NVIDIA 发布了 Alpamayo 2 Super，这是一个 34B 参数的视觉-语言-动作（VLA）模型，面向自动驾驶，采用开放商业许可。其既定设计目标是长尾事件：即传统检测与预测堆栈处理不佳的罕见多智能体场景。该模型将一个 32B VLM 主干（基于 NVIDIA Cosmos 3 Super Reasoner 构建，并通过强化学习进行后训练）与一个 2.3B 的基于扩散模型的动作解码器配对。通过对全向环视摄像头视频进行一次处理，它即可输出规划轨迹、该轨迹的因果解释以及一个元动作。
它可以部署吗
可以，而且从第一天起就可用于商业用途。权重以 OpenMDW-1.1 发布，这是 Linux 基金会针对开放模型分发的宽松许可证；源代码采用 Apache 2.0。该许可证涵盖微调、衍生模型和商业再分发。NVIDIA 正在将 OpenMDW 应用于整个 Alpamayo 系列，因此此前为研发而推出的版本现在无需额外许可即可商业部署。
输入、输出与训练数据
输入为多摄像头 RGB 视频、文本以及带时间戳的自运动历史。经过验证的公开 notebook 配置使用六个摄像头，每个摄像头四帧历史帧。自运动为 3D 平移加上一个 3×3 旋转矩阵，多时间步。
轨迹 API 返回 64 个路点，覆盖 0.1 至 6.4 秒，间隔为 0.1 秒。每个路点携带自车坐标系 XYZ 和一个 3×3 旋转矩阵。
训练数据约为 115,000 小时的多摄像头驾驶视频，附带自车运动与轨迹标注。其中包含约 3,700,000 条因果链（Chain-of-Causation，CoC）轨迹——即对驾驶决策的结构化、因果关联解释。图像训练数据超过十亿张图像。
基准测试
在 LingoQA 上，Alpamayo 2 Super 取得 79.2 的 Lingo-Judge 分数，在近 40 个受评估模型中排名第一。在 NVIDIA 的测试中，它比 Qwen2.5-VL 72B 高出 17.0 分，比 Gemini 2.5 Pro 高出 15.1 分，比 GPT-4o 高出 23.2 分。
还有两个数字对规划工作很重要。使用 AlpaSim 在 PhysicalAI-AV-NuRec 数据集的 910 个场景上进行闭环评估，得到 AlpaSim 分数为 1.50 ± 0.13。在 PhysicalAI-AV 数据集的 937 个挑战性样本上进行开环评估，6.4 秒处的 minADE₆ 为 0.911m。
一个模型产生五种输出
对于每一种驾驶情境，该模型都会生成一条轨迹、一条解释决策的 CoC 轨迹、一个诸如让行或变道之类的元动作、推理自动标注，以及带 2D 定位的视觉问答。
这种组合正是该发布在运营层面引人关注的原因。开发者可以将模型观察到的内容与其选择的动作关联起来。CoC 轨迹可与 NVIDIA Halos 安全验证工作流集成，并支持符合 ISO/PAS 8800 的 AI 安全。
NVIDIA 表示，将其用作专有车队数据的自动标注器时，该模型可将标注周期从数月压缩至数天。
交互式讲解
核心要点
- 34B VLA 模型——32B Cosmos 3 Super Reasoner 主干网络加上一个 2.3B 扩散动作专家。
- OpenMDW-1.1 权重与 Apache 2.0 代码；允许商业使用和再分发，无需额外许可。
- LingoQA Lingo-Judge 79.2，在近 40 个模型中排名第一；AlpaSim 1.50 ± 0.13；6.4s 时 minADE₆ 0.911m。
- 一次前向推理即可产出轨迹、因果链（Chain-of-Causation）轨迹、元动作、自动标注以及有依据的 VQA。
- 云端规模模型在 1× H100 80GB 上测试，峰值 72,115 MiB；可将其蒸馏用于车载推理。
查看NVIDIA 博客和 Hugging Face 模型卡。另外，欢迎在 Twitter 上关注我们，别忘了加入我们拥有 15 万+ 成员的 ML SubReddit，并订阅我们的新闻通讯。等等！你在 Telegram 上吗？现在你也可以在 Telegram 上加入我们了。
NVIDIA has released Alpamayo 2 Super, a 34B-parameter vision-language-action (VLA) model for autonomous driving, under an open commercial license. The stated design target is the long-tail events: rare, multi-agent situations that conventional detection-and-prediction stacks handle poorly. The model pairs a 32B VLM backbone, built on NVIDIA Cosmos 3 Super Reasoner and post-trained with reinforcement learning, with a 2.3B diffusion-based action decoder. From one pass over full-surround camera video it emits a planned trajectory, a causal explanation of that trajectory, and a meta-action.
Is it deployable
Yes, and for commercial use from day one. The weights are released under OpenMDW-1.1, the Linux Foundation’s permissive license for open model distributions; source code is Apache 2.0. The license covers fine-tuning, derivative models and commercial redistribution. NVIDIA is applying OpenMDW across the entire Alpamayo family, so earlier releases introduced for R&D are now deployable commercially without additional permission.
Inputs, outputs and training data
Inputs are multi-camera RGB video, text, and egomotion history with timestamps. The validated public notebook profiles use six cameras and four historical frames per camera. Egomotion is 3D translation plus a 3×3 rotation matrix, multi-timestep.
The trajectory API returns 64 waypoints spanning 0.1 to 6.4 seconds at 0.1-second intervals. Each waypoint carries ego-frame XYZ and a 3×3 rotation matrix.
Training data is roughly 115,000 hours of multi-camera driving video with egomotion and trajectory annotations. It includes about 3,700,000 Chain-of-Causation (CoC) traces — structured, causally linked explanations of driving decisions. Image training data exceeds one billion images.
Benchmarks
On LingoQA, Alpamayo 2 Super records a Lingo-Judge score of 79.2 and ranks first among nearly 40 models evaluated. In NVIDIA’s testing it beat Qwen2.5-VL 72B by 17.0 points, Gemini 2.5 Pro by 15.1, and GPT-4o by 23.2.
Two more numbers matter for planning work. Closed-loop evaluation with AlpaSim on 910 scenarios from the PhysicalAI-AV-NuRec dataset gives an AlpaSim score of 1.50 ± 0.13. Open-loop evaluation on 937 challenging samples from the PhysicalAI-AV dataset gives minADE₆ at 6.4s of 0.911m.
Five outputs from one model
For each driving situation, the model produces a trajectory, a CoC trace explaining the decision, a meta-action such as yield or lane change, reasoning auto-labels, and visual question answering with 2D grounding.
That combination is what makes the release interesting operationally. Developers can tie what the model observed to the action it chose. CoC traces integrate with NVIDIA Halos safety-validation workflows and support AI safety aligned with ISO/PAS 8800.
Used as an autolabeler on proprietary fleet data, NVIDIA says the model compresses annotation cycles from months to days.
Interactive explainer
Key Takeaways
- 34B VLA model — 32B Cosmos 3 Super Reasoner backbone plus a 2.3B diffusion action expert.
- OpenMDW-1.1 weights and Apache 2.0 code; commercial use and redistribution allowed, no extra permission needed.
- LingoQA Lingo-Judge 79.2, first among nearly 40 models; AlpaSim 1.50 ± 0.13; minADE₆ 0.911m at 6.4s.
- One pass yields trajectory, Chain-of-Causation trace, meta-action, auto-labels, and grounded VQA.
- Cloud-scale model tested on 1× H100 80GB at 72,115 MiB peak; distill it for in-car inference.
Check out theNVIDIA blog and Hugging Face model card. Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.