Nvidia unveils AI agent tools, open model at GTC Taipei

Nvidia Announces Enterprise AI Agent Tools and Nemotron 3 Ultra Model at GTC Taipei

Nvidia used its GTC Taipei keynote on June 1 to unveil a suite of enterprise AI agent software and its largest open model to date, positioning the company as a full-stack provider for the emerging market of autonomous digital coworkers.

Agent Toolkit and NemoClaw for Autonomous Engineers

The company announced an expanded Agent Toolkit that includes NemoClaw blueprints, the OpenShell secure runtime, and CUDA-X libraries accessible as agent skills — together forming an open-source foundation for enterprises to build, orchestrate, and deploy long-running autonomous agents.nvidia

“The world’s software leaders are bringing AI agents into the systems where work gets done — showing how AI coworkers help employees think faster and execute complex tasks to solve bigger problems,” CEO Jensen Huang said during the keynote.nvidia

Cadence, Dassault Systèmes, Siemens, and Synopsys are among the first to adopt NemoClaw, building autonomous AI engineers that can compress weeks of simulation and verification work into hours. Cadence is using OpenShell to secure its ChipStack AI Super Agent for autonomous chip design verification, while Siemens is integrating the tools into its Fuse EDA AI Agent for semiconductor and PCB system design.nvidia

Nemotron 3 Ultra: 550 Billion Parameters for Agentic AI

Nvidia also unveiled Nemotron 3 Ultra, a 550-billion-parameter mixture-of-experts model with approximately 55 billion active parameters per token, using a hybrid Mamba-Transformer architecture. The company claims the model delivers 5x faster inference and up to 30 percent lower cost compared with open frontier models in its class.artificialanalysis

The model supports a 1-million-token context window and is post-trained for leading agent platforms including LangChain Deep Agents, OpenHands, and OpenCode. CrowdStrike is deploying Nemotron models for vulnerability remediation agents, while Palantir is integrating them into its AI Forward Deployed Engineer platform for autonomous task execution. Nemotron 3 Ultra is expected to be available on June 4 via Hugging Face, ModelScope, OpenRouter, and build.nvidia.com.coaleypeak

DGX Station for Windows

Rounding out the announcements, Nvidia introduced DGX Station for Windows, a deskside AI supercomputer built on the GB300 Grace Blackwell Ultra Desktop Superchip, capable of running frontier models up to 1 trillion parameters locally. The system features up to 748GB of coherent memory and 20 petaflops of FP4 performance, developed in collaboration with Microsoft and expected to ship in Q4 from partners including Dell, HP, and ASUS.globenewswire