Alpamayo 2 Super: The Open Reasoning Model for Robotaxis

Tuesday, June 16, 12:00pm - 1:00pm (EDT)

Link:

Time shown in-04:00America, New York

NVIDIA Alpamayo 2 Super is here — a 32-billion-parameter open reasoning VLA model that doesn't just drive, it reasons, plans, and acts across the full autonomous vehicle stack. In this livestream, NVIDIA experts will share how the full Alpamayo platform — including AlpaGym and the Physical AI Dataset — closes the loop from real-world data to in-vehicle deployment. If you're working on robotaxis or L4 enabled autonomy, this is the livestream you won’t want to miss.

Top 3 things attendees will learn:

What's new in Alpamayo 2 Super — including the jump to 32B parameters, full 360-degree surround perception, and Meta-Action outputs — and why it matters.
How AlpaGym's closed-loop reinforcement learning framework trains models on the consequences of their driving decisions, before they ever touch a real road.
How to go from real-world fleet data to in-vehicle inference on DRIVE AGX Thor using distillation recipes — with live demos of the full pipeline.

Add to Calendar2026/06/16 09:00:002026/06/16 10:00:00America/Los_AngelesAlpamayo 2 Super: The Open Reasoning Model for RobotaxisNVIDIA Alpamayo 2 Super is here — a 32-billion-parameter open reasoning VLA model that doesn't just drive, it reasons, plans, and acts across the full autonomous vehicle stack. In this livestream, NVIDIA experts will share how the full Alpamayo platform — including AlpaGym and the Physical AI Dataset — closes the loop from real-world data to in-vehicle deployment. If you're working on robotaxis or L4 enabled autonomy, this is the livestream you won’t want to miss.

Top 3 things attendees will learn:

What's new in Alpamayo 2 Super — including the jump to 32B parameters, full 360-degree surround perception, and Meta-Action outputs — and why it matters.
How AlpaGym's closed-loop reinforcement learning framework trains models on the consequences of their driving decisions, before they ever touch a real road.
How to go from real-world fleet data to in-vehicle inference on DRIVE AGX Thor using distillation recipes — with live demos of the full pipeline.

https://www.youtube.com/watch?v=kJRVwaYwvt0Amelia Ngamelian@nvidia.comfalse30

https://www.youtube.com/watch?v=kJRVwaYwvt0