Apple Silicon costs more than OpenRouter | Hacker News

Hacker News debated a claim that Apple Silicon local LLM inference costs more than OpenRouter, with critics saying the analysis inflated power and hardware costs.

Commenters said the comparison unfairly treated a consumer Mac as if it were a 24/7 cloud server and ignored differences in utilization and workload mix.

Supporters of local inference pointed to privacy, offline use, model control, and cases where input tokens dominate costs.

Skeptics countered that Apple Silicon is often too slow for interactive workloads, while datacenter GPUs and large VRAM still win for heavy use.