diff --git a/.github/workflows/build-wheels-batch.yml b/.github/workflows/build-wheels-batch.yml
new file mode 100644
index 00000000..5208a082
--- /dev/null
+++ b/.github/workflows/build-wheels-batch.yml
@@ -0,0 +1,45 @@
+name: Batch Build Wheels
+
+on:
+  workflow_dispatch:
+    inputs:
+      versions:
+        description: 'Comma-seperated version tags of llama-cpp-python to build'
+        default: 'v0.1.77,v0.1.76'
+        required: true
+        type: string
+
+permissions:
+  contents: write
+  
+jobs:
+  define_matrix:
+    name: Define Workflow Matrix
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      PCKGVERS: ${{ inputs.versions }}
+        
+    steps:
+      - uses: actions/checkout@v3
+      
+      - name: Define Job Output
+        id: set-matrix
+        run: |
+          $x = $env:PCKGVERS.Split(',').Trim() -Join ', '
+          Write-Output "matrix={`"node`":[$x]}" >> $env:GITHUB_OUTPUT
+          
+  run_workflows:
+    name: Build ${{ matrix.version }} Wheels
+    needs: define_matrix
+    strategy:
+      max-parallel: 1
+      matrix:
+        version: ${{ fromJSON(needs.define_matrix.outputs.matrix) }}
+    uses: ./.github/workflows/build-wheels.yml
+    with:
+      version: ${{ matrix.version }}
diff --git a/.github/workflows/build-wheels-cpu.yml b/.github/workflows/build-wheels-cpu.yml
new file mode 100644
index 00000000..fbdad596
--- /dev/null
+++ b/.github/workflows/build-wheels-cpu.yml
@@ -0,0 +1,77 @@
+name: Build CPU-only Wheels
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of llama-cpp-python to build: v0.1.77'
+        default: 'v0.1.77'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of llama-cpp-python to build: v0.1.77'
+        default: 'v0.1.77'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-20.04, windows-latest]
+        pyver: ["3.7", "3.8", "3.9", "3.10", "3.11"]
+        avx: ["AVX","AVX2","AVX512","basic"]
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      AVXVER: ${{ matrix.avx }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          repository: 'abetlen/llama-cpp-python'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+          
+      - uses: actions/setup-python@v3
+        with:
+          python-version: ${{ matrix.pyver }}
+          
+      - name: Install Dependencies
+        run: |
+          python -m pip install build wheel
+          
+      - name: Build Wheel
+        run: |
+          $env:VERBOSE = '1'
+          $env:FORCE_CMAKE = '1'
+          if ($env:AVXVER -eq 'AVX') {$env:CMAKE_ARGS = '-DLLAMA_AVX2=off'}
+          if ($env:AVXVER -eq 'AVX512') {$env:CMAKE_ARGS = '-DLLAMA_AVX512=on'}
+          if ($env:AVXVER -eq 'basic') {$env:CMAKE_ARGS = '-DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_FMA=off -DLLAMA_F16C=off'}
+          python -m build --wheel -C--build-option=egg_info "-C--build-option=--tag-build=+cpu$env:AVXVER"
+          
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.6.1
+        continue-on-error: true
+        with:
+          file: ./dist/*.whl
+          tag: 'cpu'
+          file_glob: true
+          make_latest: false
+          overwrite: true
+        
+      - uses: actions/upload-artifact@v3
+        if: steps.upload-release.outcome == 'failure'
+        with:
+          name: cpu
+          path: ./dist/*.whl
diff --git a/.github/workflows/build-wheels.yml b/.github/workflows/build-wheels.yml
new file mode 100644
index 00000000..0c7febe6
--- /dev/null
+++ b/.github/workflows/build-wheels.yml
@@ -0,0 +1,112 @@
+name: Build Wheels
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version tag of llama-cpp-python to build: v0.1.77'
+        default: 'v0.1.77'
+        required: false
+        type: string
+  workflow_call:
+    inputs:
+      version:
+        description: 'Version tag of llama-cpp-python to build: v0.1.77'
+        default: 'v0.1.77'
+        required: false
+        type: string
+
+permissions:
+  contents: write
+
+jobs:
+  build_wheels:
+    name: ${{ matrix.os }} ${{ matrix.pyver }} ${{ matrix.cuda }} ${{ matrix.releasetag == 'wheels' && 'AVX2' || matrix.releasetag }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-20.04, windows-latest]
