diff --git a/pytorch/training/buildspec-2-9-ec2.yml b/pytorch/training/buildspec-2-9-ec2.yml
new file mode 100644
index 000000000000..3f2cef8d599f
--- /dev/null
+++ b/pytorch/training/buildspec-2-9-ec2.yml
@@ -0,0 +1,75 @@
+account_id: &ACCOUNT_ID <set-$ACCOUNT_ID-in-environment>
+prod_account_id: &PROD_ACCOUNT_ID 763104351884
+region: &REGION <set-$REGION-in-environment>
+framework: &FRAMEWORK pytorch
+version: &VERSION 2.9.0
+short_version: &SHORT_VERSION "2.9"
+arch_type: x86
+# autopatch_build: "True"
+
+repository_info:
+  training_repository: &TRAINING_REPOSITORY
+    image_type: &TRAINING_IMAGE_TYPE training
+    root: !join [ *FRAMEWORK, "/", *TRAINING_IMAGE_TYPE ]
+    repository_name: &REPOSITORY_NAME !join [ pr, "-", *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE ]
+    repository: &REPOSITORY !join [ *ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/, *REPOSITORY_NAME ]
+    release_repository_name: &RELEASE_REPOSITORY_NAME !join [ *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE ]
+    release_repository: &RELEASE_REPOSITORY !join [ *PROD_ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/, *RELEASE_REPOSITORY_NAME ]
+
+context:
+  training_context: &TRAINING_CONTEXT
+    start_cuda_compat:
+      source: docker/build_artifacts/start_cuda_compat.sh
+      target: start_cuda_compat.sh
+    dockerd_entrypoint:
+      source: docker/build_artifacts/dockerd_entrypoint.sh
+      target: dockerd_entrypoint.sh
+    changehostname:
+      source: docker/build_artifacts/changehostname.c
+      target: changehostname.c
+    start_with_right_hostname:
+      source: docker/build_artifacts/start_with_right_hostname.sh
+      target: start_with_right_hostname.sh
+    example_mnist_file:
+      source: docker/build_artifacts/mnist.py
+      target: mnist.py
+    deep_learning_container:
+      source: ../../src/deep_learning_container.py
+      target: deep_learning_container.py
+    setup_oss_compliance:
+      source: ../../scripts/setup_oss_compliance.sh
+      target: setup_oss_compliance.sh
+
+images:
+  BuildEC2CPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_CPU_TRAINING_PY3 false
+    image_size_baseline: 7200
+    device_type: &DEVICE_TYPE cpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-ec2" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-ec2" ]
+    # skip_build: "False"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    target: ec2
+    context:
+      <<: *TRAINING_CONTEXT
+  BuildEC2GPUPTTrainPy3cu130DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_GPU_TRAINING_PY3 false
+    image_size_baseline: 28000
+    device_type: &DEVICE_TYPE gpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    cuda_version: &CUDA_VERSION cu130
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-ec2" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-ec2" ]
+    # skip_build: "False"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile.,
+                         *DEVICE_TYPE ]
+    target: ec2
+    context:
+      <<: *TRAINING_CONTEXT
diff --git a/pytorch/training/buildspec-2-9-sm.yml b/pytorch/training/buildspec-2-9-sm.yml
new file mode 100644
index 000000000000..c4b2f8dcdff2
--- /dev/null
+++ b/pytorch/training/buildspec-2-9-sm.yml
@@ -0,0 +1,75 @@
+account_id: &ACCOUNT_ID <set-$ACCOUNT_ID-in-environment>
+prod_account_id: &PROD_ACCOUNT_ID 763104351884
+region: &REGION <set-$REGION-in-environment>
+framework: &FRAMEWORK pytorch
+version: &VERSION 2.9.0
+short_version: &SHORT_VERSION "2.9"
+arch_type: x86
+# autopatch_build: "True"
+
+repository_info:
+  training_repository: &TRAINING_REPOSITORY
+    image_type: &TRAINING_IMAGE_TYPE training
+    root: !join [ *FRAMEWORK, "/", *TRAINING_IMAGE_TYPE ]
+    repository_name: &REPOSITORY_NAME !join [ pr, "-", *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE ]
+    repository: &REPOSITORY !join [ *ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/, *REPOSITORY_NAME ]
+    release_repository_name: &RELEASE_REPOSITORY_NAME !join [ *FRAMEWORK, "-", *TRAINING_IMAGE_TYPE ]
+    release_repository: &RELEASE_REPOSITORY !join [ *PROD_ACCOUNT_ID, .dkr.ecr., *REGION, .amazonaws.com/, *RELEASE_REPOSITORY_NAME ]
