{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-23T06:40:50.084Z","headline":"NVIDIA 发布 Cosmos 3 Edge：4B 参数开源世界模型，为机器人及边缘 AI 提供实时推理与动作生成","description":"NVIDIA 在 Hugging Face 上开源了 Cosmos 3 Edge，一个 40 亿参数的世界模型，旨在帮助机器人和视觉 AI 智能体在边缘设备上理解环境、实时推理并生成动作。","url":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/","mainEntityOfPage":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/","datePublished":"2026-07-20T15:58:51.000Z","dateModified":"2026-07-20T15:58:51.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://huggingface.co/blog/nvidia/cosmos3edge","https://aihot.virxact.com/items/cmrtgmuyf3366bitl3gnpc8eo"],"canonicalUrl":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/","directAnswer":{"@type":"Answer","text":"NVIDIA 在 Hugging Face 发布 Cosmos 3 Edge。这是一款四十亿参数的开放世界模型，面向机器人和视觉 AI 智能体，可在边缘设备上理解环境、实时推理并生成机器人动作。","url":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/","dateCreated":"2026-07-20T15:58:51.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"huggingface.co source article","url":"https://huggingface.co/blog/nvidia/cosmos3edge","datePublished":"2026-07-20T15:58:51.000Z","provider":{"@type":"Organization","name":"huggingface.co","url":"https://huggingface.co/blog/nvidia/cosmos3edge"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrtgmuyf3366bitl3gnpc8eo","datePublished":"2026-07-20T15:58:51.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrtgmuyf3366bitl3gnpc8eo"}}],"aggregationSource":"Hugging Face：Blog（RSS）","originalPublisher":{"name":"huggingface.co","url":"https://huggingface.co/blog/nvidia/cosmos3edge"},"article":{"id":"cmrtgmuyf3366bitl3gnpc8eo","slug":"cmrtgmuyf3366bitl3gnpc8eo","url":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/","title":"NVIDIA 发布 Cosmos 3 Edge：4B 参数开源世界模型，为机器人及边缘 AI 提供实时推理与动作生成","title_en":"Introducing Cosmos 3 Edge","summary":"NVIDIA 在 Hugging Face 上开源了 Cosmos 3 Edge，一个 40 亿参数的世界模型，旨在帮助机器人和视觉 AI 智能体在边缘设备上理解环境、实时推理并生成动作。","source":"Hugging Face：Blog（RSS）","sourceUrl":"https://huggingface.co/blog/nvidia/cosmos3edge","aiHotUrl":"https://aihot.virxact.com/items/cmrtgmuyf3366bitl3gnpc8eo","publishedAt":"2026-07-20T15:58:51.000Z","category":"模型更新","score":70,"selected":true,"articleBody":["What Is World Modeling? ：#what-is-world-modeling Two Transformer Towers, One Shared Representation ：#two-transformer-towers-one-shared-representation A Common Representation for Action ：#a-common-representation-for-action Policy Mode ：#policy-mode Post-trained samples for improved performance ：#post-trained-samples-for-improved-performance Try Cosmos 3 Edge ：#try-cosmos-3-edge The real world is vast and to operate in it physical AI systems need to understand how a scene is changing, anticipate what could happen next, and determine how an action will affect the world.","Machines operate at the edge in factories, warehouses, hospitals and beyond. To understand and act in these environments, they need models that deliver data center–level performance on memory-constrained systems.","Today we are releasing NVIDIA Cosmos 3 Edge on Hugging Face’s Cosmos 3 repository. It’s a 4-billion-parameter open world model that helps robots and vision AI agents understand their surroundings, reason in real time and generate robot actions on edge devices.","Download the model: https://huggingface.co/nvidia/Cosmos3-Edge：https://huggingface.co/nvidia/Cosmos3-Edge","The model delivers memory-efficient, high-throughput inference across NVIDIA edge computers including