무슨 발표인가

• 340억 파라미터 비전-언어-액션(VLA) 모델로 추론·계획·행동 수행
• AlpaGym 강화학습으로 주행 결정 시뮬레이션 검증
• Cosmos-Dreams로 포토리얼 주행 시나리오 생성

원문 (영어)

Editor's note: This post, originally published May 31, 2026, has been updated to reflect NVIDIA renamed NVIDIA OmniDreams to NVIDIA Cosmos-Dreams. News Summary: NVIDIA’s most powerful open reasoning model to date, NVIDIA Alpamayo 2 Super is an open 34-billion-parameter reasoning VLA model that reasons, plans and acts across the full driving stack for safer, scalable level 4 development.

NVIDIA AlpaGym is a new high-throughput, closed-loop reinforcement learning framework that trains AV models on the consequences of their driving decisions in simulation before road deployment. NVIDIA Cosmos-Dreams is a new generative world model for photorealistic closed-loop AV scenario generation, enabling developers to simulate rare and long-tail driving scenarios at scale.

NVIDIA physical AI agent skills for AV development include Neural Reconstruction powered by NVIDIA Omniverse NuRec, enabling developers to reconstruct real-world fleet data into photorealistic 3D scenes and adapt them across vehicle sensor configurations.

NVIDIA GTC Taipei — NVIDIA today introduced NVIDIA Alpamayo 2 Super, a 34-billion-parameter reasoning‑based vision language action (VLA) model that extends the NVIDIA Alpamayo family of open AI models, simulation frameworks and physical AI datasets for safe, level 4 robotaxi development.

Alongside the model, the company announced new tools, models and agent skills that complete the pipeline from real-world data capture to closed-loop training and in-vehicle deployment, including NVIDIA AlpaGym, NVIDIA Cosmos-Dreams and new NVIDIA Omniverse NuRec models .

Alpamayo 2 Super helps accelerate autonomous vehicle (AV) development by eliminating the need to build key autonomy infrastructure from scratch.

원문: NVIDIA News — "NVIDIA Launches Alpamayo 2 Super Open Reasoning Model for Robotaxis" (2026-08-06) 공식 원문: https://nvidianews.nvidia.com/news/nvidia-alpamayo-2-super-robotaxis