LLM Reference

Nemotron 3 VoiceChat

Multimodal

About

12B speech-to-speech model for low-latency full-duplex voice conversations

Capabilities

VisionMultimodalReasoningFunction CallingTool UseJSON ModeCode Execution

Rankings

Specifications

Released2026-03-16
Parameters12B
ArchitectureDecoder Only
Specializationgeneral