PhoneLLM Open-Sources Two Voice Models with Sub-100ms Latency

Pipecat released two open-weight PhoneLLM models, full-weight fine-tunes of NVIDIA Nemotron optimized for conversational voice AI, now available on Hugging Face and Cloudflare Workers AI, with PhoneLLM Mini (30B) achieving sub-100ms first-token latency.

2026-10-16 ~ 2026-10-16 · 2 related posts

• PhoneLLM launches open voice-AI models with sub-100ms latency, tops PhoneBench at 83% — charles_irl · 2026-10-16
• PhoneLLM Mini (30B) and Max (120B) target sub-100ms voice AI latency on Workers AI — ritakozlov · 2026-10-16