diff --git a/.github/workflows/wheels.yml b/.github/workflows/wheels.yml
new file mode 100644
index 0000000000..aa375d7e50
--- /dev/null
+++ b/.github/workflows/wheels.yml
@@ -0,0 +1,111 @@
+name: Build Wheels
+
+on:
+  workflow_dispatch:
+  push:
+    tags:
+      - "*"
+
+jobs:
+  build_macos_wheels:
+    name: Build wheels on ${{matrix.os}}-${{matrix.arch}}
+    runs-on: ${{matrix.os}}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - {os: macos-13, arch: x86_64, build: "*"}
+          - {os: macos-14, arch: arm64, build: "*"}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: "recursive"
+
+      - name: Build wheels
+        uses: pypa/cibuildwheel@v2.17.0
+        if: matrix.arch == 'arm64'
+        env:
+          CIBW_ARCHS: ${{ matrix.arch }}
+          CIBW_BUILD: ${{ matrix.build }}
+          CIBW_BUILD_FRONTEND: build
+          CIBW_ENVIRONMENT_MACOS: CMAKE_ARGS='-DLLAMA_METAL=on' DYLD_LIBRARY_PATH="$(pwd)/llama_cpp"
+        with:
+          config-file: ./pyproject.toml
+
+      - name: Build wheels
+        uses: pypa/cibuildwheel@v2.17.0
+        if: matrix.arch == 'x86_64'
+        env:
+          CIBW_ARCHS: ${{ matrix.arch }}
+          CIBW_BUILD: ${{ matrix.build }}
+          CIBW_BUILD_FRONTEND: build
+          CIBW_ENVIRONMENT_MACOS: CMAKE_ARGS='-DLLAMA_METAL=off' DYLD_LIBRARY_PATH="$(pwd)/llama_cpp"
+        with:
+          config-file: ./pyproject.toml
+
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: cibw-wheels-${{matrix.os}}-${{matrix.arch}}-${{strategy.job-index}}
+          path: ./wheelhouse/*.whl
+
+
+  build_linux_wheels:
+    name: Build wheels on ${{matrix.os}}-${{matrix.arch}}
+    runs-on: ${{matrix.os}}
+    strategy:
+      fail-fast: true
+      matrix:
+        include:
+          - {os: ubuntu-latest, arch: x86_64, build: "*"}
+          - {os: ubuntu-latest, arch: aarch64, build: "*[61]-manylinux_aarch64"}
+          - {os: ubuntu-latest, arch: aarch64, build: "*[72]-manylinux_aarch64"}
+          - {os: ubuntu-latest, arch: aarch64, build: "*[83]-manylinux_aarch64"}
+          - {os: ubuntu-latest, arch: aarch64, build: "*[94]-manylinux_aarch64"}
+          - {os: ubuntu-latest, arch: aarch64, build: "*[05]-manylinux_aarch64"}
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: "recursive"
+
+      - name: Set up QEMU
+        if: matrix.arch == 'aarch64'
+        uses: docker/setup-qemu-action@v3
+        with:
+          platforms: arm64
+
+      - name: Build wheels
+        uses: pypa/cibuildwheel@v2.17.0
+        env:
+          CIBW_ARCHS: ${{ matrix.arch }}
+          CIBW_BUILD: ${{ matrix.build }}
+          CIBW_BUILD_FRONTEND: build
+          CIBW_SKIP: '*-musllinux*'
+        with:
+          config-file: ./pyproject.toml
+
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: cibw-wheels-${{matrix.os}}-${{matrix.arch}}-${{strategy.job-index}}
+          path: ./wheelhouse/*.whl
+
+
+  build_sdist:
+    name: Build source distribution
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: "recursive"
+
+      - name: Build sdist
+        run: pipx run build --sdist
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: cibw-sdist
+          path: dist/*.tar.gz
diff --git a/.gitignore b/.gitignore
index 9d68dbcd9d..c2321418c6 100644
--- a/.gitignore
+++ b/.gitignore
@@ -42,6 +42,7 @@ share/python-wheels/
 .installed.cfg
 *.egg
 MANIFEST
+wheelhouse/
 
 # PyInstaller
 #  Usually these files are written by a python script from a template
@@ -178,3 +179,4 @@ cython_debug/
 
 # downloaded model .bin files
 docker/open_llama/*.bin
+install/
diff --git a/pyproject.toml b/pyproject.toml
index 2f3d3ced0f..c712057081 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -3,7 +3,7 @@ requires = ["scikit-build-core[pyproject]>=0.5.1"]
 build-backend = "scikit_build_core.build"
 
 [project]
-name = "llama_cpp_python"
+name = "llama_cpp_python-cross"
 dynamic = ["version"]
 description = "Python bindings for the llama.cpp library"
 readme = "README.md"
@@ -73,3 +73,23 @@ Changelog = "https://llama-cpp-python.readthedocs.io/en/latest/changelog/"
 
 [tool.pytest.ini_options]
 testpaths = "tests"
+
+[tool.cibuildwheel]
+before-all = "uname -a"
+
+[tool.cibuildwheel.linux]
+manylinux-aarch64-image = "quay.io/pypa/manylinux_2_28_aarch64:latest"
+manylinux-x86_64-image = "quay.io/pypa/manylinux_2_28_x86_64:latest"
+before-all = [
+  "uname -a",
+  "yum install -y wget openblas-devel",
+  "wget https://dl.fedoraproject.org/pub/epel/8/Everything/$(uname -m)/Packages/c/ccache-3.7.7-1.el8.$(uname -m).rpm",
+  "rpm -Uvh *.rpm",
+  "dnf install -y ccache",
+]
+
+[tool.cibuildwheel.linux.environment]
+LIBRARY_PATH = "/project/llama_cpp"
+LD_LIBRARY_PATH = "/project/llama_cpp"
+RUNPATH = "/project/llama_cpp"
+CMAKE_ARGS = "-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS"