本地AI变得更加私人，但代价是什么？
评论者测试本地ai模型，发现兴奋、困惑和沮丧。隐私利益与高硬件成本相平衡。
a reviewer from the verge, antonio g. di benedetto, is experimenting with local ai models on an m5 ultra mac studio with 256gb of unified memory. they installed the open-source hermes agent, which allows for local llms, and are testing various models. the reviewer finds using local ai to be exciting, but also overwhelming and frustrating due to the numerous available models and their specialized uses. local ai offers better privacy than cloud services, but requires expensive hardware. this development could make ai more accessible to users prioritizing data privacy, but the high hardware costs may be a barrier.