NVIDIA released the open-source model Nemotron 3 Ultra, built on a Hybrid Transformer-Mamba MoE architecture with 550B total parameters and 55B active parameters. It supports up to 1M tokens of context and targets long-horizon autonomous agent workflows, with Day-0 support from vLLM.
Key Takeaways
NVIDIA released the open-source model Nemotron 3 Ultra
built on a Hybrid Transformer-Mamba MoE architecture with 550B total parameters and 55B active…
It supports up to 1M tokens of context and targets long-horizon autonomous agent workflows,with…
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: NVIDIA releases open-source reasoning model Nemotron 3 Ultra, with Day-0 support from vLLM. June 4, 2026. https://new.nat6.net/en/news/cmtynzyln033nrog085edd7v1/