+
+context:
+  training_context: &TRAINING_CONTEXT
+    start_cuda_compat:
+      source: docker/build_artifacts/start_cuda_compat.sh
+      target: start_cuda_compat.sh
+    dockerd_entrypoint:
+      source: docker/build_artifacts/dockerd_entrypoint.sh
+      target: dockerd_entrypoint.sh
+    changehostname:
+      source: docker/build_artifacts/changehostname.c
+      target: changehostname.c
+    start_with_right_hostname:
+      source: docker/build_artifacts/start_with_right_hostname.sh
+      target: start_with_right_hostname.sh
+    example_mnist_file:
+      source: docker/build_artifacts/mnist.py
+      target: mnist.py
+    deep_learning_container:
+      source: ../../src/deep_learning_container.py
+      target: deep_learning_container.py
+    setup_oss_compliance:
+      source: ../../scripts/setup_oss_compliance.sh
+      target: setup_oss_compliance.sh
+
+images:
+  BuildSageMakerCPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_CPU_TRAINING_PY3 false
+    image_size_baseline: 7200
+    device_type: &DEVICE_TYPE cpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    # skip_build: "False"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /Dockerfile., *DEVICE_TYPE ]
+    target: sagemaker
+    context:
+      <<: *TRAINING_CONTEXT
+  BuildSageMakerGPUPTTrainPy3DockerImage:
+    <<: *TRAINING_REPOSITORY
+    build: &PYTORCH_GPU_TRAINING_PY3 false
+    image_size_baseline: 28000
+    device_type: &DEVICE_TYPE gpu
+    python_version: &DOCKER_PYTHON_VERSION py3
+    tag_python_version: &TAG_PYTHON_VERSION py312
+    cuda_version: &CUDA_VERSION cu130
+    os_version: &OS_VERSION ubuntu22.04
+    tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    latest_release_tag: !join [ *VERSION, "-", *DEVICE_TYPE, "-", *TAG_PYTHON_VERSION, "-", *CUDA_VERSION, "-", *OS_VERSION, "-sagemaker" ]
+    # skip_build: "False"
+    docker_file: !join [ docker/, *SHORT_VERSION, /, *DOCKER_PYTHON_VERSION, /, *CUDA_VERSION, /Dockerfile.,
+                         *DEVICE_TYPE ]
+    target: sagemaker
+    context:
+      <<: *TRAINING_CONTEXT
diff --git a/pytorch/training/buildspec.yml b/pytorch/training/buildspec.yml
index e9f328177b4b..e7a0d5614f66 100644
--- a/pytorch/training/buildspec.yml
+++ b/pytorch/training/buildspec.yml
@@ -1 +1 @@
-buildspec_pointer: buildspec-2-8-sm.yml
+buildspec_pointer: buildspec-2-9-ec2.yml
diff --git a/pytorch/training/docker/2.9/py3/Dockerfile.cpu b/pytorch/training/docker/2.9/py3/Dockerfile.cpu
new file mode 100644
index 000000000000..422912115a07
--- /dev/null
+++ b/pytorch/training/docker/2.9/py3/Dockerfile.cpu
@@ -0,0 +1,313 @@
+ARG PYTHON=python3
+ARG PYTHON_VERSION=3.12.10
+ARG PYTHON_SHORT_VERSION=3.12
+ARG PYTORCH_VERSION=2.9.0
+
+ARG OPEN_MPI_VERSION=4.1.7
+
+ARG TORCHTNT_VERSION=0.2.4
+ARG TORCHDATA_VERSION=0.11.0
+ARG TORCHAUDIO_VERSION=2.9.0
+ARG TORCHVISION_VERSION=0.24.0
+
+FROM ubuntu:22.04 AS base_image
+
+# This arg required to stop docker build waiting for region configuration while installing tz data from ubuntu 20
+ENV DEBIAN_FRONTEND=noninteractive
+ENV LD_LIBRARY_PATH="/usr/local/lib:${LD_LIBRARY_PATH}"
+
+RUN apt-get update \
+ && apt-get upgrade -y \
+ && apt-get autoremove -y \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+#################################################################
+#   ____
+#  / ___| ___  _ __ ___  _ __ ___   ___  _ __
+# | |    / _ \| '_ ` _ \| '_ ` _ \ / _ \| '_ \
+# | |___  (_) | | | | | | | | | | | (_) | | | |
+#  \____|\___/|_| |_| |_|_| |_| |_|\___/|_| |_|
+#  ___                              ____           _
+# |_ _|_ __ ___   __ _  __ _  ___  |  _ \ ___  ___(_)_ __   ___
+#  | || '_ ` _ \ / _` |/ _` |/ _ \ | |_) / _ \/ __| | '_ \ / _ \
+#  | || | | | | | (_| | (_| |  __/ |  _ <  __/ (__| | |_) |  __/
+# |___|_| |_| |_|\__,_|\__, |\___| |_| \_\___|\___|_| .__/ \___|
+#                      |___/                        |_|
+#################################################################
+
+FROM base_image AS common
+
+LABEL maintainer="Amazon AI"
+LABEL dlc_major_version="1"
+
+ARG PYTHON
+ARG PYTHON_VERSION
+ARG PYTHON_SHORT_VERSION
+ARG PYTORCH_VERSION
+ARG TORCHTNT_VERSION
+ARG TORCHDATA_VERSION
