Unable to build python wheel with cuda 13

**Describe the bug**
A clear and concise description of what the bug is.

When running onnxruntime-genai-cuda , the CUDA provider fails to load because it requires cublasLt64_12.dll, which only exists in CUDA 12.

```
File "D:\Venna\main.py", line 76, in
main()
File "D:\Venna\main.py", line 27, in main
model = og.Model(config)
^^^^^^^^^^^^^^^^
RuntimeError: E:\_work\1\s\onnxruntime\core\session\provider_bridge_ort.cc:1844 onnxruntime::ProviderLibrary::Get [ONNXRuntimeError] : 1 : FAIL : Error loading "D:\Venna\.venv\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll" which depends on "cublasLt64_12.dll" which is missing. (Error 126: "The specified module could not be found.")
```

I then tried building onnxruntime-genai from source using CUDA 13, but the build fails during nvcc CUDA compilation.

```
.\build.bat --use_cuda --config Release --skip_examples
2025-11-16 00:15:00,435 util.run [INFO] - Running subprocess in 'D:\onnxruntime-genai'
'D:\Program Files\Microsoft Visual Studio\18\Community\Common7\IDE\CommonExtensions\Microsoft\CMake\CMake\bin\cmake.EXE' -G 'Visual Studio 17 2022' -T 'cuda=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0' -DCMAKE_BUILD_TYPE=Release -S 'D:\onnxruntime-genai' -B 'D:\onnxruntime-genai\build\Windows\Release' -DCMAKE_POSITION_INDEPENDENT_CODE=ON -DUSE_CUDA=ON -DUSE_TRT_RTX=OFF -DUSE_ROCM=OFF -DUSE_DML=OFF -DENABLE_JAVA=OFF -DBUILD_WHEEL=ON -DUSE_GUIDANCE=OFF -DPUBLISH_JAVA_MAVEN_LOCAL=OFF '-DCMAKE_CUDA_COMPILER=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc'
-- Selecting Windows SDK version 10.0.26100.0 to target Windows 10.0.26200.
-- CET shadow stack enabled (/CETCOMPAT).
-- Using ONNX Runtime package Microsoft.ML.OnnxRuntime.Gpu.Windows version 1.22.0
-- ORT_HEADER_DIR: D:/onnxruntime-genai/build/Windows/Release/_deps/ortlib-src/buildTransitive/native/include
-- ORT_LIB_DIR: D:/onnxruntime-genai/build/Windows/Release/_deps/ortlib-src/runtimes/win-x64/native
Loading Dependencies URLs ...
Loading Dependencies ...
CMake Deprecation Warning at build/Windows/Release/_deps/pybind11_project-src/CMakeLists.txt:13 (cmake_minimum_required):
Compatibility with CMake < 3.10 will be removed from a future version of
CMake.

Update the VERSION argument value. Or, use the ... syntax
to tell CMake that the project requires at least but has been updated
to work with policies introduced by or earlier.

-- pybind11 v2.13.6
-- _STATIC_MSVC_RUNTIME_LIBRARY: OFF
-- Fetch json
-- Tokenizer needed.
-- CMAKE_CUDA_COMPILER_VERSION: 13.0.88
CMake Deprecation Warning at cmake/check_cuda.cmake:20 (cmake_policy):
The OLD behavior for policy CMP0104 will be removed from a future version
of CMake.

The cmake-policies(7) manual explains that the OLD behaviors of all
policies are deprecated and that a policy should be set to OLD only under
specific short-term circumstances. Projects should be ported to the NEW
behavior and not rely on setting a policy to OLD.
Call Stack (most recent call first):
CMakeLists.txt:102 (include)

-- CMAKE_CUDA_COMPILER_VERSION: 13.0.88
------------------Enabling tests------------------
-- Including CUDA kernel tests in the build.
-- Enable STABLE_TOPK in CUDA kernel tests.
------------------Enabling Python Wheel------------------
Setting up wheel files in : D:/onnxruntime-genai/build/Windows/Release/wheel
