diff --git a/.github/workflows/release.yaml b/.github/workflows/release.yaml
index 50177050c..c6ab84fe0 100644
--- a/.github/workflows/release.yaml
+++ b/.github/workflows/release.yaml
@@ -478,243 +478,77 @@ jobs:
             dist/OllamaSetup.exe
             dist/ollama-windows-*.zip
 
-  # Linux x86 assets built using the container based build
-  build-linux-amd64:
+  build-linux:
     environment: release
     runs-on: linux
-    env:
-      PLATFORM: linux/amd64
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: Set Version
-        shell: bash
-        run: echo "VERSION=${GITHUB_REF_NAME#v}" >> $GITHUB_ENV
-      - run: |
-          ./scripts/build_linux.sh
-      - uses: actions/upload-artifact@v4
-        with:
-          name: dist-linux-amd64
-          path: |
-            dist/*linux*
-            !dist/*-cov
-
-  # Linux ARM assets built using the container based build
-  # (at present, docker isn't pre-installed on arm ubunutu images)
-  build-linux-arm64:
-    environment: release
-    runs-on: linux-arm64
-    env:
-      PLATFORM: linux/arm64
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: Set Version
-        shell: bash
-        run: echo "VERSION=${GITHUB_REF_NAME#v}" >> $GITHUB_ENV
-      - name: 'Install Docker'
-        run: |
-          # Add Docker's official GPG key:
-          env
-          uname -a
-          sudo apt-get update
-          sudo apt-get install -y ca-certificates curl
-          sudo install -m 0755 -d /etc/apt/keyrings
-          sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
-          sudo chmod a+r /etc/apt/keyrings/docker.asc
-
-          # Add the repository to Apt sources:
-          echo \
-            "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \
-            $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
-            sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
-          sudo apt-get update
-          sudo apt-get install -y docker-ce docker-ce-cli containerd.io
-          sudo usermod -aG docker $USER
-          sudo apt-get install acl
-          sudo setfacl --modify user:$USER:rw /var/run/docker.sock
-      - run: |
-          ./scripts/build_linux.sh
-      - uses: actions/upload-artifact@v4
-        with:
-          name: dist-linux-arm64
-          path: |
-            dist/*linux*
-            !dist/*-cov
-
-  # Container image build
-  build-container-image:
-    environment: release
     strategy:
       matrix:
-        runner:
-          - linux
-          - linux-arm64
-    runs-on: ${{ matrix.runner }}
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
+        include:
+          - os: linux
+            arch: amd64
+            targets: [archive, rocm]
+          - os: linux
+            arch: arm64
+            targets: [archive]
     steps:
       - uses: actions/checkout@v4
+      - uses: docker/setup-qemu-action@v3
+      - uses: docker/setup-buildx-action@v3
+      - run: |
+          apt-get update && apt-get install pigz
+          for TARGET in ${{ matrix.targets }}; do docker buildx build --platform $PLATFORM --target $TARGET --output type=local,dest=dist/$PLATFORM .; done
+          tar c -C dist/$PLATFORM . | pigz -9cv >dist/ollama-${PLATFORM//\//-}.tar.gz
+        env:
+          PLATFORM: ${{ matrix.os }}/${{ matrix.arch }}
+      - uses: actions/upload-artifact@v4
         with:
-          submodules: recursive
-      - name: 'Install Docker'
-        if: ${{ startsWith(matrix.runner, 'linux-arm64') }}
-        run: |
-          sudo apt-get update
-          sudo apt-get install -y ca-certificates curl
-          sudo install -m 0755 -d /etc/apt/keyrings
-          sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
-          sudo chmod a+r /etc/apt/keyrings/docker.asc
-          echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.asc] https://download.docker.com/linux/ubuntu \
-            $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
-            sudo tee /etc/apt/sources.list.d/docker.list > /dev/null
-          sudo apt-get update
-          sudo apt-get install -y docker-ce docker-ce-cli containerd.io
-          sudo usermod -aG docker $USER
-          sudo apt-get install acl
-          sudo setfacl --modify user:$USER:rw /var/run/docker.sock
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
-          tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
-            type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          machine=$(uname -m)
-          case ${machine} in
-            x86_64) echo ARCH=amd64; echo PLATFORM_PAIR=linux-amd64 ;;
-            aarch64) echo ARCH=arm64; echo PLATFORM_PAIR=linux-arm64 ;;
-          esac >>$GITHUB_ENV
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
-        with:
-          username: ${{ vars.DOCKER_USER }}
-          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Build and push by digest