+ARG TORCHAUDIO_VERSION
+ARG TORCHVISION_VERSION
+
+ARG OPEN_MPI_VERSION
+
+ENV LD_LIBRARY_PATH="/usr/local/lib:${LD_LIBRARY_PATH}"
+ENV LD_LIBRARY_PATH="/lib/x86_64-linux-gnu:${LD_LIBRARY_PATH}"
+
+# Python won’t try to write .pyc or .pyo files on the import of source modules
+# Force stdin, stdout and stderr to be totally unbuffered. Good for logging
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONIOENCODING=UTF-8
+ENV LANG=C.UTF-8
+ENV LC_ALL=C.UTF-8
+
+ENV DLC_CONTAINER_TYPE=training
+WORKDIR /
+
+RUN apt-get update \
+ && apt-get -y upgrade --only-upgrade systemd \
+ && apt-get install -y --no-install-recommends \
+    automake \
+    build-essential \
+    ca-certificates \
+    cmake \
+    curl \
+    emacs \
+    git \
+    jq \
+    libcurl4-openssl-dev \
+    libglib2.0-0 \
+    libgl1-mesa-glx \
+    libsm6 \
+    libssl-dev \
+    libxext6 \
+    libxrender-dev \
+    zlib1g-dev \
+    unzip \
+    vim \
+    wget \
+    libbz2-dev \
+    libreadline-dev \
+    libsqlite3-dev \
+    llvm \
+    libncurses5-dev \
+    libncursesw5-dev \
+    xz-utils \
+    tk-dev \
+    liblzma-dev \
+    libffi-dev \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+# Install Open MPI
+RUN wget https://www.open-mpi.org/software/ompi/v4.1/downloads/openmpi-${OPEN_MPI_VERSION}.tar.gz \
+ && gunzip -c openmpi-${OPEN_MPI_VERSION}.tar.gz | tar xf - \
+ && cd openmpi-${OPEN_MPI_VERSION} \
+ && ./configure --prefix=/home/.openmpi \
+ && make all install \
+ && cd .. \
+ && rm openmpi-${OPEN_MPI_VERSION}.tar.gz \
+ && rm -rf openmpi-${OPEN_MPI_VERSION}
+
+# The ENV variables declared below are changed in the previous section
+# Grouping these ENV variables in the first section causes
+# ompi_info to fail. This is only observed in CPU containers
+ENV PATH="/home/.openmpi/bin:${PATH}"
+ENV LD_LIBRARY_PATH="/home/.openmpi/lib:${LD_LIBRARY_PATH}"
+RUN ompi_info --parsable --all | grep mpi_built_with_cuda_support:value
+
+# Install OpenSSH for MPI to communicate between containers, allow OpenSSH to talk to containers without asking for confirmation
+RUN apt-get update \
+ && apt-get install -y --no-install-recommends openssh-client openssh-server \
+ && mkdir -p /var/run/sshd \
+ && cat /etc/ssh/ssh_config | grep -v StrictHostKeyChecking > /etc/ssh/ssh_config.new \
+ && echo "    StrictHostKeyChecking no" >> /etc/ssh/ssh_config.new \
+ && mv /etc/ssh/ssh_config.new /etc/ssh/ssh_config \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+# Configure OpenSSH so that nodes can communicate with each other
+RUN mkdir -p /var/run/sshd \
+ && sed 's@session\s*required\s*pam_loginuid.so@session optional pam_loginuid.so@g' -i /etc/pam.d/sshd
+
+RUN rm -rf /root/.ssh/ \
+ && mkdir -p /root/.ssh/ \
+ && ssh-keygen -q -t rsa -N '' -f /root/.ssh/id_rsa \
+ && cp /root/.ssh/id_rsa.pub /root/.ssh/authorized_keys \
+ && printf "Host *\n StrictHostKeyChecking no\n" >> /root/.ssh/config
+
+# install python
+RUN cd /tmp/ \
+&& wget https://www.python.org/ftp/python/${PYTHON_VERSION}/Python-${PYTHON_VERSION}.tgz \
+&& tar xzf Python-${PYTHON_VERSION}.tgz \
+&& cd Python-${PYTHON_VERSION} \
+&& ./configure --enable-optimizations --with-lto --with-computed-gotos --with-system-ffi \
+&& make -j "$(nproc)" \
+&& make altinstall \
+&& cd .. \
+&& rm -rf Python-${PYTHON_VERSION} \
+&& rm Python-${PYTHON_VERSION}.tgz \
+&& ln -s /usr/local/bin/python${PYTHON_SHORT_VERSION} /usr/local/bin/python \
+&& ln -s /usr/local/bin/python${PYTHON_SHORT_VERSION} /usr/local/bin/python3 \
+# This installation generate a .python_history file in the root directory leads sanity check to fail
+&& rm -f /root/.python_history
+
+# Python Path
+ENV PATH="/usr/local/bin:${PATH}"
+
+# this will add pip systemlink to pip${PYTHON_SHORT_VERSION}
+RUN python -m pip install --upgrade pip --trusted-host pypi.org --trusted-host files.pythonhosted.org
+
+# Install common packages
+RUN pip install --no-cache-dir \
+    cython \
+    cryptography \
+    pyOpenSSL \
+    pybind11 \
+    mkl \
+    mkl-include \
+    parso \
+    typing \
+    charset-normalizer \
+    packaging \
+    boto3 \
+    PyYAML \
+    numpy \
+    scipy \
+    click \
+    psutil \
+    ipython \
+    ipykernel \
