diff --git a/3rdparty/json b/3rdparty/json
index bc889afb4c..55f93686c0 160000
--- a/3rdparty/json
+++ b/3rdparty/json
@@ -1 +1 @@
-Subproject commit bc889afb4c5bf1c0d8ee29ef35eaaf4c8bef8a5d
+Subproject commit 55f93686c01528224f448c19128836e7df245f72
diff --git a/bringup_fix.sh b/bringup_fix.sh
new file mode 100644
index 0000000000..9d52d15748
--- /dev/null
+++ b/bringup_fix.sh
@@ -0,0 +1,47 @@
+ARCH=$(uname -m)
+if [ $ARCH == "x86_64" ]; then
+
+wget https://urm.nvidia.com/artifactory/sw-gpu-cuda-installer-generic-local/packaging/r13.0/cuda_nvrtc/linux-x86_64/13.0.48/cuda-nvrtc-dev-13-0_13.0.48-1_amd64.deb && \
+    dpkg -i cuda-nvrtc-dev-13-0_13.0.48-1_amd64.deb && \
+    rm cuda-nvrtc-dev-13-0_13.0.48-1_amd64.deb
+
+wget https://github.com/Kitware/CMake/releases/download/v4.0.3/cmake-4.0.3-linux-x86_64.sh && \
+    bash cmake-4.0.3-linux-x86_64.sh --skip-license --prefix=/usr/local/cmake --exclude-subdir
+
+apt update
+apt remove -y ibverbs-providers libibverbs1
+apt install -y libibverbs-dev
+apt install -y libstdc++-14-dev
+
+elif [ $ARCH == "aarch64" ]; then
+
+wget https://urm.nvidia.com/artifactory/sw-gpu-cuda-installer-generic-local/packaging/r13.0/cuda_nvrtc/linux-sbsa/13.0.48/cuda-nvrtc-dev-13-0_13.0.48-1_arm64.deb && \
+    dpkg -i cuda-nvrtc-dev-13-0_13.0.48-1_arm64.deb && \
+    rm cuda-nvrtc-dev-13-0_13.0.48-1_arm64.deb
+
+wget https://github.com/Kitware/CMake/releases/download/v4.0.3/cmake-4.0.3-linux-aarch64.sh && \
+    bash cmake-4.0.3-linux-aarch64.sh --skip-license --prefix=/usr/local/cmake --exclude-subdir
+
+apt update
+apt remove -y ibverbs-providers libibverbs1
+apt install -y libibverbs-dev
+apt install -y libstdc++-14-dev
+
+else
+    echo "Unsupported architecture: $ARCH"
+    exit 1
+fi
+
+cd /usr/local/lib/python3.12/dist-packages/ && \
+    ls -la | grep pytorch_triton && \
+    mv pytorch_triton-3.3.1+gitc8757738.dist-info triton-3.3.1+gitc8757738.dist-info && \
+    cd triton-3.3.1+gitc8757738.dist-info && \
+    echo "Current directory: $(pwd)" && \
+    echo "Files in directory:" && \
+    ls -la && \
+    sed -i 's/^Name: pytorch-triton/Name: triton/' METADATA && \
+    sed -i 's|pytorch_triton-3.3.1+gitc8757738.dist-info/|triton-3.3.1+gitc8757738.dist-info/|g' RECORD && \
+    echo "METADATA after update:" && \
+    grep "^Name:" METADATA
+
+# pip install git+https://github.com/triton-lang/triton.git@main
diff --git a/cpp/CMakeLists.txt b/cpp/CMakeLists.txt
index 6732db6eaa..dd3f5423fd 100644
--- a/cpp/CMakeLists.txt
+++ b/cpp/CMakeLists.txt
@@ -472,7 +472,6 @@ print(os.path.dirname(torch.__file__),end='');"
       endif()
     endif()
   endif()
-
 else()
   if(NOT WIN32)
     if(NOT USE_CXX11_ABI)
diff --git a/cpp/tensorrt_llm/deep_ep/CMakeLists.txt b/cpp/tensorrt_llm/deep_ep/CMakeLists.txt
index 088391aef4..68d74a8858 100644
--- a/cpp/tensorrt_llm/deep_ep/CMakeLists.txt
+++ b/cpp/tensorrt_llm/deep_ep/CMakeLists.txt
@@ -127,6 +127,8 @@ ExternalProject_Add(
                 ${DEEP_EP_SOURCE_DIR}/third-party/nvshmem.patch
   COMMAND sed "s/TRANSPORT_VERSION_MAJOR 3/TRANSPORT_VERSION_MAJOR 103/" -i
           src/CMakeLists.txt
+  COMMAND sed "s/_STANDARD 11/_STANDARD 17/" -i src/device/CMakeLists.txt
+  COMMAND sed "s/_STANDARD 11/_STANDARD 17/" -i src/CMakeLists.txt
   COMMAND patch -p1 --forward --batch -i
           ${CMAKE_CURRENT_SOURCE_DIR}/nvshmem_fast_build.patch
   CMAKE_CACHE_ARGS
diff --git a/cpp/tensorrt_llm/executor/tensor.cpp b/cpp/tensorrt_llm/executor/tensor.cpp
index 8b7e15ac17..c38feb0e34 100644
--- a/cpp/tensorrt_llm/executor/tensor.cpp
+++ b/cpp/tensorrt_llm/executor/tensor.cpp
@@ -63,9 +63,9 @@ DataType Tensor::getDataType() const
     case nvinfer1::DataType::kBF16: return DataType::kBF16;
     case nvinfer1::DataType::kINT64: return DataType::kINT64;
     case nvinfer1::DataType::kINT4: [[fallthrough]] /* do nothing */;
-    case nvinfer1::DataType::kFP4: /* do nothing */;
+    case nvinfer1::DataType::kFP4: [[fallthrough]] /* do nothing */;
+    default: TLLM_THROW("Unsupported data type");
     }
-    TLLM_THROW("Unsupported data type");
 }
 