-        id: build
-        uses: docker/build-push-action@v6
-        with:
-          context: "."
-          platforms: linux/${{ env.ARCH }}
-          build-args: |
-            GOFLAGS
-          outputs: type=image,name=${{ env.FINAL_IMAGE_REPO }},push-by-digest=true,name-canonical=true,push=true
-      - name: Export digest
-        run: |
-          mkdir -p /tmp/digests
-          digest="${{ steps.build.outputs.digest }}"
-          touch "/tmp/digests/${digest#sha256:}"
-      - name: Upload digest
-        uses: actions/upload-artifact@v4
-        with:
-          name: digests-${{ env.PLATFORM_PAIR }}
-          path: /tmp/digests/*
-          if-no-files-found: error
-          retention-days: 1
-  merge:
+          name: dist-${{ matrix.os }}-${{ matrix.arch }}
+          path: |
+            dist/ollama-${{ matrix.os }}-${{ matrix.arch }}.tar.gz
+
+  build-docker:
     environment: release
-    runs-on: linux
-    needs:
-      - build-container-image
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        include:
+          - flavor: |
+              latest=auto
+            platforms: linux/amd64,linux/arm64
+            build-args: [GOFLAGS]
+          - flavor: |
+              suffix=-rocm,onlatest=false
+            platforms: linux/amd64
+            build-args: [GOFLAGS, FLAVOR=rocm]
     steps:
       - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - name: Download digests
-        uses: actions/download-artifact@v4
-        with:
-          path: /tmp/digests
-          pattern: digests-*
-          merge-multiple: true
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
-          tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
-            type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          machine=$(uname -m)
-          case ${machine} in
-            x86_64) echo ARCH=amd64; echo PLATFORM_PAIR=linux-amd64 ;;
-            aarch64) echo ARCH=arm64; echo PLATFORM_PAIR=linux-arm64 ;;
-          esac >>$GITHUB_ENV
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
+      - uses: docker/setup-qemu-action@v2
+      - uses: docker/setup-buildx-action@v2
+      - uses: docker/login-action@v3
         with:
           username: ${{ vars.DOCKER_USER }}
           password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Create manifest list and push
-        working-directory: /tmp/digests
-        run: |
-          docker buildx imagetools create $(jq -cr '.tags | map("-t " + .) | join(" ")' <<< "$DOCKER_METADATA_OUTPUT_JSON") \
-            $(printf '${{ env.FINAL_IMAGE_REPO }}@sha256:%s ' *)
-      - name: Inspect image
-        run: |
-          docker buildx imagetools inspect ${{ env.FINAL_IMAGE_REPO }}:${{ steps.meta.outputs.version }}          
-  build-container-image-rocm:
-    environment: release
-    runs-on: linux
-    env:
-      FINAL_IMAGE_REPO: ollama/ollama
-      ARCH: amd64
-      PLATFORM_PAIR: linux-amd64
-    steps:
-      - uses: actions/checkout@v4
+      - id: metadata
+        uses: docker/metadata-action@v4
         with:
-          submodules: recursive
-      - name: Docker meta
-        id: meta
-        uses: docker/metadata-action@v5
-        with:
-          images: ${{ env.FINAL_IMAGE_REPO }}
-          flavor: |
-            latest=false
+          flavor: ${{ matrix.flavor }}
+          images: |
+            ollama/ollama
           tags: |
-            type=ref,enable=true,priority=600,prefix=0.0.0-pr,suffix=,event=pr
             type=semver,pattern={{version}}
-      - name: Set Version
-        shell: bash
-        run: |
-          echo GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ env.DOCKER_METADATA_OUTPUT_VERSION }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'" >>$GITHUB_ENV
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Login to Docker Hub
-        uses: docker/login-action@v3
+      - uses: docker/build-push-action@v6
         with:
-          username: ${{ vars.DOCKER_USER }}
-          password: ${{ secrets.DOCKER_ACCESS_TOKEN }}
-      - name: Build and push by digest
-        id: build
-        uses: docker/build-push-action@v6
-        with:
-          context: "."
-          target: runtime-rocm
-          build-args: |
-            GOFLAGS
-          tags: ${{ env.FINAL_IMAGE_REPO }}:${{ env.DOCKER_METADATA_OUTPUT_VERSION}}-rocm
+          context: .
           push: true
+          platforms: ${{ matrix.platforms }}
+          build-args: ${{ matrix.build-args }}
+          tags: ${{ steps.metadata.outputs.tags }}
+          labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: type=registry,ref=ollama/ollama:latest
+          cache-to: type=inline
+          provenance: false
+        env:
+          GOFLAGS="'-ldflags=-w -s \"-X=github.com/ollama/ollama/version.Version=${{ steps.metadata.outputs.version }}\" \"-X=github.com/ollama/ollama/server.mode=release\"'"
 