+    pillow \
+    h5py \
+    fsspec \
+    "idna>=3.7" \
+    "tqdm>=4.66.3" \
+    "requests>=2.32.0" \
+    "setuptools>=70.0.0" \
+    "urllib3>=2.5.0" \
+    "awscli" \
+    opencv-python==4.11.0.86 \
+    mpi4py \
+    jinja2>=3.1.6 \
+    tornado>=6.5.1
+
+# Install PyTorch
+RUN pip install --no-cache-dir -U torch==${PYTORCH_VERSION} \
+    torchvision==${TORCHVISION_VERSION} \
+    torchaudio==${TORCHAUDIO_VERSION} \
+    --index-url https://download.pytorch.org/whl/cpu \
+    && pip install --no-cache-dir -U torchtnt==${TORCHTNT_VERSION} \
+    torchdata==${TORCHDATA_VERSION} \
+    s3torchconnector \
+    fastai \
+    accelerate \
+    # pin numpy requirement for fastai dependency
+    # requires explicit declaration of spacy, thic, blis
+    spacy \
+    # pin thinc due to incompatibility with numpy 1.26.4 (sagemaker doesn't support latest numpy)
+    thinc \
+    blis \
+    numpy \
+ && pip uninstall -y dataclasses
+
+RUN curl -o /license.txt https://aws-dlc-licenses.s3.amazonaws.com/pytorch-2.9/license.txt
+
+COPY deep_learning_container.py /usr/local/bin/deep_learning_container.py
+
+RUN chmod +x /usr/local/bin/deep_learning_container.py
+
+COPY bash_telemetry.sh /usr/local/bin/bash_telemetry.sh
+RUN chmod +x /usr/local/bin/bash_telemetry.sh
+RUN echo 'source /usr/local/bin/bash_telemetry.sh' >> /etc/bash.bashrc
+
+# Removing the cache as it is needed for security verification
+RUN rm -rf /root/.cache | true
+
+########################################################
+#  _____ ____ ____    ___
+# | ____/ ___|___ \  |_ _|_ __ ___   __ _  __ _  ___
+# |  _|| |     __) |  | || '_ ` _ \ / _` |/ _` |/ _ \
+# | |__| |___ / __/   | || | | | | | (_| | (_| |  __/
+# |_____\____|_____| |___|_| |_| |_|\__,_|\__, |\___|
+#                                         |___/
+#  ____           _
+# |  _ \ ___  ___(_)_ __   ___
+# | |_) / _ \/ __| | '_ \ / _ \
+# |  _ <  __/ (__| | |_) |  __/
+# |_| \_\___|\___|_| .__/ \___|
+#                  |_|
+########################################################
+
+FROM common AS ec2
+
+WORKDIR /
+
+COPY setup_oss_compliance.sh setup_oss_compliance.sh
+RUN bash setup_oss_compliance.sh ${PYTHON} && rm setup_oss_compliance.sh
+
+COPY dockerd_entrypoint.sh /usr/local/bin/dockerd_entrypoint.sh
+RUN chmod +x /usr/local/bin/dockerd_entrypoint.sh
+ENTRYPOINT ["bash", "-m", "dockerd_entrypoint.sh"]
+
+# Starts framework
+CMD ["/bin/bash"]
+
+#################################################################
+#  ____                   __  __       _
+# / ___|  __ _  __ _  ___|  \/  | __ _| | _____ _ __
+# \___ \ / _` |/ _` |/ _ \ |\/| |/ _` | |/ / _ \ '__|
+#  ___) | (_| | (_| |  __/ |  | | (_| |   <  __/ |
+# |____/ \__,_|\__, |\___|_|  |_|\__,_|_|\_\___|_|
+#              |___/
+#  ___                              ____           _
+# |_ _|_ __ ___   __ _  __ _  ___  |  _ \ ___  ___(_)_ __   ___
+#  | || '_ ` _ \ / _` |/ _` |/ _ \ | |_) / _ \/ __| | '_ \ / _ \
+#  | || | | | | | (_| | (_| |  __/ |  _ <  __/ (__| | |_) |  __/
+# |___|_| |_| |_|\__,_|\__, |\___| |_| \_\___|\___|_| .__/ \___|
+#                      |___/                        |_|
+#################################################################
+
+FROM common AS sagemaker
+
+LABEL maintainer="Amazon AI"
+LABEL dlc_major_version="1"
+
+ENV SAGEMAKER_TRAINING_MODULE=sagemaker_pytorch_container.training:main
+
+WORKDIR /
+
+# Install SM packages
+RUN pip install --no-cache-dir -U \
+    smclarify \
+    "sagemaker>=2" \
+    sagemaker-experiments \
+    sagemaker-pytorch-training \
+    sagemaker-training
+
+# Install extra packages
+RUN pip install --no-cache-dir -U \
+    bokeh \
+    imageio \
+    numba \
+    pandas \
+    plotly \
+    scikit-learn \
+    seaborn \
+    shap \
+    cloudpickle
+
+# Copy workaround script for incorrect hostname
+COPY changehostname.c /
+COPY start_with_right_hostname.sh /usr/local/bin/start_with_right_hostname.sh
+
+RUN chmod +x /usr/local/bin/start_with_right_hostname.sh
+
+COPY setup_oss_compliance.sh setup_oss_compliance.sh
+RUN bash setup_oss_compliance.sh ${PYTHON} && rm setup_oss_compliance.sh
+
+ENTRYPOINT ["bash", "-m", "start_with_right_hostname.sh"]
+CMD ["/bin/bash"]
