diff --git a/.buildkite/release-pipeline.yaml b/.buildkite/release-pipeline.yaml
index a1de41652c9a..8c6ef7817aaf 100644
--- a/.buildkite/release-pipeline.yaml
+++ b/.buildkite/release-pipeline.yaml
@@ -1,24 +1,22 @@
 steps:
   # aarch64 + CUDA builds. PyTorch 2.8 aarch64 + CUDA wheel is only available on CUDA 12.9
   - label: "Build arm64 wheel - CUDA 12.9"
+    depends_on: ~
     id: build-wheel-arm64-cuda-12-9
     agents:
       queue: arm64_cpu_queue_postmerge
     commands:
       # #NOTE: torch_cuda_arch_list is derived from upstream PyTorch build files here:
       # https://github.com/pytorch/pytorch/blob/main/.ci/aarch64_linux/aarch64_ci_build.sh#L7
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.9.1 --build-arg torch_cuda_arch_list='8.7 9.0 10.0+PTX 12.0' --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
+      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.9.1 --build-arg VLLM_MAIN_CUDA_VERSION=12.9 --build-arg torch_cuda_arch_list='8.7 9.0 10.0+PTX 12.0' --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
       - "mkdir artifacts"
       - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
       - "bash .buildkite/scripts/upload-wheels.sh"
     env:
       DOCKER_BUILDKIT: "1"
 
-  - block: "Build CUDA 12.8 wheel"
-    key: block-build-cu128-wheel
-
   - label: "Build wheel - CUDA 12.8"
-    depends_on: block-build-cu128-wheel
+    depends_on: ~
     id: build-wheel-cuda-12-8
     agents:
       queue: cpu_queue_postmerge
@@ -30,12 +28,8 @@ steps:
     env:
       DOCKER_BUILDKIT: "1"
 
-  - block: "Build CUDA 12.6 wheel"
-    key: block-build-cu126-wheel
-    depends_on: ~
-
   - label: "Build wheel - CUDA 12.6"
-    depends_on: block-build-cu126-wheel
+    depends_on: ~
     id: build-wheel-cuda-12-6
     agents:
       queue: cpu_queue_postmerge
@@ -102,8 +96,6 @@ steps:
     depends_on:
       - create-multi-arch-manifest
       - build-wheel-cuda-12-8
-      - build-wheel-cuda-12-6
-      - build-wheel-cuda-12-9
     id: annotate-release-workflow
     agents:
       queue: cpu_queue_postmerge
diff --git a/.buildkite/scripts/annotate-release.sh b/.buildkite/scripts/annotate-release.sh
index 94e0ac2398f3..fde48603ad3c 100755
--- a/.buildkite/scripts/annotate-release.sh
+++ b/.buildkite/scripts/annotate-release.sh
@@ -14,18 +14,33 @@ buildkite-agent annotate --style 'info' --context 'release-workflow' << EOF
 To download the wheel:
 \`\`\`
 aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}/vllm-${RELEASE_VERSION}-cp38-abi3-manylinux1_x86_64.whl .
+aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}/vllm-${RELEASE_VERSION}-cp38-abi3-manylinux2014_aarch64.whl .
+
 aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu126/vllm-${RELEASE_VERSION}+cu126-cp38-abi3-manylinux1_x86_64.whl .
-aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu118/vllm-${RELEASE_VERSION}+cu118-cp38-abi3-manylinux1_x86_64.whl . 
+aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu129/vllm-${RELEASE_VERSION}+cu129-cp38-abi3-manylinux1_x86_64.whl .
 \`\`\`
 
 To download and upload the image:
 
 \`\`\`
-docker pull public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}
-docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT} vllm/vllm-openai
-docker tag vllm/vllm-openai vllm/vllm-openai:latest
-docker tag vllm/vllm-openai vllm/vllm-openai:v${RELEASE_VERSION}
-docker push vllm/vllm-openai:latest
-docker push vllm/vllm-openai:v${RELEASE_VERSION}
+docker pull public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}-x86_64
+docker pull public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}-aarch64
+
+docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}-x86_64 vllm/vllm-openai:x86_64
+docker tag vllm/vllm-openai:x86_64 vllm/vllm-openai:latest-x86_64
+docker tag vllm/vllm-openai:x86_64 vllm/vllm-openai:v${RELEASE_VERSION}-x86_64
+docker push vllm/vllm-openai:latest-x86_64
+docker push vllm/vllm-openai:v${RELEASE_VERSION}-x86_64
+
+docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}-aarch64 vllm/vllm-openai:aarch64
+docker tag vllm/vllm-openai:aarch64 vllm/vllm-openai:latest-aarch64
+docker tag vllm/vllm-openai:aarch64 vllm/vllm-openai:v${RELEASE_VERSION}-aarch64
+docker push vllm/vllm-openai:latest-aarch64
+docker push vllm/vllm-openai:v${RELEASE_VERSION}-aarch64
+
+docker manifest create vllm/vllm-openai:latest vllm/vllm-openai:latest-x86_64 vllm/vllm-openai:latest-aarch64 --amend
+docker manifest create vllm/vllm-openai:v${RELEASE_VERSION} vllm/vllm-openai:v${RELEASE_VERSION}-x86_64 vllm/vllm-openai:v${RELEASE_VERSION}-aarch64 --amend
+docker manifest push vllm/vllm-openai:latest
+docker manifest push vllm/vllm-openai:v${RELEASE_VERSION}
 \`\`\`
 EOF 
\ No newline at end of file
diff --git a/docker/Dockerfile b/docker/Dockerfile
index 307e9658f717..d4761e84f28e 100644
--- a/docker/Dockerfile
+++ b/docker/Dockerfile
@@ -196,6 +196,7 @@ ARG SCCACHE_S3_NO_CREDENTIALS=0
 
