NVIDIA releases Nemotron 3 Ultra, a 550B open model

Hugging Face Models

Nemotron 3 Ultra has 550B total parameters (55B active) with a hybrid Mamba-Transformer MoE architecture and 1M context. It is optimized for long-running agents, claiming up to 5x faster inference and 30% lower cost, with open weights released under OpenMDW 1.1.

NvidiaNemotronMixture of experts

See 14 more sources

America's leading open-source model, Nemotron 3 Ultra, is now available on Perplexity for all Pro...16 days agoAravind Srinivas

Nemotron 3 Ultra is now available for Pro and Max subscribers on Perplexity and Computer. It's...16 days agoPerplexity AI

RT @alvarobartt: NVIDIA Nemotron 3 Ultra is here to supercharge your agents! 550B MoE, 55B active...17 days agoHugging Face

RT @HuggingPapers: NVIDIA just released Nemotron 3 Ultra on Hugging Face 550B total params, 55B...17 days agoHugging Face

[AINews] not much happened today17 days ago

Nemotron 3 Ultra and Latent MoE Scaling4 days agoSebastian Raschka

Introducing NVIDIA Nemotron 3 Ultra18 days agoNVIDIA

Nemotron 3 Ultra now available on AI Gateway17 days agoJerilyn Zheng

Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning6 days agoNVIDIA (Allan), : (Allan), Aaron Blakeman (Allan), Aaron Thomas (Allan), Aastha Jhunjhunwala (Allan), Abhibha Gupta (Allan), Abhinav Khattar (Allan), Adam Rajfer (Allan), Adi Renduchintala (Allan), Adil Asif (Allan), Aditya Vavre (Allan), Adriana Flores Mi

Nemotron Office Hours: The Nemotron 3 Model Family | Nemotron Labs3 days agoNVIDIA Developer

Ask the Experts: Nemotron 3 Ultra | Nemotron Labs16 days agoNVIDIA Developer

Nemotron 3 Ultra Tutorial: Build an Autonomous Research Agent with NemoHermes and OpenCode16 days agoNVIDIA Developer

nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 · Hugging Face18 days agojacek2023Discuss

NVIDIA announces Nemotron 3 Ultra21 days agothemixtergamesDiscuss

Hugging Face Models·nvidia

Jun 3, 2:54 PM