Alpamayo R1

Alpamayo R1 is NVIDIA's first open reasoning vision-language-action (VLA) model designed for autonomous vehicles. It is part of the broader NVIDIA Physical AI Stack and sits within NVIDIA's open Alpamayo portfolio — a collection of reasoning VLA models targeting the autonomous driving domain. [NVIDIA AI Models 2026 Guide]

A defining capability of Alpamayo R1 is its ability to allow vehicles to understand their surroundings and explain their reasoning in natural language to passengers, closing the gap between black-box decision-making and interpretable, communicative autonomy. [NVIDIA AI Models 2026 Guide]

• Reasoning VLA architecture — combines visual perception, language understanding, and action generation in a single model pipeline.
• Natural-language explanation — vehicles can articulate driving decisions to passengers in plain language.
• Level 4 autonomy target — designed to support the development of fully autonomous (no human fallback required) vehicle systems.
• AlpaSim integration — paired with AlpaSim, NVIDIA's fully open simulation blueprint for high-fidelity AV testing, allowing teams to validate systems in simulation before physical deployment. [NVIDIA AI Models 2026 Guide]

Alpamayo R1 runs on the NVIDIA DRIVE platform and is built to leverage the Rubin Hardware Platform infrastructure. It is part of the same NVIDIA Open Model Ecosystem that includes GR00T N1.7 for robotics and Cosmos 2.5 for physical AI simulation. [NVIDIA AI Models 2026 Guide]

Several major automotive and research organizations are building toward Level 4 autonomy using Alpamayo R1 and AlpaSim: [NVIDIA AI Models 2026 Guide]

Mercedes-Benz's all-new CLA is notable as the first production vehicle planned to ship with Alpamayo on the NVIDIA DRIVE platform. [NVIDIA AI Models 2026 Guide]

In the autonomous driving AI space, Alpamayo R1 is positioned as the leading open-source reasoning VLA for AV. The primary commercial competitor in terms of deployed autonomous miles is Waymo (Google/Alphabet), which leads on real-world deployment but operates as a closed-source, proprietary system. Alpamayo R1's open nature is therefore a significant differentiator for automotive OEMs and research teams that require transparency and customizability. [NVIDIA AI Models 2026 Guide]

This open-source advantage is consistent with NVIDIA's broader strategy described in the NVIDIA Open Model Ecosystem — releasing models freely to drive adoption of NVIDIA hardware.

Alpamayo R1 is available free on Hugging Face, GitHub, and NVIDIA Build (build.nvidia.com) under NVIDIA's commercial-friendly open model licenses (Apache 2.0, MIT, or NVIDIA Open Model License). Commercial use is permitted. [NVIDIA AI Models 2026 Guide]

For managed cloud deployment, Alpamayo R1 can be accessed via NVIDIA NIM (Inference Microservices), reducing the need for teams to manage their own GPU infrastructure. See also NVIDIA Build Inference Endpoints for API-based access options.

Alpamayo R1 is one of three pillars in NVIDIA's physical AI strategy alongside GR00T N1.7 (humanoid robotics) and Cosmos 2.5 (world simulation). Together, these form the NVIDIA Physical AI Stack — NVIDIA's thesis that the next frontier of AI is not text generation but AI operating in the physical world. [NVIDIA AI Models 2026 Guide]