Nvidia said Alpamayo 2 Super became available for commercial use on Hugging Face on August 4 under the OpenMDW-1.1 license for robotaxis and other autonomous vehicles.

The company described it as a 34 billion-parameter vision-language-action model for Level 4 autonomous driving.

It processes multi-camera video, navigation inputs, and driving context to generate trajectories, reasoning traces, meta-actions, scene question-and-answer responses, and auto-labels.

Nvidia said it is part of a wider Alpamayo stack for data labeling, training, simulation, and in-vehicle deployment.

OpenMDW-1.1 is a permissive model license from the Linux Foundation that allows use and redistribution for any purpose, while requiring distributors to keep the license and notices with model materials.

Model outputs carry no license obligations.

Nvidia said the model ranked first on LingoQA, a Wayve benchmark for visual question answering in autonomous driving with 28,000 scenarios and 419,000 annotations.

It also said Alpamayo 2 Super outperformed Qwen2.5-VL 72B, Gemini 2.5 Pro, and GPT-4o in its Lingo-Judge tests.

The release may matter to users of Nvidia’s autonomous vehicle platform.

Nvidia said DRIVE Hyperion works with Alpamayo and has been selected by automakers and robotaxi operators including Mercedes-Benz, Hyundai Motor Group, General Motors, Uber, Toyota, BYD, Geely, Volvo Cars, and JLR.

🔗 Source: Nvidia

Recent Nvidia developments