   # Aggregate all the assets and ship a release
   release:
diff --git a/CMakePresets.json b/CMakePresets.json
index f1b6d41b4..2bda4eb10 100644
--- a/CMakePresets.json
+++ b/CMakePresets.json
@@ -3,7 +3,7 @@
   "configurePresets": [
     {
       "name": "Default",
-      "binaryDir": "${sourceDir}/build",
+      "binaryDir": "${sourceDir}/dist/build",
       "cacheVariables": {
         "CMAKE_BUILD_TYPE": "Release"
       }
diff --git a/Dockerfile b/Dockerfile
index 47228df61..eb741fdf5 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,201 +1,161 @@
-ARG GOLANG_VERSION=1.22.8
-ARG CUDA_VERSION_11=11.3.1
-ARG CUDA_VERSION_12=12.4.0
-ARG ROCM_VERSION=6.1.2
-ARG JETPACK_6=r36.2.0
-ARG JETPACK_5=r35.4.1
+# vim: filetype=dockerfile
 
-### To create a local image for building linux binaries on mac or windows with efficient incremental builds
-#
-# docker build --platform linux/amd64 -t builder-amd64 -f Dockerfile --target unified-builder-amd64 .
-# docker run --platform linux/amd64 --rm -it -v $(pwd):/go/src/github.com/ollama/ollama/ builder-amd64
-#
-### Then incremental builds will be much faster in this container
-#
-# make -j 10 dist
-#
-FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS unified-builder-amd64
-ARG GOLANG_VERSION
-ARG CUDA_VERSION_11
-ARG CUDA_VERSION_12
-COPY ./scripts/rh_linux_deps.sh /
-ENV PATH /opt/rh/devtoolset-10/root/usr/bin:/usr/local/cuda/bin:$PATH
-ENV LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64
-RUN GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
-RUN yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-rhel7.repo && \
-    dnf clean all && \
-    dnf install -y \
-    zsh \
-    cuda-toolkit-$(echo ${CUDA_VERSION_11} | cut -f1-2 -d. | sed -e "s/\./-/g") \
-    cuda-toolkit-$(echo ${CUDA_VERSION_12} | cut -f1-2 -d. | sed -e "s/\./-/g")
-# TODO intel oneapi goes here...
-ENV GOARCH amd64
-ENV CGO_ENABLED 1
-WORKDIR /go/src/github.com/ollama/ollama/
-ENTRYPOINT [ "zsh" ]
+ARG FLAVOR=${TARGETARCH}
 
