Pipecat PhoneLLM Alpha 1
Pipecat PhoneLLM Alpha 1A low-latency, open-weights LLM tuned for multi-turn voice agents, accurate tool calling, and conversational customer-support workflows.
Model details
View repositoryPhoneLLM Alpha 1 is an open-weights language model from Pipecat, optimized for low-latency, multi-turn voice-agent applications. It is designed for accurate tool calling and natural customer-support conversations, making it well suited to real-time voice AI systems that require responsive, context-aware interactions.
The model is a full-parameter fine-tune of NVIDIA Nemotron 3 Nano 30B-A3B, with 30B total parameters and 3.5B active parameters. It supports English and a 262,144-token context window, and can be served with vLLM on NVIDIA H100 GPUs. PhoneLLM Alpha 1 is released under the BSD 2-Clause license, with the underlying Nemotron model remaining subject to NVIDIA’s license and attribution requirements.
