"Nvidia driver too old" error on deployment of the cog model to replicate.

[HG2407]

I am getting Nvidia driver too old when I deployed my model on a40 large gpu. It seems like it is an issue in the deployed machine and not on my end. Should I try changing gpus ?

[HG2407]

On my end, the nvidia driver (525.85.12) is compatible with the used cuda version (11.8).