-### To create a local image for building linux binaries on mac or linux/arm64 with efficient incremental builds
-# Note: this does not contain jetson variants
-#
-# docker build --platform linux/arm64 -t builder-arm64 -f Dockerfile --target unified-builder-arm64 .
-# docker run --platform linux/arm64 --rm -it -v $(pwd):/go/src/github.com/ollama/ollama/ builder-arm64
-#
-FROM --platform=linux/arm64 rockylinux:8 AS unified-builder-arm64
-ARG GOLANG_VERSION
-ARG CUDA_VERSION_11
-ARG CUDA_VERSION_12
-COPY ./scripts/rh_linux_deps.sh /
-RUN GOLANG_VERSION=${GOLANG_VERSION} sh /rh_linux_deps.sh
-RUN yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/sbsa/cuda-rhel8.repo && \
-    dnf config-manager --set-enabled appstream && \
-    dnf clean all && \
-    dnf install -y \
-    zsh \
-    cuda-toolkit-$(echo ${CUDA_VERSION_11} | cut -f1-2 -d. | sed -e "s/\./-/g") \
-    cuda-toolkit-$(echo ${CUDA_VERSION_12} | cut -f1-2 -d. | sed -e "s/\./-/g")
-ENV PATH /opt/rh/gcc-toolset-10/root/usr/bin:$PATH:/usr/local/cuda/bin
-ENV LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64
-ENV LIBRARY_PATH=/usr/local/cuda/lib64/stubs:/opt/amdgpu/lib64
-ENV GOARCH arm64
-ENV CGO_ENABLED 1
-WORKDIR /go/src/github.com/ollama/ollama/
-ENTRYPOINT [ "zsh" ]
+ARG ROCMVERSION=6.1.2
+ARG JETPACK5VERSION=r35.4.1
+ARG JETPACK6VERSION=r36.2.0
+ARG CMAKEVERSION=3.31.2
 
-FROM --platform=linux/amd64 unified-builder-amd64 AS build-amd64
-COPY . .
-ARG OLLAMA_SKIP_CUDA_GENERATE
-ARG OLLAMA_SKIP_ROCM_GENERATE
-ARG OLLAMA_FAST_BUILD
-ARG VERSION
-ARG CUSTOM_CPU_FLAGS
+FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCMVERSION}-complete AS base-amd64
+RUN sed -i -e 's/mirror.centos.org/vault.centos.org/g' -e 's/^#.*baseurl=http/baseurl=http/g' -e 's/^mirrorlist=http/#mirrorlist=http/g' /etc/yum.repos.d/*.repo \
+    && yum install -y yum-utils devtoolset-10-gcc devtoolset-10-gcc-c++ \
+    && yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-rhel7.repo \
+    && curl -s -L https://github.com/ccache/ccache/releases/download/v4.10.2/ccache-4.10.2-linux-x86_64.tar.xz | tar -Jx -C /usr/local/bin --strip-components 1
+ENV PATH=/opt/rh/devtoolset-10/root/usr/bin:/opt/rh/devtoolset-11/root/usr/bin:$PATH
+
+FROM --platform=linux/arm64 rockylinux:8 AS base-arm64
+# install epel-release for ccache
+RUN yum install -y yum-utils epel-release \
+    && yum install -y clang ccache \
+    && yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/sbsa/cuda-rhel8.repo
+ENV CC=clang CXX=clang++
+
+FROM base-${TARGETARCH} AS base
+ARG CMAKEVERSION
+RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
+COPY CMakeLists.txt CMakePresets.json .
+COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
+ENV LDFLAGS=-s
+
+FROM base AS cpu
+# amd64 uses gcc which requires devtoolset-11 for AVX extensions while arm64 uses clang
+RUN if [ "$(uname -m)" = "x86_64" ]; then yum install -y devtoolset-11-gcc devtoolset-11-gcc-c++; fi
+ENV PATH=/opt/rh/devtoolset-11/root/usr/bin:$PATH
 RUN --mount=type=cache,target=/root/.ccache \
-    if grep "^flags" /proc/cpuinfo|grep avx>/dev/null; then \
-        make -j $(nproc) dist ; \
-    else \
-        make -j 5 dist ; \
-    fi
-RUN cd dist/linux-$GOARCH && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH.tgz
-RUN if [ -z ${OLLAMA_SKIP_ROCM_GENERATE} ] ; then \
-    cd dist/linux-$GOARCH-rocm && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH-rocm.tgz ;\
-    fi
+    cmake --preset 'Default' && cmake --build --parallel --preset 'Default'
 
