfloat16), tensor(float), tensor(float16)|
 |Round|*in* X:**T**<br> *out* Y:**T**|11+|**T** = tensor(double), tensor(float), tensor(float16)|
 |ScaledTanh|*in* input:**T**<br> *out* output:**T**|1+|**T** = tensor(double), tensor(float), tensor(float16)|
-|Scan|*in* initial_state_and_scan_inputs:**V**<br> *out* final_state_and_scan_outputs:**V**<br><br>or<br><br>*in* sequence_lens:**I**<br> *in* initial_state_and_scan_inputs:**V**<br> *out* final_state_and_scan_outputs:**V**|19+|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|Scan|*in* initial_state_and_scan_inputs:**V**<br> *out* final_state_and_scan_outputs:**V**<br><br>or<br><br>*in* sequence_lens:**I**<br> *in* initial_state_and_scan_inputs:**V**<br> *out* final_state_and_scan_outputs:**V**|25+|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[23, 24]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[21, 22]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
+|||[19, 20]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[16, 18]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[11, 15]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |||[9, 10]|**V** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
@@ -923,7 +941,10 @@ Do not modify directly.*
 |Sign|*in* input:**T**<br> *out* output:**T**|13+|**T** = tensor(bfloat16), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)|
 |SimplifiedLayerNormalization|*in* X:**T**<br> *in* scale:**V**<br> *out* Y:**V**<br> *out* inv_std_var:**U**|1+|**T** = tensor(bfloat16), tensor(double), tensor(float), tensor(float16)<br/> **U** = tensor(double), tensor(float)<br/> **V** = tensor(bfloat16), tensor(double), tensor(float), tensor(float16)|
 |Sin|*in* input:**T**<br> *out* output:**T**|7+|**T** = tensor(double), tensor(float), tensor(float16)|
-|Size|*in* data:**T**<br> *out* size:**T1**|13+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
+|Size|*in* data:**T**<br> *out* size:**T1**|25+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
+|||[23, 24]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
+|||[21, 22]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
+|||[13, 20]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
 |||[1, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T1** = tensor(int64)|
 |Slice|*in* data:**T**<br> *in* starts:**Tind**<br> *in* ends:**Tind**<br> *in* axes:**Tind**<br> *in* steps:**Tind**<br> *out* output:**T**<br><br>or<br><br>*in* data:**T**<br> *out* output:**T**|13+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **Tind** = tensor(int32), tensor(int64)|
 |||[11, 12]|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **Tind** = tensor(int32), tensor(int64)|