NVIDIA RTX PRO GPUs, NVIDIA DGX, NVIDIA GeForce RTX™ GPUs, NVIDIA Jetson including the newly announced Jetson T2000 and T3000 modules.","Designed as a compact open model that can serve as a small vision language model (VLM) with best-in-class throughput and accuracy with real-time inference.","As a post-trained world action model (WAM), Cosmos 3 Edge operates at robot-control resolution (640×360 observations), delivering real-time reasoning and generating 32 actions per inference on NVIDIA Jetson Thor - while achieving real-time control at 15 Hz.","Among similar size (4B parameters) models, Cosmos 3 Edge ranks #1 on VANTAGE-Bench for vision analytics and state-of-the-art for robot policy learning, setting the state of the art for smart infrastructure and robotics.","A world model learns how an environment changes over time. It represents objects, motion, spatial relationships, and the effects of actions.","Consider a robot reaching for an object. Recognizing the object is only the first step. The robot must also understand where the object is, how its gripper is moving, what may happen when contact occurs, and which action is most likely to complete the task successfully.","A world model helps the robot reason about these relationships. It can predict the visual result of an action, infer the action that caused a change, or generate an action to produce a desired outcome.","Cosmos 3 Edge brings these capabilities together in one model, on-device. Its shared representation enables physical AI systems to understand the current world state, simulate possible futures, and connect those futures to actions.","Cosmos 3 combines two transformer towers:","The two towers maintain separate normalization layers and multilayer perceptrons. They share multimodal attention layers, which align information across language, video, audio, and action.","This design allows the model to reason about a scene before generating an output. Depending on the task, Cosmos 3 can produce reasoning tokens from the autoregressive tower or denoised video and action tokens from the diffusion tower.","The attention pattern is also adapted to each type of information. Language uses causal attention, where each token attends to the tokens that came before it. Diffusion tokens attend more broadly to the available context, supporting coherent prediction and generation.","Together, these components give the model a common representation of what is happening, what may happen next, and how an action could influence the outcome.","：https://cdn-uploads.huggingface.co/production/uploads/6a45813c717ace23b8c3b562/EXkYn60a1TSwkBXaoKmNh.png","Physical systems describe actions differently. A vehicle may represent an action through ego pose and movement. A camera uses camera motion. A robot arm uses the pose of its end effector, while a hand or gripper also needs to represent grasp state.","Cosmos 3 maps these different embodiments into a common action representation.","Actions are encoded as compact geometric vectors that capture:","This creates a direct connection between control and the visual structure of the world. The model can associate changes in pixels with physical motion, spatial relationships, and control inputs.","As a result, generated video becomes more than a visual prediction. It can represent how the world is expected to change in response to an action. This gives developers training data grounded in motion, control, and cause and