 MemoryType Tensor::getMemoryType() const
diff --git a/cpp/tensorrt_llm/runtime/iBuffer.cpp b/cpp/tensorrt_llm/runtime/iBuffer.cpp
index f676fee088..77707a0e4c 100644
--- a/cpp/tensorrt_llm/runtime/iBuffer.cpp
+++ b/cpp/tensorrt_llm/runtime/iBuffer.cpp
@@ -101,9 +101,9 @@ char const* IBuffer::getDataTypeName(DataType dataType)
     case nvinfer1::DataType::kINT8: return DataTypeTraits<nvinfer1::DataType::kINT8>::name;
     case nvinfer1::DataType::kFP8: return DataTypeTraits<nvinfer1::DataType::kFP8>::name;
     case nvinfer1::DataType::kINT4: [[fallthrough]] /* do nothing */;
-    case nvinfer1::DataType::kFP4: /* do nothing */;
+    case nvinfer1::DataType::kFP4: [[fallthrough]] /* do nothing */;
+    default: TLLM_THROW("Unknown data type");
     }
-    TLLM_THROW("Unknown data type");
 }
 
 char const* IBuffer::getDataTypeName() const
diff --git a/docker/Dockerfile.multi b/docker/Dockerfile.multi
index c832481da9..c41b600d1a 100644
--- a/docker/Dockerfile.multi
+++ b/docker/Dockerfile.multi
@@ -1,7 +1,7 @@
 # Multi-stage Dockerfile
-ARG BASE_IMAGE=nvcr.io/nvidia/pytorch
+ARG BASE_IMAGE=gitlab-master.nvidia.com:5005/dl/dgx/pytorch
 ARG TRITON_IMAGE=nvcr.io/nvidia/tritonserver
-ARG BASE_TAG=25.06-py3
+ARG BASE_TAG=25.08-py3.32224057-base
 ARG TRITON_BASE_TAG=25.06-py3
 ARG DEVEL_IMAGE=devel
 
@@ -74,8 +74,10 @@ ENV PYTORCH_CUDA_ALLOC_CONF="garbage_collection_threshold:0.99999"
 RUN pip3 uninstall -y opencv && rm -rf /usr/local/lib/python3*/dist-packages/cv2/
 RUN pip3 install opencv-python-headless --force-reinstall --no-deps --no-cache-dir
 
+COPY bringup_fix.sh bringup_fix.sh
+RUN bash ./bringup_fix.sh && rm bringup_fix.sh
+
 # WARs against security issues inherited from pytorch:25.06
-# * https://github.com/advisories/GHSA-8qvm-5x2c-j2w7
 RUN pip3 install --upgrade --no-cache-dir \
     "protobuf>=4.25.8"
 
