diff --git a/onnxruntime-gcc13.patch b/onnxruntime-gcc13.patch
deleted file mode 100644
index 63e086a24be..00000000000
--- a/onnxruntime-gcc13.patch
+++ /dev/null
@@ -1,26 +0,0 @@
-diff --git a/onnxruntime/core/optimizer/selectors_actions/selector_action_transformer.cc b/onnxruntime/core/optimizer/selectors_actions/selector_action_transformer.cc
-index b68cbaf..b1d6c51 100644
---- a/onnxruntime/core/optimizer/selectors_actions/selector_action_transformer.cc
-+++ b/onnxruntime/core/optimizer/selectors_actions/selector_action_transformer.cc
-@@ -147,7 +147,7 @@ static Status MatchAndProcess(
-       RuntimeOptimizationRecord::ProducedOpIdVector produced_op_ids{};
-       produced_op_ids.reserve(action_saved_state.produced_node_op_schemas.size());
- 
--      for (const auto op_schema : action_saved_state.produced_node_op_schemas) {
-+      for (const auto& op_schema : action_saved_state.produced_node_op_schemas) {
-         produced_op_ids.push_back(utils::MakeOpId(*op_schema));
-         if (save_context->record_produced_node_op_schema) {
-           status = save_context->record_produced_node_op_schema(*op_schema);
-diff --git a/onnxruntime/core/session/inference_session.cc b/onnxruntime/core/session/inference_session.cc
-index 023cbcb..8fc17f0 100644
---- a/onnxruntime/core/session/inference_session.cc
-+++ b/onnxruntime/core/session/inference_session.cc
-@@ -907,7 +907,7 @@ common::Status InferenceSession::SaveToOrtFormat(const std::filesystem::path& fi
-   ORT_RETURN_IF_ERROR(kernel_type_str_resolver.RegisterGraphNodeOpSchemas(model_->MainGraph()));
-   ORT_RETURN_IF_ERROR(standalone::RegisterCustomOpNodeSchemas(kernel_type_str_resolver, model_->MainGraph()));
- 
--  for (const auto op_schema : saved_runtime_optimization_produced_node_op_schemas_) {
-+  for (const auto& op_schema : saved_runtime_optimization_produced_node_op_schemas_) {
-     ORT_RETURN_IF_ERROR(kernel_type_str_resolver.RegisterOpSchema(*op_schema));
-   }
- 
diff --git a/onnxruntime/cms-changes.patch b/onnxruntime/cms-changes.patch
new file mode 100644
index 00000000000..1f1009a6b46
--- /dev/null
+++ b/onnxruntime/cms-changes.patch
@@ -0,0 +1,47 @@
+diff --git a/onnxruntime/core/mlas/lib/platform.cpp b/onnxruntime/core/mlas/lib/platform.cpp
+index eccde79848e61..4a107f173da07 100644
+--- a/onnxruntime/core/mlas/lib/platform.cpp
++++ b/onnxruntime/core/mlas/lib/platform.cpp
+@@ -23,6 +23,7 @@ Module Name:
+ #include "kleidiai/mlasi_kleidiai.h"
+ #endif
+ 
++#include <string>
+ #include <thread>
+ #include <mutex>
+ 
+@@ -355,8 +356,11 @@ Return Value:
+         //
+ 
+         uint64_t xcr0 = MlasReadExtendedControlRegister(_XCR_XFEATURE_ENABLED_MASK);
++        const char *cpu_opt = std::getenv("MLAS_DYNAMIC_CPU_ARCH");
++        if (cpu_opt == nullptr) cpu_opt = "99";
++        auto opt = std::stoi(cpu_opt);
+ 
+-        if ((xcr0 & 0x6) == 0x6) {
++        if ((opt > 0) && (xcr0 & 0x6) == 0x6) {
+ 
+             this->GemmFloatKernel = MlasGemmFloatKernelAvx;
+ 
+@@ -390,7 +394,7 @@ Return Value:
