NVIDIA just shipped a 30B audio model that does everything. Nemotron Audex 30B-A3B handles: - Audio understanding, ASR, translation - Text-to-speech in one MoE model - Recommended inference setups per task Open weights on Hugging Face, ready for vLLM, SGLang, Transformers. https://t.co/q9rtTKihIH
@clementdelangue·Jul 12, 2026·2 sources·neutral
Read articleAI Summary
NVIDIA released the Nemotron Audex 30B-A3B audio model with open weights on Hugging Face.
All Sources
RT @so_sthbryan: NVIDIA just shipped a 30B audio model that does everything. Nemotron Audex 30B-A3B handles: - Audio understanding, ASR, t…
@clementdelangue
Jul 12, 2026
NVIDIA just shipped a 30B audio model that does everything. Nemotron Audex 30B-A3B handles: - Audio understanding, ASR, translation - Text-to-speech in one MoE model - Recommended inference setups per task Open weights on Hugging Face, ready for vLLM, SGLang, Transformers. https://t.co/q9rtTKihIH
@clementdelangue
Jul 12, 2026