NVIDIA delivers industry-leading gpt-oss-120b performance of 1.5M tokens per second on a single NVIDIA Blackwell GB200 NVL72 system, optimized for the world’s largest AI inference infrastructure.
Monday, August 11, 4-5 p.m. PT
Join NVIDIA AI research leaders as they chart the next frontier in computer graphics and physical AI.
Delivering 1.5 M TPS Inference on NVIDIA GB200 NVL72, NVIDIA accelerates OpenAI gpt-oss models enabling faster, more cost-effective AI inference deployment—from cloud to edge.
Groundbreaking open-weight models are now available with local optimizations for NVIDIA GeForce RTX and RTX PRO GPUs.
Dynamo adds support for popular AWS services, unlocking new levels of performance, scalability, and cost-efficiency for serving large language models.
The new AI infrastructure will include an NVIDIA AI Technology Center to foster local AI research, nurture talent, and drive innovation in Indonesia with NVIDIA Inception startups.
OpenAI, NVIDIA Propel AI Innovation With New Optimized Open Models
NVIDIA Research Special Address at SIGGRAPH
NVIDIA Accelerates OpenAI gpt-oss Models for Industry Leading Inference
OpenAI’s New Open-Source Models Accelerated on RTX AI PCs
NVIDIA Dynamo Delivers Cost-Efficient Inference at Scale With AWS
Indosat to Build AI Center of Excellence With Cisco and NVIDIA
NVIDIA CEO Jensen Huang takes the stage to share what’s next in AI factories, agentic AI, and physical AI powering the new industrial revolution. Watch Now
Deci was acquired by NVIDIA Corporation of Santa Clara, CA in May 2024 and was dissolved as a separate corporate entity. For legacy support content please refer to the following: Deci AI Documentation