ainewsblitz.com

Breaking

NVIDIA Releases Nemotron 3.5 Lightning, an Open MoE Model Built for Long-Running AI Agents That Runs on a Single GPU

NVIDIA Releases Nemotron 3.5 Lightning, an Open MoE Model Built for Long-Running AI Agents That Runs on a Single GPU
  • AI Agents
  • Open Source
  • Foundation Models

NVIDIA has released Nemotron 3.5 Lightning, an openly licensed 30-billion-parameter Mixture-of-Experts model designed to power the high-volume execution work of always-on AI agents, with weights, training data and recipes published under a permissive license and support for running on a single GPU. The model, launched on August 11, 2026, is available for free download, fine-tuning and modification on Hugging Face and can scale from a workstation-class device such as DGX Spark up to full data-center deployments.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.

$20
Read this article
$29/month
Unlimited — all 27,885 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year