-# Jetsons need to be built in discrete stages
-FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK_5} AS runners-jetpack5-arm64
-ARG GOLANG_VERSION
-RUN apt-get update && apt-get install -y git curl ccache && \
-    curl -s -L https://dl.google.com/go/go${GOLANG_VERSION}.linux-arm64.tar.gz | tar xz -C /usr/local && \
-    ln -s /usr/local/go/bin/go /usr/local/bin/go && \
-    ln -s /usr/local/go/bin/gofmt /usr/local/bin/gofmt && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-WORKDIR /go/src/github.com/ollama/ollama/
-COPY . .
-ARG CGO_CFLAGS
-ENV GOARCH arm64
-ARG VERSION
+FROM base AS cuda-11
+ARG CUDA11VERSION=11.3
+RUN yum install -y cuda-toolkit-${CUDA11VERSION//./-}
+ENV PATH=/usr/local/cuda-11/bin:$PATH
 RUN --mount=type=cache,target=/root/.ccache \
-    make -j 5 dist_cuda_v11 \
-        CUDA_ARCHITECTURES="72;87" \
-        GPU_RUNNER_VARIANT=_jetpack5 \
-        DIST_LIB_DIR=/go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack5/lib/ollama \
-        DIST_GPU_RUNNER_DEPS_DIR=/go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack5/lib/ollama/cuda_jetpack5
+    cmake --preset 'CUDA 11' && cmake --build --parallel --preset 'CUDA 11'
 
-FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK_6} AS runners-jetpack6-arm64
-ARG GOLANG_VERSION
-RUN apt-get update && apt-get install -y git curl ccache && \
-    curl -s -L https://dl.google.com/go/go${GOLANG_VERSION}.linux-arm64.tar.gz | tar xz -C /usr/local && \
-    ln -s /usr/local/go/bin/go /usr/local/bin/go && \
-    ln -s /usr/local/go/bin/gofmt /usr/local/bin/gofmt && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-WORKDIR /go/src/github.com/ollama/ollama/
-COPY . .
-ARG CGO_CFLAGS
-ENV GOARCH arm64
-ARG VERSION
+FROM base AS cuda-12
+ARG CUDA12VERSION=12.4
+RUN yum install -y cuda-toolkit-${CUDA12VERSION//./-}
+ENV PATH=/usr/local/cuda-12/bin:$PATH
 RUN --mount=type=cache,target=/root/.ccache \
-    make -j 5 dist_cuda_v12 \
-        CUDA_ARCHITECTURES="87" \
-        GPU_RUNNER_VARIANT=_jetpack6 \
-        DIST_LIB_DIR=/go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack6/lib/ollama \
-        DIST_GPU_RUNNER_DEPS_DIR=/go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack6/lib/ollama/cuda_jetpack6
+    cmake --preset 'CUDA 12' && cmake --build --parallel --preset 'CUDA 12'
 
-FROM --platform=linux/arm64 unified-builder-arm64 AS build-arm64
-COPY . .
-ARG OLLAMA_SKIP_CUDA_GENERATE
-ARG OLLAMA_FAST_BUILD
-ARG VERSION
+FROM base AS rocm-6
 RUN --mount=type=cache,target=/root/.ccache \
-    make -j 5 dist
-COPY --from=runners-jetpack5-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
-COPY --from=runners-jetpack6-arm64 /go/src/github.com/ollama/ollama/dist/ dist/
-RUN cd dist/linux-$GOARCH && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH.tgz
-RUN cd dist/linux-$GOARCH-jetpack5 && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH-jetpack5.tgz
-RUN cd dist/linux-$GOARCH-jetpack6 && \
-    tar -cf - . | pigz --best > ../ollama-linux-$GOARCH-jetpack6.tgz
+    cmake --preset 'ROCm 6' && cmake --build --parallel --preset 'ROCm 6'
 
