For applications that aren't latency sensitive, I run inference on a free 4 core Ampere server from Oracle. Once you ditch the "fast" prerequisite, a lot of har

For applications that aren't latency sensitive, I run inference on a free 4 core Ampere server from Oracle. Once you ditch the "fast" prerequisite, a lot of hardware becomes viable.