@@ -103,7 +105,6 @@ RUN bash ./install_nixl.sh && rm install_nixl.sh
 COPY docker/common/install_etcd.sh install_etcd.sh
 RUN bash ./install_etcd.sh && rm install_etcd.sh
 
-
 FROM ${DEVEL_IMAGE} AS wheel
 WORKDIR /src/tensorrt_llm
 COPY benchmarks benchmarks
diff --git a/docker/common/install_tensorrt.sh b/docker/common/install_tensorrt.sh
index 6d118b62c4..e2e3c6218d 100644
--- a/docker/common/install_tensorrt.sh
+++ b/docker/common/install_tensorrt.sh
@@ -49,31 +49,31 @@ install_ubuntu_requirements() {
     rm cuda-keyring_1.1-1_all.deb
 
     apt-get update
-    if [[ $(apt list --installed | grep libcudnn9) ]]; then
-      apt-get remove --purge -y libcudnn9*
-    fi
-    if [[ $(apt list --installed | grep libnccl) ]]; then
-      apt-get remove --purge -y --allow-change-held-packages libnccl*
-    fi
-    if [[ $(apt list --installed | grep libcublas) ]]; then
-      apt-get remove --purge -y --allow-change-held-packages libcublas*
-    fi
-    if [[ $(apt list --installed | grep cuda-nvrtc-dev) ]]; then
-      apt-get remove --purge -y --allow-change-held-packages cuda-nvrtc-dev*
-    fi
+  #   if [[ $(apt list --installed | grep libcudnn9) ]]; then
+  #     apt-get remove --purge -y libcudnn9*
+  #   fi
+  #   if [[ $(apt list --installed | grep libnccl) ]]; then
+  #     apt-get remove --purge -y --allow-change-held-packages libnccl*
+  #   fi
+  #   if [[ $(apt list --installed | grep libcublas) ]]; then
+  #     apt-get remove --purge -y --allow-change-held-packages libcublas*
+  #   fi
+  #   if [[ $(apt list --installed | grep cuda-nvrtc-dev) ]]; then
+  #     apt-get remove --purge -y --allow-change-held-packages cuda-nvrtc-dev*
+  #   fi
 
-    CUBLAS_CUDA_VERSION=$(echo $CUDA_VER | sed 's/\./-/g')
-    NVRTC_CUDA_VERSION=$(echo $CUDA_VER | sed 's/\./-/g')
+  #   CUBLAS_CUDA_VERSION=$(echo $CUDA_VER | sed 's/\./-/g')
+  #   NVRTC_CUDA_VERSION=$(echo $CUDA_VER | sed 's/\./-/g')
 
-    apt-get install -y --no-install-recommends \
-        libcudnn9-cuda-12=${CUDNN_VER} \
-        libcudnn9-dev-cuda-12=${CUDNN_VER} \
-	libcudnn9-headers-cuda-12=${CUDNN_VER} \
-        libnccl2=${NCCL_VER} \
-        libnccl-dev=${NCCL_VER} \
-        libcublas-${CUBLAS_CUDA_VERSION}=${CUBLAS_VER} \
-        libcublas-dev-${CUBLAS_CUDA_VERSION}=${CUBLAS_VER} \
-        cuda-nvrtc-dev-${NVRTC_CUDA_VERSION}=${NVRTC_VER}
+  #   apt-get install -y --no-install-recommends \
+  #       libcudnn9-cuda-12=${CUDNN_VER} \
+  #       libcudnn9-dev-cuda-12=${CUDNN_VER} \
+	# libcudnn9-headers-cuda-12=${CUDNN_VER} \
+  #       libnccl2=${NCCL_VER} \
+  #       libnccl-dev=${NCCL_VER} \
+  #       libcublas-${CUBLAS_CUDA_VERSION}=${CUBLAS_VER} \
+  #       libcublas-dev-${CUBLAS_CUDA_VERSION}=${CUBLAS_VER} \
+  #       cuda-nvrtc-dev-${NVRTC_CUDA_VERSION}=${NVRTC_VER}
 