-FROM --platform=linux/amd64 scratch AS dist-amd64
-COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/ollama-linux-*.tgz /
-FROM --platform=linux/arm64 scratch AS dist-arm64
-COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/ollama-linux-*.tgz /
-FROM dist-$TARGETARCH AS dist
+FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK5VERSION} AS jetpack-5
+ARG CMAKEVERSION
+RUN apt-get update && apt-get install -y curl ccache \
+    && curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
+COPY CMakeLists.txt CMakePresets.json .
+COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
+RUN --mount=type=cache,target=/root/.ccache \
+    cmake --preset 'JetPack 5' && cmake --build --parallel --preset 'JetPack 5'
 
+FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK6VERSION} AS jetpack-6
+ARG CMAKEVERSION
+RUN apt-get update && apt-get install -y curl ccache \
+    && curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKEVERSION}/cmake-${CMAKEVERSION}-linux-$(uname -m).tar.gz | tar xz -C /usr/local --strip-components 1
+COPY CMakeLists.txt CMakePresets.json .
+COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
+RUN --mount=type=cache,target=/root/.ccache \
+    cmake --preset 'JetPack 6' && cmake --build --parallel --preset 'JetPack 6'
 
-# For amd64 container images, filter out cuda/rocm to minimize size
-FROM build-amd64 AS runners-cuda-amd64
-RUN rm -rf \
-    ./dist/linux-amd64/lib/ollama/libggml_hipblas.so \
-    ./dist/linux-amd64/lib/ollama/runners/rocm*
+FROM base AS build
+ARG GOVERSION=1.23.4
+RUN curl -fsSL https://golang.org/dl/go${GOVERSION}.linux-$(case $(uname -m) in x86_64) echo amd64 ;; aarch64) echo arm64 ;; esac).tar.gz | tar xz -C /usr/local
+ENV PATH=/usr/local/go/bin:$PATH
+WORKDIR /go/src/github.com/ollama/ollama
+COPY . .
+ARG GOFLAGS="'-ldflags=-w -s'"
+ENV CGO_ENABLED=1
+RUN --mount=type=cache,target=/root/.cache/go-build \
+    go build -trimpath -buildmode=pie -o /bin/ollama .
 
-FROM build-amd64 AS runners-rocm-amd64
-RUN rm -rf \
-    ./dist/linux-amd64/lib/ollama/libggml_cuda*.so \
-    ./dist/linux-amd64/lib/ollama/libcu*.so* \
-    ./dist/linux-amd64/lib/ollama/runners/cuda*
+FROM --platform=linux/amd64 scratch AS amd64
+COPY --from=cuda-11 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.11 \
+    /usr/local/cuda/lib64/libcublasLt.so.11 \
+    /usr/local/cuda/lib64/libcudart.so.11.0 \
+    /lib/ollama/cuda_v11/
+COPY --from=cuda-12 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.12 \
+    /usr/local/cuda/lib64/libcublasLt.so.12 \
+    /usr/local/cuda/lib64/libcudart.so.12 \
+    /lib/ollama/cuda_v12/
 
-FROM --platform=linux/amd64 ubuntu:22.04 AS runtime-amd64
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
-COPY --from=runners-cuda-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
+FROM --platform=linux/arm64 scratch AS arm64
+COPY --from=cuda-11 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.11 \
+    /usr/local/cuda/lib64/libcublasLt.so.11 \
+    /usr/local/cuda/lib64/libcudart.so.11.0 \
+    /lib/ollama/cuda_v11/
+COPY --from=cuda-12 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.12 \
+    /usr/local/cuda/lib64/libcublasLt.so.12 \
+    /usr/local/cuda/lib64/libcudart.so.12 \
+    /lib/ollama/cuda_v12/
+COPY --from=jetpack-5 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.11 \
+    /usr/local/cuda/lib64/libcublasLt.so.11 \
+    /usr/local/cuda/lib64/libcudart.so.11.0 \
+    /lib/ollama/cuda_jetpack5/
+COPY --from=jetpack-6 --chmod=644 \
+    dist/build/lib/libggml-cuda.so \
+    /usr/local/cuda/lib64/libcublas.so.12 \
+    /usr/local/cuda/lib64/libcublasLt.so.12 \
+    /usr/local/cuda/lib64/libcudart.so.12 \
+    /lib/ollama/cuda_jetpack6/
 