------------------Enabling model benchmark------------------
-- Configuring done (9.9s)
-- Generating done (0.2s)
-- Build files have been written to: D:/onnxruntime-genai/build/Windows/Release
2025-11-16 00:15:10,752 util.run [INFO] - Running subprocess in 'D:\onnxruntime-genai'
'D:\Program Files\Microsoft Visual Studio\18\Community\Common7\IDE\CommonExtensions\Microsoft\CMake\CMake\bin\cmake.EXE' --build 'D:\onnxruntime-genai\build\Windows\Release' --config Release
MSBuild version 18.0.2+995a3dce4 for .NET Framework

Checking File Globs
noexcep_operators.vcxproj -> D:\onnxruntime-genai\build\Windows\Release\lib\Release\noexcep_operators.lib
ocos_operators.vcxproj -> D:\onnxruntime-genai\build\Windows\Release\lib\Release\ocos_operators.lib
Compiling CUDA source file ..\..\..\src\cuda\beam_search_scorer_cuda.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\/src/cuda/beam_search_scorer_cuda.cu.obj "D:\onnxruntime-genai\src\cuda\beam_s
earch_scorer_cuda.cu"
beam_search_scorer_cuda.cu
tmpxft_00008450_00000000-7_beam_search_scorer_cuda.cudafe1.cpp
Compiling CUDA source file ..\..\..\src\cuda\beam_search_topk.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\beam_search_topk.obj "D:\onnxruntime-genai\src\cuda\beam_search_topk.cu"
beam_search_topk.cu
tmpxft_000012ac_00000000-7_beam_search_topk.cudafe1.cpp
Compiling CUDA source file ..\..\..\src\cuda\cuda_sampling.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\cuda_sampling.obj "D:\onnxruntime-genai\src\cuda\cuda_sampling.cu"
cuda_sampling.cu
tmpxft_00000ab8_00000000-7_cuda_sampling.cudafe1.cpp
Compiling CUDA source file ..\..\..\src\cuda\cuda_topk.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\cuda_topk.obj "D:\onnxruntime-genai\src\cuda\cuda_topk.cu"
cuda_topk.cu
tmpxft_00005db4_00000000-7_cuda_topk.cudafe1.cpp
Compiling CUDA source file ..\..\..\src\cuda\model_kernels.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\model_kernels.obj "D:\onnxruntime-genai\src\cuda\model_kernels.cu"
model_kernels.cu
tmpxft_00006ed8_00000000-7_model_kernels.cudafe1.cpp
Compiling CUDA source file ..\..\..\src\cuda\search_cuda.cu...

D:\onnxruntime-genai\build\Windows\Release>"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\bin\nvcc.exe" --use-local-env -ccbin "D:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\MSVC\14.44.35207\bin\HostX64\x64" -x cu -I"C:\Program Files\NVIDIA GPU C
omputing Toolkit\CUDA\v13.0\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\cccl" -I"D:\onnxruntime-genai\build\Windows\Release\_deps\ortlib-src\buildTransitive\native\include" -I"D:\onnxruntime-genai\src" -I"C:\Program Files\NVIDIA GPU Computing T
oolkit\CUDA\v13.0\include" --keep-dir onnxrunt.A5C0B7CE\x64\Release -maxrregcount=0 --machine 64 --compile -cudart static -allow-unsupported-compiler -Xcudafe --diag_suppress=2803 --expt-relaxed-constexpr -std=c++17 --generate-code=arch=compute_75,code=[compute_75,sm_7
5] -diag-suppress=1650 -diag-suppress=221 -Wno-deprecated-gpu-targets -Xcompiler="/EHsc -Ob2 /WX" -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=
0 -D"CMAKE_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DNDEBUG -D_DISABLE_CONSTEXPR_MUTEX_CONSTRUCTOR -DUSE_WINML=0 -DUSE_CUDA=1 -DUSE_ROCM=0 -DUSE_DML=0 -DUSE_CXX17=1 -DBUILDING_ORT_GENAI_C -DUSE_GUIDANCE=0 -DTEST_PHI2=0 -D"CMAKE