     apt-get clean
     rm -rf /var/lib/apt/lists/*
@@ -130,12 +130,17 @@ install_tensorrt() {
         if [ -z "$ARCH" ];then ARCH=$(uname -m);fi
         if [ "$ARCH" = "arm64" ];then ARCH="aarch64";fi
         if [ "$ARCH" = "amd64" ];then ARCH="x86_64";fi
-        RELEASE_URL_TRT="https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/${TRT_VER_SHORT}/tars/TensorRT-${TRT_VER}.Linux.${ARCH}-gnu.cuda-${TRT_CUDA_VERSION}.tar.gz"
+
+        if [ "$ARCH" = "x86_64" ]; then
+        RELEASE_URL_TRT="http://cuda-repo/release-candidates/Libraries/TensorRT/v10.14/10.14.0.19-6374d0f7/13.0-r580/Linux-x64-manylinux_2_28/tar/TensorRT-10.14.0.19.Linux.x86_64-gnu.cuda-13.0.tar.gz"
+        else
+        RELEASE_URL_TRT="http://cuda-repo/release-candidates/Libraries/TensorRT/v10.14/10.14.0.19-6374d0f7/13.0-r580/Linux-aarch64-manylinux_2_35/tar/TensorRT-10.14.0.19.Ubuntu-22.04.aarch64-gnu.cuda-13.0.tar.gz"
+        fi
     fi
 
     wget --no-verbose ${RELEASE_URL_TRT} -O /tmp/TensorRT.tar
     tar -xf /tmp/TensorRT.tar -C /usr/local/
-    mv /usr/local/TensorRT-${TRT_VER} /usr/local/tensorrt
+    mv /usr/local/TensorRT-* /usr/local/tensorrt
     pip3 install --no-cache-dir /usr/local/tensorrt/python/tensorrt-*-cp${PARSED_PY_VERSION}-*.whl
     rm -rf /tmp/TensorRT.tar
     echo 'export LD_LIBRARY_PATH=/usr/local/tensorrt/lib:$LD_LIBRARY_PATH' >> "${ENV}"
diff --git a/jenkins/current_image_tags.properties b/jenkins/current_image_tags.properties
index dee2ee7218..87a2808834 100644
--- a/jenkins/current_image_tags.properties
+++ b/jenkins/current_image_tags.properties
@@ -11,7 +11,7 @@
 #
 # NB: Typically, the suffix indicates the PR whose CI pipeline generated the images. In case that
 #     images are adopted from PostMerge pipelines, the abbreviated commit hash is used instead.
-LLM_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.06-py3-x86_64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202508051130-6090
-LLM_SBSA_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:pytorch-25.06-py3-aarch64-ubuntu24.04-trt10.11.0.33-skip-tritondevel-202508051130-6090
+LLM_DOCKER_IMAGE=gitlab-master.nvidia.com:5005/xiweny/images:gb110_bringup_x86_64
+LLM_SBSA_DOCKER_IMAGE=gitlab-master.nvidia.com:5005/xiweny/images:gb110_bringup_sbsa
 LLM_ROCKYLINUX8_PY310_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.1-devel-rocky8-x86_64-rocky8-py310-trt10.11.0.33-skip-tritondevel-202508051130-6090
 LLM_ROCKYLINUX8_PY312_DOCKER_IMAGE=urm.nvidia.com/sw-tensorrt-docker/tensorrt-llm:cuda-12.9.1-devel-rocky8-x86_64-rocky8-py312-trt10.11.0.33-skip-tritondevel-202508051130-6090
diff --git a/requirements.txt b/requirements.txt
index d5a3a8ecb9..9612f7d1f6 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -21,7 +21,7 @@ pandas
 h5py==3.12.1
 StrEnum
 sentencepiece>=0.1.99
-tensorrt~=10.11.0
+tensorrt
 # https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-25-06.html#rel-25-06 uses 2.8.0a0.
 torch>=2.7.1,<=2.8.0a0
 torchvision
diff --git a/triton_backend/inflight_batcher_llm/CMakeLists.txt b/triton_backend/inflight_batcher_llm/CMakeLists.txt
index 62da21994e..0f26015922 100644
--- a/triton_backend/inflight_batcher_llm/CMakeLists.txt
+++ b/triton_backend/inflight_batcher_llm/CMakeLists.txt
@@ -289,7 +289,7 @@ target_link_libraries(
 FetchContent_Declare(
   json
   GIT_REPOSITORY https://github.com/nlohmann/json.git
-  GIT_TAG v3.11.2)
+  GIT_TAG v3.12.0)
 
 FetchContent_MakeAvailable(json)