>>105978580
posting from my ai workstation
it costs less than 1 gpu out of 8 that you need to run these models :)