New AWS Instances Available!
ARM-based Graviton3 AWS instances are now available in the platform! Kick off a model packaging request to get a deployable container for these instances and assess whether you can achieve better performance per dollar on these instances compared to other CPUs.
OctoML regularly adds support for new hardware instances, to help you run automated benchmarking experiments and decide which hardware is the most optimal for your model(s) based on performance and price tradeoffs. You analyze these results easily on our Compare tab.