effect.","As a policy, Cosmos 3 predicts an action together with its expected visual consequence.","The model can generate what the system should do and simulate what is likely to happen next. This connects world modeling directly to robot policy training and evaluation.","Current state in, action and visual consequence out.","These modes enable one model to learn cause, effect, and policy. Action can flow through the model in both directions, allowing Cosmos 3 Edge to predict the effects of an action or infer an action from its effects.","Prompt: Pick up the banana and put it in the plate.","We are also releasing Cosmos 3 Edge Policy (DROID): A robot manipulation policy post-trained on the DROID dataset for pick-and-place tasks, with accompanying post-training scripts.","Developers can use a small cluster of H100 or NVIDIA DGX Station to efficiently fine-tune Cosmos 3 Edge for their target workloads before deploying to NVIDIA edge and accelerated computing platforms.","Alongside the base models, we're releasing reference post-trained checkpoints and training recipes to help developers adapt and optimize Cosmos 3 for their own applications.","Cosmos 3 is an open world foundation model platform. As models are post-trained with high-quality, domain-specific data, their accuracy continue to improve for specialized applications. Cosmos serves as the starting point for building domain-adapted world models using open frameworks. Post-training enables developers to improve model performance while preserving output quality. To demonstrate this workflow, we're also releasing the Cosmos 3 Super 4-Step Distillation checkpoint along with a post-training script, providing a faster reference implementation for 64B model that helps developers adapt Cosmos to their own domains and achieve better downstream performance.","Cosmos 3 Super 4-Step (Text-to-Image & Image-to-Video): These are distribution-matching distilled checkpoint and scripts that reduce diffusion from 35–50 denoising steps to just 4, delivering up to 25× faster inference while preserving image and video quality and fidelity.","These examples serve as reference implementations that developers can build on. Using the open training scripts, you can post-train Cosmos 3 Edge for custom robot policies or distill Cosmos 3 Super for faster image and video generation tailored to your application.","Cosmos 3 Edge connects understanding, prediction, simulation, and action through a shared world representation.Developers can deploy it closer to their sensors on-device. The model can be used as a reasoner or as an action generator.","Use open Cosmos Frameworks to customize and specialize the models.","Checkout Cosmos 3: https://huggingface.co/collections/nvidia/cosmos3：https://huggingface.co/collections/nvidia/cosmos3","We continue to advance Cosmos 3 for Physical AI, with upcoming improvements in interactive world generation, driving scenario simulation, and robotics policies.","We are also investing in faster inference and more efficient post-training across a broader range of hardware, reducing the time and compute required to build downstream models.","This includes optimizing Cosmos 3 checkpoints with open inference and optimization frameworks such as vLLM, with more optimizations and developer tooling coming