+        pyver: ["3.7", "3.8", "3.9", "3.10", "3.11"]
+        cuda: ["11.6.2", "11.7.1", "11.8.0", "12.0.1", "12.1.1", "12.2.0"]
+        releasetag: ["AVX","wheels","AVX512","basic"]
+    defaults:
+      run:
+        shell: pwsh
+    env:
+      CUDAVER: ${{ matrix.cuda }}
+      AVXVER: ${{ matrix.releasetag }}
+      PCKGVER: ${{ inputs.version }}
+
+    steps:
+      - uses: actions/checkout@v3
+        with:
+          repository: 'abetlen/llama-cpp-python'
+          ref: ${{ inputs.version }}
+          submodules: 'recursive'
+          
+      - uses: actions/setup-python@v3
+        with:
+          python-version: ${{ matrix.pyver }}
+        
+      - name: Setup Mamba
+        uses: conda-incubator/setup-miniconda@v2.2.0
+        with:
+          activate-environment: "build"
+          python-version: ${{ matrix.pyver }}
+          miniforge-variant: Mambaforge
+          miniforge-version: latest
+          use-mamba: true
+          add-pip-as-python-dependency: true
+          auto-activate-base: false
+          
+      - name: Install Dependencies
+        run: |
+          $cudaVersion = $env:CUDAVER
+          $cudaChannels = ''
+          $cudaNum = [int]$cudaVersion.substring($cudaVersion.LastIndexOf('.')+1)
+          while ($cudaNum -ge 0) { $cudaChannels += '-c nvidia/label/cuda-' + $cudaVersion.Remove($cudaVersion.LastIndexOf('.')+1) + $cudaNum + ' '; $cudaNum-- }
+          mamba install -y 'cuda' $cudaChannels.TrimEnd().Split()
+          python -m pip install build wheel
+          
+      - name: Build Wheel
+        run: |
+          $packageVersion = [version]$env:PCKGVER.TrimStart('v')
+          $cudaVersion = $env:CUDAVER.Remove($env:CUDAVER.LastIndexOf('.')).Replace('.','')
+          $env:CUDA_PATH = $env:CONDA_PREFIX
+          $env:CUDA_HOME = $env:CONDA_PREFIX
+          if ($IsLinux) {$env:LD_LIBRARY_PATH = $env:CONDA_PREFIX + '/lib:' + $env:LD_LIBRARY_PATH}
+          $env:VERBOSE = '1'
+          $env:FORCE_CMAKE = '1'
+          $env:CMAKE_ARGS = '-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=all'
+          if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.70') {$env:CMAKE_ARGS = '-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=35-real;37-real;52;61-real;70-real;72-real;75-real;80-real;86-real;89-real;90'}
+          if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.70' -and [version]$env:CUDAVER -ge [version]'12.0') {$env:CMAKE_ARGS = '-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=52;61-real;70-real;72-real;75-real;80-real;86-real;89-real;90'}
+          if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.70' -and [version]$env:CUDAVER -lt [version]'11.8') {$env:CMAKE_ARGS = '-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=35-real;37-real;52;61-real;70-real;72-real;75-real;80-real;86'}
+          if ($packageVersion -lt [version]'0.1.66') {$env:CUDAFLAGS = '-arch=all'}
+          if ($env:AVXVER -eq 'AVX') {$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DLLAMA_AVX2=off'}
+          if ($env:AVXVER -eq 'AVX512') {$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DLLAMA_AVX512=on'}
+          if ($env:AVXVER -eq 'basic') {$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_FMA=off -DLLAMA_F16C=off'}
+          python -m build --wheel -C--build-option=egg_info "-C--build-option=--tag-build=+cu$cudaVersion"
+          
+      - name: Upload files to a GitHub release
+        id: upload-release
+        uses: svenstaro/upload-release-action@2.6.1
+        continue-on-error: true
+        with:
+          file: ./dist/*.whl
+          tag: ${{ matrix.releasetag }}
+          file_glob: true
+          make_latest: false
+          overwrite: true
+        
+      - uses: actions/upload-artifact@v3
+        if: steps.upload-release.outcome == 'failure'
+        with:
+          name: ${{ matrix.releasetag == 'wheels' && 'AVX2' || matrix.releasetag }}
+          path: ./dist/*.whl
+          
+  build_cpu:
+    name: Build CPU-only Wheels
+    needs: build_wheels
+    uses: ./.github/workflows/build-wheels-cpu.yml
+    with:
+      version: ${{ inputs.version }}