PhoneLLM Mini (30B) and Max (120B) target sub-100ms voice AI latency on Workers AI
ritakozlov · x · 2026-10-16
Pipecat's team launched two PhoneLLM models optimized for conversational voice AI, available on Cloudflare Workers AI: PhoneLLM Mini (30B) with sub-100ms time-to-first-audio, fitting an RTX 5090 and running 100+ concurrent agents per B200, and PhoneLLM Max (120B), a "microthinking" model with sub-200ms TTFAT and SOTA task completion at 1/10th the latency and cost of comparable models.
Both are full-weights fine-tunes of NVIDIA Nemotron models (Mini on Nemotron 3 Nano).
Related event: PhoneLLM Open-Sources Two Voice Models with Sub-100ms Latency(2 posts)→
More from Models
- Liquid AI's d1-ultra, #1 on Artificial Analysis Decision Index, lands on Vercel — JosephJacks_ · 2026-10-16
- Weekly AI roundup: Claude Haiku 5.5 cuts prices ~90%, Gemini 4 Argon output jumps to 1M tokens — DeepLearningAI · 2026-10-16
- Tencent Hunyuan says reported agent activity came from third-party model evals, not deliberate targeting — TencentHunyuan · 2026-10-16
- Law firm tests show Claude beating paid legal AI tools like Legora — jkubicki · 2026-10-16
- ChatGPT spontaneously generates rainbow persona calling itself a GPT-5.1 'Cognitive Construct' — SamsaraSiddhartha · 2026-10-16
- NVIDIA Nemotron 3.5 Bolt 74B-A9B Lands on Ollama, Fits 24GB via Native 2-bit Quantization — ollama · 2026-10-16