CybernewsOpinion

Experiment: I ran a datacenter-class 120B parameter model for just $800

A cheap $800 computer can run a 120 billion parameter AI model, GPT-OSS-120B, locally at over 10 tokens per second, utilizing ...