Possible to upgrade CUDA in the Lambda Stack?

post by stepheno-iambic on Nov 26, 2025

We’re using the Lambda Stack on a Vector One workstation. The Lambda Stack currently has CUDA 12.8 according to what we have running and according to the Lambda Stack website: Lambda Stack: an AI software stack that's always up-to-date

Does anyone know if it is possible to install CUDA 12.9 over the Lambda Stack, that is keeping the rest of the Lambda Stack software but just upgrading the CUDA driver?

post by Nicholas_C on Dec 1, 2025

You can move the /etc/apt/preferences.d/lambda-repository file or change the priority level within the file to lower it as needed. Additionally, you can create a python virtual environment and customize the environment to your needs while still inheriting the site packages. You can refer to our docs page on the venv: Virtual environments and Docker containers - Lambda Docs

post by stepheno-iambic on Dec 2, 2025

Great, thanks! I’ll try this.