 # Flag to control whether to use pre-built vLLM wheels
 ARG VLLM_USE_PRECOMPILED=""
+ARG VLLM_MAIN_CUDA_VERSION=""
 
 # if USE_SCCACHE is set, use sccache to speed up compilation
 RUN --mount=type=cache,target=/root/.cache/uv \
@@ -213,6 +214,7 @@ RUN --mount=type=cache,target=/root/.cache/uv \
         && export SCCACHE_IDLE_TIMEOUT=0 \
         && export CMAKE_BUILD_TYPE=Release \
         && export VLLM_USE_PRECOMPILED="${VLLM_USE_PRECOMPILED}" \
+        && export VLLM_MAIN_CUDA_VERSION="${VLLM_MAIN_CUDA_VERSION}" \
         && export VLLM_DOCKER_BUILD_CONTEXT=1 \
         && sccache --show-stats \
         && python3 setup.py bdist_wheel --dist-dir=dist --py-limited-api=cp38 \
diff --git a/setup.py b/setup.py
index eb313b7d219c..67f65d9b9f56 100644
--- a/setup.py
+++ b/setup.py
@@ -56,8 +56,6 @@ def load_module_from_path(module_name, path):
     # fallback to cpu
     VLLM_TARGET_DEVICE = "cpu"
 
-MAIN_CUDA_VERSION = "12.8"
-
 
 def is_sccache_available() -> bool:
     return which("sccache") is not None and \
@@ -507,7 +505,7 @@ def get_vllm_version() -> str:
             version += f"{sep}precompiled"
         else:
             cuda_version = str(get_nvcc_cuda_version())
-            if cuda_version != MAIN_CUDA_VERSION:
+            if cuda_version != envs.VLLM_MAIN_CUDA_VERSION:
                 cuda_version_str = cuda_version.replace(".", "")[:3]
                 # skip this for source tarball, required for pypi
                 if "sdist" not in sys.argv:
@@ -515,7 +513,7 @@ def get_vllm_version() -> str:
     elif _is_hip():
         # Get the Rocm Version
         rocm_version = get_rocm_version() or torch.version.hip
-        if rocm_version and rocm_version != MAIN_CUDA_VERSION:
+        if rocm_version and rocm_version != envs.VLLM_MAIN_CUDA_VERSION:
             version += f"{sep}rocm{rocm_version.replace('.', '')[:3]}"
     elif _is_tpu():
         version += f"{sep}tpu"
diff --git a/vllm/envs.py b/vllm/envs.py
index bb10c7cc2ac2..10ac28413bbc 100755
--- a/vllm/envs.py
+++ b/vllm/envs.py
@@ -70,6 +70,7 @@
     VLLM_VIDEO_LOADER_BACKEND: str = "opencv"
     VLLM_MM_INPUT_CACHE_GIB: int = 4
     VLLM_TARGET_DEVICE: str = "cuda"
+    VLLM_MAIN_CUDA_VERSION: str = "12.8"
     MAX_JOBS: Optional[str] = None
     NVCC_THREADS: Optional[str] = None
     VLLM_USE_PRECOMPILED: bool = False
@@ -247,6 +248,11 @@ def get_vllm_port() -> Optional[int]:
     "VLLM_TARGET_DEVICE":
     lambda: os.getenv("VLLM_TARGET_DEVICE", "cuda").lower(),
 
+    # Main CUDA version of vLLM, supporting [12.6, 12.8, 12.9],
+    # 12.8 is the default. This follows PyTorch but can be overridden.
+    "VLLM_MAIN_CUDA_VERSION":
+    lambda: os.getenv("VLLM_MAIN_CUDA_VERSION", "").lower() or "12.8",
+
     # Maximum number of compilation jobs to run in parallel.
     # By default this is the number of CPUs
     "MAX_JOBS":