+
diff --git a/pytorch/training/docker/2.9/py3/cu130/Dockerfile.gpu b/pytorch/training/docker/2.9/py3/cu130/Dockerfile.gpu
new file mode 100644
index 000000000000..5cd5bf502de0
--- /dev/null
+++ b/pytorch/training/docker/2.9/py3/cu130/Dockerfile.gpu
@@ -0,0 +1,287 @@
+ARG PYTHON=python3
+ARG PYTHON_VERSION=3.12.10
+ARG PYTHON_SHORT_VERSION=3.12
+ARG PYTORCH_VERSION=2.9.0
+ARG TORCHTNT_VERSION=0.2.4
+ARG TORCHAUDIO_VERSION=2.9.0
+ARG TORCHVISION_VERSION=0.24.0
+ARG TORCHDATA_VERSION=0.11.0
+
+ARG GDRCOPY_VERSION=2.5.1
+ARG TE_VERSION=2.9
+ARG FLASH_ATTN_VERSION=2.8.3
+
+#################################################################
+#   ____
+#  / ___| ___  _ __ ___  _ __ ___   ___  _ __
+# | |    / _ \| '_ ` _ \| '_ ` _ \ / _ \| '_ \
+# | |___  (_) | | | | | | | | | | | (_) | | | |
+#  \____|\___/|_| |_| |_|_| |_| |_|\___/|_| |_|
+#  ___                              ____           _
+# |_ _|_ __ ___   __ _  __ _  ___  |  _ \ ___  ___(_)_ __   ___
+#  | || '_ ` _ \ / _` |/ _` |/ _ \ | |_) / _ \/ __| | '_ \ / _ \
+#  | || | | | | | (_| | (_| |  __/ |  _ <  __/ (__| | |_) |  __/
+# |___|_| |_| |_|\__,_|\__, |\___| |_| \_\___|\___|_| .__/ \___|
+#                      |___/                        |_|
+#################################################################
+
+FROM public.ecr.aws/deep-learning-containers/base:13.0.0-gpu-py312-ubuntu22.04-ec2 AS common
+# base has EFA, PYTHON and CUDA 13.0
+
+LABEL maintainer="Amazon AI"
+LABEL dlc_major_version="1"
+
+ARG PYTHON
+ARG PYTORCH_VERSION
+ARG TORCHDATA_VERSION
+ARG TORCHAUDIO_VERSION
+ARG TORCHVISION_VERSION
+ARG TORCHTNT_VERSION
+ARG TE_VERSION
+ARG FLASH_ATTN_VERSION
+ARG GDRCOPY_VERSION
+
+ENV CUDA_HOME="/usr/local/cuda"
+ENV PATH="${CUDA_HOME}/bin:${PATH}"
+ENV EFA_PATH="/opt/amazon/efa"
+ENV OPEN_MPI_PATH="/opt/amazon/openmpi"
+
+# Python won’t try to write .pyc or .pyo files on the import of source modules
+# Force stdin, stdout and stderr to be totally unbuffered. Good for logging
+ENV PYTHONDONTWRITEBYTECODE=1
+ENV PYTHONUNBUFFERED=1
+ENV PYTHONIOENCODING=UTF-8
+ENV LANG=C.UTF-8
+ENV LC_ALL=C.UTF-8
+
+ENV TORCH_NVCC_FLAGS="-Xfatbin -compress-all"
+
+ENV DLC_CONTAINER_TYPE=training
+WORKDIR /
+
+RUN apt-get update \
+ && apt-get -y upgrade --only-upgrade systemd \
+ && apt-get install -y --allow-change-held-packages --no-install-recommends \
+    libgl1-mesa-glx \
+    build-essential \
+    ca-certificates \
+    zlib1g-dev \
+    openssl \
+    python3-dev \
+    pkg-config \
+    check \
+    llvm \
+    xz-utils \
+ && rm -rf /var/lib/apt/lists/* \
+ && apt-get clean
+
+ENV PATH="${OPEN_MPI_PATH}/bin:${EFA_PATH}/bin:${PATH}"
+ENV LD_LIBRARY_PATH="/usr/local/lib:/opt/amazon/ofi-nccl/lib:/opt/amazon/openmpi/lib:/opt/amazon/efa/lib:/usr/local/cuda/lib64:${LD_LIBRARY_PATH}"
+# Python Path
+ENV PATH="/usr/local/bin:${PATH}"
+
+# Install common conda packages
+RUN pip install --no-cache-dir \
+    cython \
+    cryptography \
+    pyOpenSSL \
+    pybind11 \
+    mkl \
+    mkl-include \
+    parso \
+    typing \
+    charset-normalizer \
+    packaging \
+    PyYAML \
+    numpy \
+    scipy \
+    click \
+    psutil \
+    ipython \
+    ipykernel \
+    pillow \
+    h5py \
+    fsspec \
+    "idna>=3.7" \
+    "tqdm>=4.66.3" \
+    "requests>=2.32.0" \
+    "setuptools>=70.0.0" \
+    "urllib3>=2.5.0" \
+    ninja \
+    opencv-python==4.11.0.86 \
+    mpi4py \
+    jinja2>=3.1.6 \
+    tornado>=6.5.1
+
+# Install PyTorch
+RUN pip install --no-cache-dir -U torch==${PYTORCH_VERSION} \
+    torchvision==${TORCHVISION_VERSION} \
+    torchaudio==${TORCHAUDIO_VERSION} \
+    --index-url https://download.pytorch.org/whl/cu130 \
+    && pip install --no-cache-dir -U torchtnt==${TORCHTNT_VERSION} \
+    torchdata==${TORCHDATA_VERSION} \
+    triton \
+    s3torchconnector \
+    fastai \
+    accelerate \
+    # pin numpy requirement for fastai dependency
+    # requires explicit declaration of spacy, thic, blis
+    spacy \
+    # pin thinc due to incompatibility with numpy 1.26.4 (sagemaker doesn't support latest numpy)
+    thinc \
+    blis \
+    numpy \
+ && pip uninstall -y dataclasses
+
+# Install flash attn and NVIDIA transformer engine.