_INTDIR=\"Release\"" -Donnxruntime_genai_cuda_EXPORTS -Xcompiler "/EHsc /W4 /nologo /O2 /FS /MD " -Xcompiler "/Fdonnxruntime-genai-cuda.dir\Release\vc143.pdb" -o onnxruntime-genai-cuda.dir\Release\/src/cuda/search_cuda.cu.obj "D:\onnxruntime-genai\src\cuda\search_cuda.cu"
search_cuda.cu
tmpxft_00001388_00000000-7_search_cuda.cudafe1.cpp
beam_search_scorer_cuda.cpp
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\curand_poisson.h(649,55): error C2220: the following warning is treated as an error [D:\onnxruntime-genai\build\Windows\Release\onnxruntime-genai-cuda.vcxproj]
(compiling source file '../../../src/cuda/beam_search_scorer_cuda.cpp')

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\curand_poisson.h(649,55): warning C4996: 'double4::x': use double4_16a or double4_32a [D:\onnxruntime-genai\build\Windows\Release\onnxruntime-genai-cuda.vcxproj]
(compiling source file '../../../src/cuda/beam_search_scorer_cuda.cpp')

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\curand_poisson.h(650,55): warning C4996: 'double4::y': use double4_16a or double4_32a [D:\onnxruntime-genai\build\Windows\Release\onnxruntime-genai-cuda.vcxproj]
(compiling source file '../../../src/cuda/beam_search_scorer_cuda.cpp')

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\curand_poisson.h(651,55): warning C4996: 'double4::z': use double4_16a or double4_32a [D:\onnxruntime-genai\build\Windows\Release\onnxruntime-genai-cuda.vcxproj]
(compiling source file '../../../src/cuda/beam_search_scorer_cuda.cpp')

C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v13.0\include\curand_poisson.h(652,55): warning C4996: 'double4::w': use double4_16a or double4_32a [D:\onnxruntime-genai\build\Windows\Release\onnxruntime-genai-cuda.vcxproj]
(compiling source file '../../../src/cuda/beam_search_scorer_cuda.cpp')

ortcustomops.vcxproj -> D:\onnxruntime-genai\build\Windows\Release\lib\Release\ortcustomops.lib
Traceback (most recent call last):
File "D:\onnxruntime-genai\\build.py", line 793, in
build(arguments, environment)
File "D:\onnxruntime-genai\\build.py", line 650, in build
util.run(make_command, env=env)
File "D:\onnxruntime-genai\tools\python\util\run.py", line 56, in run
completed_process = subprocess.run(
^^^^^^^^^^^^^^^
File "C:\Users\Lenovo\miniforge3\Lib\subprocess.py", line 571, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['D:\\Program Files\\Microsoft Visual Studio\\18\\Community\\Common7\\IDE\\CommonExtensions\\Microsoft\\CMake\\CMake\\bin\\cmake.EXE', '--build', 'D:\\onnxruntime-genai\\build\\Windows\\Release', '--config', 'Release']' returned non-zero exit status 1.
```

**Desktop (please complete the following information):**
- OS: Windows
- Version 11
```
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2025 NVIDIA Corporation
Built on Wed_Aug_20_13:58:20_Pacific_Daylight_Time_2025
Cuda compilation tools, release 13.0, V13.0.88
Build cuda_13.0.r13.0/compiler.36424714_0
```

Start with build.bat and cmake/check_cuda.cmake, then inspect the CUDA sources named in the build output, including src/cuda/beam_search_scorer_cuda.cu, beam_search_topk.cu, cuda_sampling.cu, cuda_topk.cu, and model_kernels.cu. Reproduce the CUDA 13 Windows wheel build and identify the compilation failure. Done means the CUDA build completes and the resulting wheel's CUDA provider loads without the reported missing-library error.

Written by the indexing model from the issue text.