Overview
- Nvidia publicly released Alpamayo 2 Super as an open 32-billion-parameter vision-language-action model designed to handle perception, reasoning, planning, and action for Level 4 robotaxi systems.
- The model more than triples the size of Nvidia’s earlier Alpamayo family, moving from about 10 billion parameters to 32 billion to increase multimodal reasoning and decision capability.
- Nvidia bundled the model into its autonomous vehicle toolkit by pairing it with the DRIVE Hyperion reference architecture, the AlpaSim simulator, and curated physical AI datasets to support training and verification.
- Early ecosystem work has already begun, with verification and tooling partners such as Foretellix and operators like PlusAI starting integrations and scenario-based validation around the Alpamayo family.
- By open-licensing the model and providing simulation and data pipelines, Nvidia aims to lower barriers for automakers and startups to build and test Level 4 robotaxi software and to position its stack as a common foundation for the industry.