+# Optionally set NVTE_FRAMEWORK to avoid bringing in additional frameworks during TE install
+ENV NVTE_FRAMEWORK=pytorch
+
+# Install flash-attn using instructions from https://github.com/Dao-AILab/flash-attention#installation-and-features
+# Set MAX_JOBS=4 to avoid OOM issues in installation process
+RUN MAX_JOBS=4 pip install --no-cache-dir flash-attn==${FLASH_ATTN_VERSION} --no-build-isolation
+
+RUN pip install --no-cache-dir nvidia-mathdx
+    
+# Install TE using instructions from https://docs.nvidia.com/deeplearning/transformer-engine/user-guide/installation.html
+RUN pip install --no-cache-dir git+https://github.com/NVIDIA/TransformerEngine.git@release_v${TE_VERSION} --no-build-isolation
+
+RUN curl -o /license.txt https://aws-dlc-licenses.s3.amazonaws.com/pytorch-2.9/license.txt
+
+COPY start_cuda_compat.sh /usr/local/bin/start_cuda_compat.sh
+RUN chmod +x /usr/local/bin/start_cuda_compat.sh
+
+COPY bash_telemetry.sh /usr/local/bin/bash_telemetry.sh
+RUN chmod +x /usr/local/bin/bash_telemetry.sh
+RUN echo 'source /usr/local/bin/bash_telemetry.sh' >> /etc/bash.bashrc
+
+# Install GDRCopy which is a dependency of SM Distributed DataParallel binary
+# The test binaries requires cuda driver library which could be found in conda
+# So update the linker path to point to it to avoid -Lcuda not found
+RUN cd /tmp \
+ && git clone https://github.com/NVIDIA/gdrcopy.git -b v${GDRCOPY_VERSION} \
+ && cd gdrcopy \
+ && sed -ie '13s@$@ -L $(CUDA)/lib64/stubs@' tests/Makefile \
+ && CUDA=${CUDA_HOME} make install \
+ && rm -rf /tmp/gdrcopy
+
+# Install common packages used by both EC2 and SageMaker
+RUN apt-get update \
+ && apt-get install -y --no-install-recommends \
+    curl \
+    wget \
+    git \
+    jq \
+    emacs \
+    vim \
+    unzip \
+ && rm -rf /var/lib/apt/lists/* \
+ && apt-get clean
+
+# Removing the cache as it is needed for security verification
+RUN rm -rf /root/.cache | true
+
+########################################################
+#  _____ ____ ____    ___
+# | ____/ ___|___ \  |_ _|_ __ ___   __ _  __ _  ___
+# |  _|| |     __) |  | || '_ ` _ \ / _` |/ _` |/ _ \
+# | |__| |___ / __/   | || | | | | | (_| | (_| |  __/
+# |_____\____|_____| |___|_| |_| |_|\__,_|\__, |\___|
+#                                         |___/
+#  ____           _
+# |  _ \ ___  ___(_)_ __   ___
+# | |_) / _ \/ __| | '_ \ / _ \
+# |  _ <  __/ (__| | |_) |  __/
+# |_| \_\___|\___|_| .__/ \___|
+#                  |_|
+########################################################
+
+FROM common AS ec2
+
+ARG PYTHON
+
+WORKDIR /
+
+COPY dockerd_entrypoint.sh /usr/local/bin/dockerd_entrypoint.sh
+RUN chmod +x /usr/local/bin/dockerd_entrypoint.sh
+
+COPY setup_oss_compliance.sh setup_oss_compliance.sh
+RUN bash setup_oss_compliance.sh ${PYTHON} && rm setup_oss_compliance.sh
+
+# This arg required to stop docker build waiting for region configuration while installing tz data from ubuntu 20
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update \
+ && apt-get upgrade -y \
+ && apt-get autoremove -y \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+ENTRYPOINT ["bash", "-m", "dockerd_entrypoint.sh"]
+CMD ["/bin/bash"]
+
+#################################################################
+#  ____                   __  __       _
+# / ___|  __ _  __ _  ___|  \/  | __ _| | _____ _ __
+# \___ \ / _` |/ _` |/ _ \ |\/| |/ _` | |/ / _ \ '__|
+#  ___) | (_| | (_| |  __/ |  | | (_| |   <  __/ |
+# |____/ \__,_|\__, |\___|_|  |_|\__,_|_|\_\___|_|
+#              |___/
+#  ___                              ____           _
+# |_ _|_ __ ___   __ _  __ _  ___  |  _ \ ___  ___(_)_ __   ___
+#  | || '_ ` _ \ / _` |/ _` |/ _ \ | |_) / _ \/ __| | '_ \ / _ \
+#  | || | | | | | (_| | (_| |  __/ |  _ <  __/ (__| | |_) |  __/
+# |___|_| |_| |_|\__,_|\__, |\___| |_| \_\___|\___|_| .__/ \___|
+#                      |___/                        |_|
+#################################################################
+
+FROM common AS sagemaker
+
+LABEL maintainer="Amazon AI"
+LABEL dlc_major_version="1"
+
+ENV SAGEMAKER_TRAINING_MODULE=sagemaker_pytorch_container.training:main
+
+ARG PYTHON
+
+WORKDIR /
+
+# Install SM packages
+RUN pip install --no-cache-dir -U \
+    smclarify \
+    "sagemaker>=2" \
+    sagemaker-experiments \
+    sagemaker-pytorch-training \
+    sagemaker-training
+
+# Install extra packages
+RUN pip install --no-cache-dir -U \
+    bokeh \
+    imageio \
+    numba \
+    pandas \
+    plotly \
+    shap \
+    scikit-learn \
+    seaborn \
+    cloudpickle
+
+COPY setup_oss_compliance.sh setup_oss_compliance.sh
+RUN bash setup_oss_compliance.sh ${PYTHON} && rm setup_oss_compliance.sh
+
+# Copy workaround script for incorrect hostname
+COPY changehostname.c /
+COPY start_with_right_hostname.sh /usr/local/bin/start_with_right_hostname.sh
+RUN chmod +x /usr/local/bin/start_with_right_hostname.sh
+
+# This arg required to stop docker build waiting for region configuration while installing tz data from ubuntu 20
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update \
+ && apt-get upgrade -y \
+ && apt-get autoremove -y \
+ && apt-get clean \
