Allocating GPU cluster to calculation manually by user

In live sessions with stakeholders and “future design” functionalities performance can be very important. It is possible that the “sprint GPU” is far superior in performance (50x faster) then the “heavy or super” GPU. Tygron however picks the GPU cluster itself. There is no way to allocate a “slower” GPU version, like the sprint, to increase the performance. Being able to do so would be great for these kind of live sessions