+             __cpuid_count(7, 0, Cpuid7[0], Cpuid7[1], Cpuid7[2], Cpuid7[3]);
+ #endif
+ 
+-            if (((Cpuid1[2] & 0x1000) != 0) && ((Cpuid7[1] & 0x20) != 0)) {
++            if ((opt > 1) && ((Cpuid1[2] & 0x1000) != 0) && ((Cpuid7[1] & 0x20) != 0)) {
+ 
+                 this->Avx2Supported_ = true;
+ 
+diff --git a/onnxruntime/core/providers/cpu/tensor/scatter.cc b/onnxruntime/core/providers/cpu/tensor/scatter.cc
+index c7a2005924836..c938ca9bf073d 100644
+--- a/onnxruntime/core/providers/cpu/tensor/scatter.cc
++++ b/onnxruntime/core/providers/cpu/tensor/scatter.cc
+@@ -326,7 +326,7 @@ Status ScatterData(
+   // and so on
+   std::vector<int64_t> dim_block_size(num_dims);
+ 
+-  dim_block_size.back() = 1;
++  dim_block_size[num_dims-1] = 1;
+   if (num_dims > 1) {
+     // We start at num_dims - 2 because we already pre-populated
+     // the last element above
diff --git a/onnxruntime.spec b/onnxruntime/spec
similarity index 84%
rename from onnxruntime.spec
rename to onnxruntime/spec
index e59b1ddcb8a..80ba97847b8 100644
--- a/onnxruntime.spec
+++ b/onnxruntime/spec
@@ -1,11 +1,11 @@
-### RPM external onnxruntime 1.20.1
+### RPM external onnxruntime 1.25.1
 ## INITENV +PATH PYTHON3PATH %{i}/${PYTHON3_LIB_SITE_PACKAGES}
 ## INCLUDE cuda-flags
-%define github_user cms-externals
-%define branch cms/v%{realversion}
-%define tag efe7f6a3859bedbad40a2991480be4e7584b1582
+%define github_user microsoft
+%define branch main
+%define tag v%{realversion}
 Source: git+https://github.com/%{github_user}/%{n}.git?obj=%{branch}/%{tag}&export=%{n}-%{realversion}&submodules=1&output=/%{n}-%{realversion}.tgz
-Patch0: onnxruntime-gcc13
+Patch0: onnxruntime/cms-changes
 
 BuildRequires: cmake ninja
 Requires: protobuf py3-numpy py3-onnx zlib libpng py3-pybind11 re2 eigen
@@ -26,7 +26,6 @@ fi
 cmake ../%{n}-%{realversion}/cmake \
    -G Ninja \
    -Wno-dev \
-   -DPYTHON_EXECUTABLE=${PYTHON3_ROOT}/bin/python3 \
    -DCMAKE_BUILD_TYPE=Release \
    -DCMAKE_INSTALL_PREFIX="%{i}" \
    -DCMAKE_INSTALL_LIBDIR=lib \
@@ -43,19 +42,13 @@ cmake ../%{n}-%{realversion}/cmake \
    -DCMAKE_TRY_COMPILE_PLATFORM_VARIABLES="CMAKE_CUDA_RUNTIME_LIBRARY" \
 %endif
    -Donnxruntime_BUILD_CSHARP=OFF \
-   -Donnxruntime_USE_OPENMP=OFF \
-   -Donnxruntime_USE_TVM=OFF \
-   -Donnxruntime_USE_LLVM=OFF \
    -Donnxruntime_ENABLE_MICROSOFT_INTERNAL=OFF \
-   -Donnxruntime_USE_NUPHAR=OFF \
    -Donnxruntime_USE_TENSORRT=OFF \
    -Donnxruntime_CROSS_COMPILING=OFF \
    -Donnxruntime_USE_FULL_PROTOBUF=ON \
    -Donnxruntime_DISABLE_CONTRIB_OPS=OFF \
-   -Donnxruntime_PREFER_SYSTEM_LIB=ON \
    -Donnxruntime_BUILD_UNIT_TESTS=OFF \
    -DCMAKE_PREFIX_PATH="${ZLIB_ROOT};${LIBPNG_ROOT};${PROTOBUF_ROOT};${PY3_PYBIND11_ROOT};${RE2_ROOT};${EIGEN_ROOT}" \
-   -DRE2_INCLUDE_DIR="${RE2_ROOT}/include" \
    -DCMAKE_CXX_FLAGS="-Wno-error=stringop-overflow -Wno-error=maybe-uninitialized -Wno-error=overloaded-virtual"
 
 # -Wno-error=stringop-overflow is needed to work around a false positive string overflow,