alt Hacker News> Sorry to shatter your bubble, but this is patently false, LLMs are far more efficient on hardware that simultaneously serves many requests at once.
You might want to read this: https://arxiv.org/abs/2502.05317v2
> Sorry to shatter your bubble, but this is patently false, LLMs are far more efficient on hardware that simultaneously serves many requests at once.
You might want to read this: https://arxiv.org/abs/2502.05317v2