NVIDIA Launches Alpamayo, an Open Source AI Model Family for Autonomous Vehicles
During CES 2026, the American corporation unveiled a new family of tools aimed at the development of autonomous vehicles. According to the press release, Alpamayo includes artificial intelligence models, simulation tools, and datasets made available in open source.
Introducing Alpamayo 1
At the heart of this family lies Alpamayo 1, introduced by the company as the first VLA thought-chain reasoning model designed for autonomous vehicle research. The model utilizes a 10-billion parameter architecture and processes video inputs to generate trajectories with reasoning traces, elucidating the logic behind each decision. The model weights and open source inference scripts are available on Hugging Face, the company indicates. Developers can adapt Alpamayo 1 to create more compact models for vehicle development, or use it as a foundation for development tools such as reasoning-based evaluators and automatic labeling systems, NVIDIA specifies. The company announces that future models in the family will feature more detailed reasoning capabilities and commercial usage options.
AlpaSim and Comprehensive Datasets
The Alpamayo family also includes AlpaSim, presented as an end-to-end open source simulation framework for the development of high-fidelity autonomous vehicles. According to the press release, NVIDIA also provides datasets comprising over 1,700 hours of driving data covering complex scenarios, accessible via Hugging Face. The company states that these resources enable developers to refine, distill, and test models aimed at enhancing the safety, robustness, and scalability of autonomous systems. NVIDIA aims these tools at addressing the challenges posed by rare and complex situations, often referred to as long tail, which remain among the most difficult to master for autonomous systems.
Industry Leaders Show Interest
According to the press release, mobility leaders and industry experts, including Lucid, Jaguar Land Rover, Uber, and Berkeley DeepDrive, are showing interest in Alpamayo to develop reasoning-based autonomous vehicle architectures targeting level 4 autonomy. Kai Stepper, Vice President of ADAS and Autonomous Driving at Lucid Motors, states that advanced simulation environments, rich datasets, and reasoning models are important elements of this evolution. Thomas Müller, Executive Director of Product Engineering at JLR, emphasizes that the development of open and transparent AI is crucial for advancing autonomous mobility responsibly, adding that by making models like Alpamayo open source, NVIDIA is helping to accelerate innovation in the autonomous driving ecosystem.