-FROM --platform=linux/arm64 ubuntu:22.04 AS runtime-arm64
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/bin/ /bin/
-COPY --from=build-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64/lib/ /lib/
-COPY --from=runners-jetpack5-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack5/lib/ /lib/
-COPY --from=runners-jetpack6-arm64 /go/src/github.com/ollama/ollama/dist/linux-arm64-jetpack6/lib/ /lib/
+FROM --platform=linux/arm64 scratch AS rocm
+COPY --from=rocm-6 --chmod=644 \
+    dist/build/lib/libggml-hip.so \
+    /opt/rocm/lib/libamdhip64.so.6 \
+    /opt/rocm/lib/libhipblas.so.2 \
+    /opt/rocm/lib/librocblas.so.4 \
+    /opt/rocm/lib/libamd_comgr.so.2 \
+    /opt/rocm/lib/libhsa-runtime64.so.1 \
+    /opt/rocm/lib/librocprofiler-register.so.0 \
+    /opt/amdgpu/lib64/libdrm_amdgpu.so.1 \
+    /opt/amdgpu/lib64/libdrm.so.2 \
+    /usr/lib64/libnuma.so.1 \
+    /lib/ollama/rocm/
+COPY --from=rocm-6 /opt/rocm/lib/rocblas/ /lib/ollama/rocm/rocblas/
 
+FROM ${FLAVOR} AS archive
+COPY --from=cpu --chmod=644 \
+    dist/build/lib/libggml-base.so \
+    dist/build/lib/libggml-cpu-*.so \
+    /lib/ollama/
+COPY --from=build /bin/ollama /bin/ollama
 
