Reading up on Smallest.ai

Smallest AI argues that specialized small models operating on compressed representations, rather than monolithic large models, are the future of real-time voice AI.

A developer built a sub-500ms latency voice agent by co-locating and aggressively pipelining STT, LLM, and TTS stages. Commenters debate trade-offs vs. end-to-end models, endpoint detection alternatives, and the difficulty of human-like turn-taking.