+ && rm -rf /var/lib/apt/lists/*
+
+ENTRYPOINT ["bash", "-m", "start_with_right_hostname.sh"]
+CMD ["/bin/bash"]
\ No newline at end of file
diff --git a/test/dlc_tests/conftest.py b/test/dlc_tests/conftest.py
index dc6fc2ea624e..7e7522995fca 100644
--- a/test/dlc_tests/conftest.py
+++ b/test/dlc_tests/conftest.py
@@ -55,6 +55,7 @@
     # ECR repo name fixtures
     # PyTorch
     "pytorch_training",
+    "pytorch_training___2__9",
     "pytorch_training___2__8",
     "pytorch_training___2__7",
     "pytorch_training___2__6",
diff --git a/test/dlc_tests/container_tests/bin/efa/testEFA b/test/dlc_tests/container_tests/bin/efa/testEFA
index 4b676249d816..fd66b0ab2ccc 100755
--- a/test/dlc_tests/container_tests/bin/efa/testEFA
+++ b/test/dlc_tests/container_tests/bin/efa/testEFA
@@ -89,7 +89,7 @@ check_efa_nccl_all_reduce(){
     
     RETURN_VAL=${PIPESTATUS[0]}
     # In case, if you would like see logs, uncomment below line
-    # RESULT=$(cat ${TRAINING_LOG})
+    #RESULT=$(cat ${TRAINING_LOG})
 
     if [ ${RETURN_VAL} -eq 0 ]; then
         echo "***************************** check_efa_nccl_all_reduce passed *****************************"
diff --git a/test/dlc_tests/container_tests/bin/transformerengine/testPTTransformerEngine b/test/dlc_tests/container_tests/bin/transformerengine/testPTTransformerEngine
index e057b957ab54..490c727c4840 100755
--- a/test/dlc_tests/container_tests/bin/transformerengine/testPTTransformerEngine
+++ b/test/dlc_tests/container_tests/bin/transformerengine/testPTTransformerEngine
@@ -36,6 +36,26 @@ elif [ $(version $TE_VERSION) -lt $(version "2.0") ]; then
   pytest -v -s $TE_PATH/tests/pytorch/test_fused_optimizer.py
   pytest -v -s $TE_PATH/tests/pytorch/test_multi_tensor.py
   pytest -v -s $TE_PATH/tests/pytorch/test_fusible_ops.py
+elif [ $(version $TE_VERSION) -lt $(version "3.0") ]; then
+  pip install pytest==8.2.1 onnxruntime onnx expecttest
+  pytest -v -s $TE_PATH/tests/pytorch/test_sanity.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_recipe.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_deferred_init.py
+  # Disabled test due to bug: https://github.com/NVIDIA/TransformerEngine/issues/1165
+  # PYTORCH_JIT=0 NVTE_TORCH_COMPILE=0 NVTE_ALLOW_NONDETERMINISTIC_ALGO=0 pytest -v -s $TE_PATH/tests/pytorch/test_numerics.py
+  PYTORCH_JIT=0 NVTE_TORCH_COMPILE=0 NVTE_ALLOW_NONDETERMINISTIC_ALGO=0 pytest -v -s $TE_PATH/tests/pytorch/test_cuda_graphs.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_jit.py
+  # Skip test_fused_attn.py as it doesn't exist in TE 2.8
+  # NVTE_TORCH_COMPILE=0 pytest -v -s $TE_PATH/tests/pytorch/fused_attn/test_fused_attn.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_fused_rope.py
+  # Disable onnx test due lack of TE prioritization on onnx: https://github.com/NVIDIA/TransformerEngine/issues/528
+  # NVTE_TORCH_COMPILE=0 pytest -v -s $TE_PATH/tests/pytorch/test_onnx_export.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_float8tensor.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_gqa.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_fused_optimizer.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_multi_tensor.py
+  pytest -v -s $TE_PATH/tests/pytorch/test_fusible_ops.py
+
 else
   pip install pytest==8.2.1 onnxruntime onnx expecttest
   pytest -v -s $TE_PATH/tests/pytorch/test_sanity.py
diff --git a/test/dlc_tests/ec2/pytorch/training/test_pytorch_training_2_9.py b/test/dlc_tests/ec2/pytorch/training/test_pytorch_training_2_9.py
new file mode 100644
index 000000000000..7441b4933750
--- /dev/null
+++ b/test/dlc_tests/ec2/pytorch/training/test_pytorch_training_2_9.py
@@ -0,0 +1,137 @@
+import pytest
+
+import test.test_utils as test_utils
+
+from test.test_utils import ec2
+
+from test.dlc_tests.ec2.pytorch.training import common_cases
+from test.dlc_tests.ec2 import smclarify_cases
+
+
+@pytest.mark.usefixtures("sagemaker")
+@pytest.mark.integration("pytorch_gpu_tests")
+@pytest.mark.model("N/A")
+@pytest.mark.team("conda")
+@pytest.mark.parametrize(
+    "ec2_instance_type, region", common_cases.PT_EC2_GPU_INSTANCE_TYPE_AND_REGION, indirect=True
+)
+def test_pytorch_2_9_gpu(
+    pytorch_training___2__9, ec2_connection, region, gpu_only, ec2_instance_type
+):
+    pytorch_training = pytorch_training___2__9
+    if test_utils.is_image_incompatible_with_instance_type(pytorch_training, ec2_instance_type):
+        pytest.skip(
+            f"Image {pytorch_training} is incompatible with instance type {ec2_instance_type}"
+        )
+
+    test_cases = [
+        (common_cases.pytorch_standalone, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_mnist, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_linear_regression_gpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_gloo, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_nccl, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_mpi, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_torchaudio, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_torchdata, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_cudnn_match_gpu, (pytorch_training, ec2_connection, region)),
