# PyTorch + CUDA 11.6 **URL:** \ **Category:** deployment\ **Created:** [April 20, 2022, 12:15pm UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647 "2022-04-20T12:15:29Z")\ **Posts on this page:** 7\ **Page:** 1
**Author:** ![yellowishlight](https://discuss.pytorch.org/letter_avatar_proxy/v4/letter/y/9de0a6/32.png) [@yellowishlight](https://discuss.pytorch.org/u/yellowishlight)\ **Post date:** [April 20, 2022, 12:15pm UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/1 "2022-04-20T12:15:29Z")
Hello everyone, As a follow-up to this question [PyTorch + CUDA 11.4](https://discuss.pytorch.org/t/pytorch-cuda-11-4/137900) I have installed these Nvidia drivers version 510.60.02 along with Cuda 11.6. I’d like to install Pytorch in a conda virtual environment, and I’ve found in the Pytorch website that we couldn’t choose a stable version that relies on the latest versions of Cuda (the older version is 11.3) [Start Locally | PyTorch](https://pytorch.org/get-started/locally) How can I install/build Pytorch with the latest version of Cuda in my conda virtual env ? ---
**Author:** ![ptrblck](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/ptrblck/32/1823_2.png) [@ptrblck](https://discuss.pytorch.org/u/ptrblck)\ **Post date:** [April 21, 2022, 4:37am UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/2 "2022-04-21T04:37:47Z")
You could build PyTorch from source following [these instructions](https://github.com/pytorch/pytorch#from-source) or install the nightly binaries with CUDA11.6 via: ```python pip install torch --pre --extra-index-url https://download.pytorch.org/whl/nightly/cu116 ``` ---
**Author:** ![yellowishlight](https://discuss.pytorch.org/letter_avatar_proxy/v4/letter/y/9de0a6/32.png) [@yellowishlight](https://discuss.pytorch.org/u/yellowishlight)\ **Post date:** [April 21, 2022, 11:58am UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/3 "2022-04-21T11:58:33Z")
@ptrblck I installed it using the nightly binaries. Thanks a lot for your help ! ---
**Author:** ![Phillip\_J\_Murphy](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/phillip_j_murphy/32/49064_2.png) [@Phillip\_J\_Murphy](https://discuss.pytorch.org/u/Phillip_J_Murphy)\ **Post date:** [April 27, 2022, 6:02pm UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/4 "2022-04-27T18:02:21Z")
Thank you so much. Also worked for torchvision. ---
**Author:** ![skoenman](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/skoenman/32/50034_2.png) [@skoenman](https://discuss.pytorch.org/u/skoenman)\ **Post date:** [May 28, 2022, 7:29am UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/5 "2022-05-28T07:29:01Z")
Just for future ref one can get all packages under that link… torchserve aswell. ---
**Author:** ![jianlong\_Huang](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/jianlong_huang/32/49175_2.png) [@jianlong\_Huang](https://discuss.pytorch.org/u/jianlong_Huang)\ **Post date:** [August 30, 2022, 1:50am UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/7 "2022-08-30T01:50:52Z")
Hi, sir. I have one question: If I installed cuda 113 version pytorch, but my GPU is cuda 11.6. Is this mismatch of version affect the reproducibility of code. Since I set all random seeds and no deterministic algorithms used in code, but the training loss between different running times is still around 1%. ---
**Author:** ![ptrblck](https://discuss.pytorch.org/user_avatar/discuss.pytorch.org/ptrblck/32/1823_2.png) [@ptrblck](https://discuss.pytorch.org/u/ptrblck)\ **Post date:** [August 30, 2022, 3:48am UTC](https://discuss.pytorch.org/t/pytorch-cuda-11-6/149647/8 "2022-08-30T03:48:28Z")
No, your locally installed libraries should not introduce non-determinism as they should not be used since the PyTorch binaries ship with their own CUDA runtime, cuDNN, cublas etc. and will use these.