Is it possible to use a GPU type not listed in Vertex AI Custom Job documentation?

Google Cloud Build with AI Custom ML & MLOps
, ,
1

Hi everyone,

I’m running custom training jobs on Vertex AI, and I’d like to use a specific GPU type that is not mentioned in the official Vertex AI documentation (for example, a newer or less common GPU model).

Is it possible to request or enable access to such a GPU through Vertex AI Custom Jobs — for example via a support request, quota modification, or partner manager? Or are we strictly limited to the GPU types explicitly listed in the Vertex AI supported machine types and accelerators documentation?

If there’s any official process or exception path to gain access to additional GPU types (e.g., private preview or regional availability), I’d appreciate any details or links.

Thanks!