[KoolKat23]

Definitely not when it comes to LLM's, the larger more useful local models are not that fast and latency is not an issue, just look at this Google models voice function or even openai's advanced voice.