+        (common_cases.pytorch_curand_gpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_telemetry_bashrc_gpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_telemetry_entrypoint_gpu, (pytorch_training, ec2_connection)),
+    ]
+
+    if "sagemaker" in pytorch_training:
+        test_cases.append(
+            (smclarify_cases.smclarify_metrics_gpu, (pytorch_training, ec2_connection)),
+        )
+
+    # AMP must be run on multi_gpu
+    if ec2.is_instance_multi_gpu(ec2_instance_type):
+        test_cases.append((common_cases.pytorch_amp, (pytorch_training, ec2_connection)))
+
+    test_utils.execute_serial_test_cases(test_cases, test_description="PT 2.9 GPU")
+
+
+@pytest.mark.usefixtures("sagemaker")
+@pytest.mark.integration("pytorch_gpu_heavy_tests")
+@pytest.mark.model("N/A")
+@pytest.mark.team("conda")
+@pytest.mark.parametrize(
+    "ec2_instance_type, region",
+    common_cases.PT_EC2_HEAVY_GPU_INSTANCE_TYPE_AND_REGION,
+    indirect=True,
+)
+@pytest.mark.skipif(
+    test_utils.is_pr_context() and not ec2.are_heavy_instance_ec2_tests_enabled(),
+    reason="Skip GPU Heavy tests in PR context unless explicitly enabled",
+)
+def test_pytorch_2_9_gpu_heavy(
+    pytorch_training___2__9, ec2_connection, region, gpu_only, ec2_instance_type
+):
+    pytorch_training = pytorch_training___2__9
+    if test_utils.is_image_incompatible_with_instance_type(pytorch_training, ec2_instance_type):
+        pytest.skip(
+            f"Image {pytorch_training} is incompatible with instance type {ec2_instance_type}"
+        )
+
+    test_cases = [
+        (common_cases.pytorch_gdrcopy, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_transformer_engine, (pytorch_training, ec2_connection)),
+    ]
+
+    test_utils.execute_serial_test_cases(test_cases, test_description="PT 2.9 GPU Heavy")
+
+
+@pytest.mark.usefixtures("sagemaker")
+@pytest.mark.integration("inductor")
+@pytest.mark.model("N/A")
+@pytest.mark.team("training-compiler")
+@pytest.mark.parametrize(
+    "ec2_instance_type, region",
+    common_cases.PT_EC2_GPU_INDUCTOR_INSTANCE_TYPE_AND_REGION,
+    indirect=True,
+)
+def test_pytorch_2_9_gpu_inductor(
+    pytorch_training___2__9, ec2_connection, region, gpu_only, ec2_instance_type
+):
+    pytorch_training = pytorch_training___2__9
+    if test_utils.is_image_incompatible_with_instance_type(pytorch_training, ec2_instance_type):
+        pytest.skip(
+            f"Image {pytorch_training} is incompatible with instance type {ec2_instance_type}"
+        )
+
+    test_cases = [
+        (common_cases.pytorch_gloo_inductor_gpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_mpi_inductor_gpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_nccl_inductor, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_amp_inductor, (pytorch_training, ec2_connection)),
+    ]
+
+    test_utils.execute_serial_test_cases(test_cases, test_description="PT 2.9 GPU Inductor")
+
+
+@pytest.mark.usefixtures("sagemaker")
+@pytest.mark.integration("pytorch_cpu_tests")
+@pytest.mark.model("N/A")
+@pytest.mark.team("conda")
+@pytest.mark.parametrize("ec2_instance_type", common_cases.PT_EC2_CPU_INSTANCE_TYPE, indirect=True)
+def test_pytorch_2_9_cpu(pytorch_training___2__9, ec2_connection, cpu_only):
+    pytorch_training = pytorch_training___2__9
+
+    test_cases = [
+        (common_cases.pytorch_standalone, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_mnist, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_linear_regression_cpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_gloo, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_mpi, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_torchaudio, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_training_torchdata, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_telemetry_bashrc_cpu, (pytorch_training, ec2_connection)),
+        (common_cases.pytorch_telemetry_entrypoint_cpu, (pytorch_training, ec2_connection)),
+    ]
+
+    if "sagemaker" in pytorch_training:
+        test_cases += [
+            (smclarify_cases.smclarify_metrics_cpu, (pytorch_training, ec2_connection)),
+        ]
+
+    test_utils.execute_serial_test_cases(test_cases, test_description="PT 2.9 CPU")
diff --git a/test/sagemaker_tests/pytorch/training/conftest.py b/test/sagemaker_tests/pytorch/training/conftest.py
index 20e381a9ffbc..196096c79056 100644
--- a/test/sagemaker_tests/pytorch/training/conftest.py
+++ b/test/sagemaker_tests/pytorch/training/conftest.py
@@ -432,7 +432,11 @@ def skip_smppy_test(
     """For each currency release, we can skip smppy tests if the Profiler binary does not exist.
     However, when the Profiler binaries are added, be sure to fix the test logic such that the tests are not skipped.
     """
-    skip_dict = {">=2.7.1,<2.8": ["cpu", "cu128"], ">=2.8,<2.9": ["cpu", "cu129"]}
+    skip_dict = {
+        ">=2.7.1,<2.8": ["cpu", "cu128"],
+        ">=2.8,<2.9": ["cpu", "cu129"],
+        ">=2.9,<3.0": ["cpu", "cu130"],
+    }
     if _validate_pytorch_framework_version(
         request, processor, ecr_image, "skip_smppy_test", skip_dict
     ):