NVIDIA · chzblych · Jan 17, 2026 · Jan 17, 2026 · Jan 17, 2026
@@ -11,8 +11,8 @@ This branch is a prototype and not stable for production use. PRs are not accept
 [![Documentation](https://img.shields.io/badge/docs-latest-brightgreen.svg?style=flat)](https://nvidia.github.io/TensorRT-LLM/)
 [![python](https://img.shields.io/badge/python-3.12-green)](https://www.python.org/downloads/release/python-3123/)
 [![python](https://img.shields.io/badge/python-3.10-green)](https://www.python.org/downloads/release/python-31012/)
-[![cuda](https://img.shields.io/badge/cuda-13.0.0-green)](https://developer.nvidia.com/cuda-downloads)
-[![torch](https://img.shields.io/badge/torch-2.9.0-green)](https://pytorch.org)
+[![cuda](https://img.shields.io/badge/cuda-13.1.0-green)](https://developer.nvidia.com/cuda-downloads)
+[![torch](https://img.shields.io/badge/torch-2.9.1-green)](https://pytorch.org)
 [![version](https://img.shields.io/badge/release-1.3.0rc0-green)](https://github.com/NVIDIA/TensorRT-LLM/blob/main/tensorrt_llm/version.py)
 [![license](https://img.shields.io/badge/license-Apache%202-blue)](https://github.com/NVIDIA/TensorRT-LLM/blob/main/LICENSE)
 

@@ -9,15 +9,15 @@
    Before the pre-built Python wheel can be installed via `pip`, a few
    prerequisites must be put into place:
 
-   Install CUDA Toolkit 13.0 following the [CUDA Installation Guide for Linux](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)
+   Install CUDA Toolkit 13.1 following the [CUDA Installation Guide for Linux](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/)
    and make sure `CUDA_HOME` environment variable is properly set.
 
-   The `cuda-compat-13-0` package may be required depending on your system's NVIDIA GPU
+   The `cuda-compat-13-1` package may be required depending on your system's NVIDIA GPU
    driver version. For additional information, refer to the [CUDA Forward Compatibility](https://docs.nvidia.com/deploy/cuda-compatibility/forward-compatibility.html).
 
    ```bash
    # By default, PyTorch CUDA 12.8 package is installed. Install PyTorch CUDA 13.0 package to align with the CUDA version used for building TensorRT LLM wheels.
-   pip3 install torch==2.9.0 torchvision --index-url https://download.pytorch.org/whl/cu130
+   pip3 install torch==2.9.1 torchvision --index-url https://download.pytorch.org/whl/cu130
 
    sudo apt-get -y install libopenmpi-dev
 
@@ -39,6 +39,9 @@
    ```bash
    pip3 install --upgrade pip setuptools && pip3 install tensorrt_llm
    ```
+
+   > **Note:** The TensorRT LLM wheel on PyPI is built with PyTorch 2.9.1. This version may be incompatible with the NVIDIA NGC PyTorch 25.12 container, which uses a more recent PyTorch build from the main branch. If you are using this container or a similar environment, please install the pre-built wheel located at `/app/tensorrt_llm` inside the TensorRT LLM NGC Release container instead.
+
    **This project will download and install additional third-party open source software projects. Review the license terms of these open source projects before use.**
 
 2. Sanity check the installation by running the following in Python (tested on Python 3.12):

@@ -154,9 +154,9 @@ The following table shows the supported software for TensorRT-LLM.
 * -
   - Software Compatibility
 * - Container
-  - [25.10](https://docs.nvidia.com/deeplearning/frameworks/support-matrix/index.html)
+  - [25.12](https://docs.nvidia.com/deeplearning/frameworks/support-matrix/index.html)
 * - TensorRT
-  - [10.13](https://docs.nvidia.com/deeplearning/tensorrt/release-notes/index.html)
+  - [10.14](https://docs.nvidia.com/deeplearning/tensorrt/release-notes/index.html)
 * - Precision
   -
     - Blackwell (SM100/SM103/SM120) - FP32, FP16, BF16, FP8, FP4, INT8, INT4

@@ -454,7 +454,7 @@ def runLLMBuild(pipeline, buildFlags, tarName, is_linux_x86_64)
     def llmPath = sh (script: "realpath ${LLM_ROOT}",returnStdout: true).trim()
     // TODO: Remove after the cmake version is upgraded to 3.31.8
     // Get triton tag from docker/dockerfile.multi
-    def tritonShortTag = "r25.10"
+    def tritonShortTag = "r25.12"
     sh "cd ${LLM_ROOT}/triton_backend/inflight_batcher_llm && mkdir build && cd build && cmake .. -DTRTLLM_DIR=${llmPath} -DTRITON_COMMON_REPO_TAG=${tritonShortTag} -DTRITON_CORE_REPO_TAG=${tritonShortTag} -DTRITON_THIRD_PARTY_REPO_TAG=${tritonShortTag} -DTRITON_BACKEND_REPO_TAG=${tritonShortTag} -DUSE_CXX11_ABI=ON && make -j${buildJobs} install"
 
     // Step 3: packaging wheels into tarfile

@@ -1988,7 +1988,7 @@ def launchTestListCheck(pipeline)
             def llmPath = sh (script: "realpath .", returnStdout: true).trim()
             def llmSrc = "${llmPath}/TensorRT-LLM/src"
             trtllm_utils.llmExecStepWithRetry(pipeline, script: "pip3 install -r ${llmSrc}/requirements-dev.txt")
-            sh "NVIDIA_TRITON_SERVER_VERSION=25.10 LLM_ROOT=${llmSrc} LLM_BACKEND_ROOT=${llmSrc}/triton_backend python3 ${llmSrc}/scripts/check_test_list.py --l0 --qa --waive"
+            sh "NVIDIA_TRITON_SERVER_VERSION=25.12 LLM_ROOT=${llmSrc} LLM_BACKEND_ROOT=${llmSrc}/triton_backend python3 ${llmSrc}/scripts/check_test_list.py --l0 --qa --waive"
         } catch (InterruptedException e) {
             throw e
         } catch (Exception e) {
@@ -3431,7 +3431,9 @@ def launchTestJobs(pipeline, testFilter)
     ]
 
     aarch64SanityCheckConfigs = [
-        /* //Disable PY312-UB2404 temporarily since lack of official PyTorch for CUDA 13.1.
+        // Workaround PyTorch 2.9.1 vs. 2.10.0a0 incompatibility issue. Once resolved, change back to:
+        // 1. DLFW_IMAGE -> UBUNTU_24_04_IMAGE
+        // 2. Extra PyTorch CUDA install: false -> true
         "PY312-UB2404": [
             LLM_DOCKER_IMAGE,
             "GH200",
@@ -3440,7 +3442,7 @@ def launchTestJobs(pipeline, testFilter)
             "",
             DLFW_IMAGE,
             false, // Extra PyTorch CUDA 13.0 install
-        ],*/
+        ],
         "PY312-DLFW": [
             LLM_DOCKER_IMAGE,
             "GH200",