-# ROCm libraries larger so we keep it distinct from the CPU/CUDA image
-FROM --platform=linux/amd64 ubuntu:22.04 AS runtime-rocm
-# Frontload the rocm libraries which are large, and rarely change to increase chance of a common layer
-# across releases
-COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64-rocm/lib/ /lib/
-RUN apt-get update && \
-    apt-get install -y ca-certificates && \
-    apt-get clean && rm -rf /var/lib/apt/lists/*
-COPY --from=build-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/bin/ /bin/
-COPY --from=runners-rocm-amd64 /go/src/github.com/ollama/ollama/dist/linux-amd64/lib/ /lib/
-
-EXPOSE 11434
-ENV OLLAMA_HOST 0.0.0.0
-
-ENTRYPOINT ["/bin/ollama"]
-CMD ["serve"]
-
-FROM runtime-$TARGETARCH
-EXPOSE 11434
-ENV OLLAMA_HOST 0.0.0.0
+FROM ubuntu:20.04
+RUN apt-get update \
+    && apt-get install -y ca-certificates \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+COPY --from=archive /bin/ /usr/bin/
 ENV PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
-ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
+COPY --from=archive /lib/ollama/ /usr/lib/ollama/
+ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/lib/ollama
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility
 ENV NVIDIA_VISIBLE_DEVICES=all
-
+ENV OLLAMA_HOST=0.0.0.0:11434
+EXPOSE 11434
 ENTRYPOINT ["/bin/ollama"]
 CMD ["serve"]
diff --git a/Dockerfile2 b/Dockerfile2
deleted file mode 100644
index e1008c1b9..000000000
--- a/Dockerfile2
+++ /dev/null
@@ -1,59 +0,0 @@
-ARG CUDA_11_VERSION=11.3
-ARG CUDA_12_VERSION=12.4
-ARG ROCM_VERSION=6.1.2
-ARG JETPACK_5_VERSION=r35.4.1
-ARG JETPACK_6_VERSION=r36.2.0
-ARG CMAKE_VERSION=3.31.2
-
-FROM --platform=linux/amd64 rocm/dev-centos-7:${ROCM_VERSION}-complete AS base
-ARG CMAKE_VERSION
-RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-x86_64.tar.gz | tar xz -C /usr --strip-components 1
-RUN sed -i -e 's/mirror.centos.org/vault.centos.org/g' -e 's/^#.*baseurl=http/baseurl=http/g' -e 's/^mirrorlist=http/#mirrorlist=http/g' /etc/yum.repos.d/*.repo \
-    && yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel7/x86_64/cuda-rhel7.repo
-
-# FROM --platform=linux/arm64 rockylinux:8 AS base
-# ARG CMAKE_VERSION
-# RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-aarch64.tar.gz | tar xz -C /usr --strip-components 1
-# RUN yum-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/sbsa/cuda-rhel8.repo
-
-FROM base AS amd64
-ARG CUDA_11_VERSION
-ARG CUDA_12_VERSION
-RUN yum install -y cuda-toolkit-${CUDA_11_VERSION//./-} \
-    && yum install -y cuda-toolkit-${CUDA_12_VERSION//./-}
-COPY CMakeLists.txt CMakeLists.txt
-COPY ml/backend/ggml/ggml ml/backend/ggml/ggml
-
-FROM --platform=linux/amd64 amd64 AS cuda_11
-ENV PATH=/usr/local/cuda-${CUDA_11_VERSION}/bin:$PATH
-RUN cmake --build --parallel --preset 'CUDA 11'
-
-FROM --platform=linux/amd64 amd64 AS cuda_12
-ENV PATH=/usr/local/cuda-${CUDA_12_VERSION}/bin:$PATH
-RUN cmake --build --parallel --preset 'CUDA 11'
-
-FROM --platform=linux/amd64 amd64 AS rocm
-RUN cmake --build --parallel --preset 'ROCm 6'
-
-FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK_5_VERSION} AS jetpack_5
-ARG CMAKE_VERSION
-RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-aarch64.tar.gz | tar xz -C /usr --strip-components 1
-COPY CMakeLists.txt .
-COPY ml/backend/ggml/ggml .
-RUN cmake --build --parallel --preset 'JetPack 5'
-
-FROM --platform=linux/arm64 nvcr.io/nvidia/l4t-jetpack:${JETPACK_6_VERSION} AS jetpack_6
-ARG CMAKE_VERSION
-RUN curl -fsSL https://github.com/Kitware/CMake/releases/download/v${CMAKE_VERSION}/cmake-${CMAKE_VERSION}-linux-aarch64.tar.gz | tar xz -C /usr --strip-components 1
-COPY CMakeLists.txt .
-COPY ml/backend/ggml/ggml .
-RUN cmake --build --parallel --preset 'JetPack 6'
-
-FROM --platform=linux/amd64 golang:1.23
-COPY --from=cuda_11 build/libggml-cuda.so libggml-cuda-11.so
-COPY --from=cuda_12 build/libggml-cuda.so libggml-cuda-12.so
-COPY --from=rocm build/libggml-hip.so libggml-hip.so
-
-# FROM --platform=linux/arm64 golang:1.23
-# COPY --from=jetpack_5 build/libggml-cuda.so libggml-cuda-jetpack-5.so
-# COPY --from=jetpack_6 build/libggml-cuda.so libggml-cuda-jetpack-6.so
diff --git a/scripts/build_linux.sh b/scripts/build_linux.sh
index 894d9dd2e..a0c3d2f00 100755
--- a/scripts/build_linux.sh
+++ b/scripts/build_linux.sh
@@ -18,7 +18,7 @@ docker buildx build \
         --output type=local,dest=./dist/ \
         --platform=${PLATFORM} \
         ${OLLAMA_COMMON_BUILD_ARGS} \
-        --target dist \
+        --target archive \
         -f Dockerfile \
         .
 
@@ -26,4 +26,4 @@ docker buildx build \
 if echo $PLATFORM | grep "," > /dev/null ; then 
         mv -f ./dist/linux_*64/ollama* ./dist/
         rmdir ./dist/linux_*64
-fi
\ No newline at end of file
+fi