soon."],"articleImages":[{"sourceUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1605114051380-noauth.jpeg","alt":"","afterParagraph":0,"url":"/media/articles/cmrtgmuyf3366bitl3gnpc8eo/3ec7f542e3247cfe.webp"},{"sourceUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1649143001781-624bebf604abc7ebb01789af.jpeg","alt":"","afterParagraph":0,"url":"/media/articles/cmrtgmuyf3366bitl3gnpc8eo/a8acbfbb3df0c118.webp"},{"sourceUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/63107b18e87051f3e3e0f598/R9onir4Y0MZuq1jEWCZ2-.jpeg","alt":"","afterParagraph":0,"url":"/media/articles/cmrtgmuyf3366bitl3gnpc8eo/322b5e5764e17a08.webp"},{"sourceUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6395df81ff28fb463caa0191/YjDHw_KiboojHP12Vt2je.png","alt":"","afterParagraph":0,"url":"/media/articles/cmrtgmuyf3366bitl3gnpc8eo/084151ecd49fedf9.webp"}],"mediaStatus":"ok","articleBodyZh":["什么是世界建模？：#what-is-world-modeling 两个变换器塔，一个共享表示：#two-transformer-towers-one-shared-representation 动作的通用表示：#a-common-representation-for-action 策略模式：#policy-mode 后训练样本以提高性能：#post-trained-samples-for-improved-performance 试用 Cosmos 3 Edge：#try-cosmos-3-edge 真实世界广阔，而要在其中操作，物理AI系统需要理解场景如何变化，预测下一步可能发生的情况，并确定某个动作将如何影响世界。","机器在工厂、仓库、医院等边缘环境中运行。为了理解并在这些环境中执行操作，它们需要能够在受限内存系统上提供数据中心级性能的模型。","今天，我们在 Hugging Face 的 Cosmos 3 仓库上发布了 NVIDIA Cosmos 3 Edge。它是一个具有40亿参数的开放世界模型，帮助机器人和视觉AI代理理解周围环境、实时推理，并在边缘设备上生成机器人动作。","下载模型：https://huggingface.co/nvidia/Cosmos3-Edge：https://huggingface.co/nvidia/Cosmos3-Edge","该模型在 NVIDIA 边缘计算设备上提供内存高效、高吞吐量的推理，包括 NVIDIA RTX PRO GPU、NVIDIA DGX、NVIDIA GeForce RTX™ GPU、NVIDIA Jetson（包括新发布的 Jetson T2000 和 T3000 模块）。","设计为紧凑的开放模型，可作为小型视觉语言模型（VLM）使用，具有一流的吞吐量和精度，并支持实时推理。","作为后训练的世界动作模型（WAM），Cosmos 3 Edge 在机器人控制分辨率（640×360 观测）下运行，提供实时推理，并在 NVIDIA Jetson Thor 上每次推理生成32个动作，同时实现15 Hz的实时控制。","在相似规模（40亿参数）模型中，Cosmos 3 Edge 在 VANTAGE-Bench 的视觉分析中排名第一，并在机器人策略学习方面达到最先进水平，为智能基础设施和机器人技术设定了最新标准。","世界模型学习环境随时间的变化。它表示对象、运动、空间关系以及动作的效果。","考虑一个机器人去拿一个物体。识别物体只是第一步。机器人还必须理解物体的位置，它的抓取器如何移动，接触时可能发生的情况，以及哪种动作最可能成功完成任务。","世界模型帮助机器人推理这些关系。它可以预测一个动作的视觉结果，推断引起变化的动作，或者生成一个动作以产生期望的结果。","Cosmos 3 Edge 将这些能力整合到一个设备上的模型中。它的共享表示使物理 AI 系统能够理解当前世界状态，模拟可能的未来，并将这些未来与动作关联起来。","Cosmos 3 结合了两个变压器塔：","这两个塔保持独立的归一化层和多层感知器。它们共享多模态注意力层，用于跨语言、视频、音频和动作对信息进行对齐。","这种设计允许模型在生成输出之前先推理场景。根据任务不同，Cosmos 3 可以从自回归塔生成推理标记，或者从扩散塔生成去噪的视频和动作标记。","注意力模式也适应于每种类型的信息。语言使用因果注意力，每个标记关注前面的标记。扩散标记更广泛地关注可用上下文，支持一致的预测和生成。","这些组件共同赋予模型一个关于正在发生的事情、接下来可能发生的事情，以及一个动作如何影响结果的通用表示。","：https://cdn-uploads.huggingface.co/production/uploads/6a45813c717ace23b8c3b562/EXkYn60a1TSwkBXaoKmNh.png","物理系统以不同方式描述动作。一辆车可能通过自我姿势和移动来表示动作。相机使用相机运动。机器人手臂使用其末端执行器的位姿，而手或抓手还需要表示抓握状态。","Cosmos 3 将这些不同的体现映射到一个通用的动作表示中。","动作被编码为紧凑的几何向量，捕捉：","这在控制和世界的视觉结构之间创建了直接的联系。模型可以将像素的变化与物理运动、空间关系和控制输入相关联。","因此，生成的视频不仅仅是视觉预测。它可以表示世界在响应某个动作时预期会如何变化。这为开发者提供了以运动、控制以及因果关系为基础的训练数据。","作为一种策略，Cosmos 3 会预测一个动作及其预期的视觉后果。","模型可以生成系统应该采取的动作，并模拟接下来可能发生的情况。这将世界建模直接与机器人策略训练和评估连接起来。","输入当前状态，输出动作及视觉后果。","这些模式使得一个模型能够学习因果、效果和策略。动作可以双向流经模型，使得 Cosmos 3 Edge 可以预测动作的效果或从效果推断出动作。","提示：拿起香蕉并放到盘子里。","我们还发布了 Cosmos 3 Edge Policy（DROID）：一种针对搬运任务在 DROID 数据集上进行后训练的机器人操作策略，并附带后训练脚本。","开发者可以使用少量的 H100 或 NVIDIA DGX Station 集群，高效地微调 Cosmos 3 Edge 以适应目标工作负载，然后部署到 NVIDIA 边缘和加速计算平台。","除了基础模型，我们还发布了参考的后训练检查点和训练方案，帮助开发者调整和优化 Cosmos 3，以适应自己的应用场景。","Cosmos 3 是一个开放世界基础模型平台。随着模型使用高质量、特定领域的数据进行后训练，其在专业应用中的准确性会持续提升。Cosmos 作为使用开放框架构建领域适应世界模型的起点。后训练使开发者能够在保持输出质量的同时提升模型性能。为了展示这一工作流程，我们还发布了 Cosmos 3 超级四步蒸馏检查点以及后训练脚本，为 64B 模型提供更快的参考实现，帮助开发者将 Cosmos 适配到自己的领域并获得更好的下游性能。","Cosmos 3 超级 4 步骤（文本到图像 & 图像到视频）：这些是分布匹配蒸馏的检查点和脚本，将扩散过程从 35–50 步降到仅 4 步，在保持图像和视频质量及保真度的同时，实现最高 25 倍的推理加速。","这些示例作为参考实现，供开发者在此基础上构建。使用开放训练脚本，您可以对 Cosmos 3 Edge 进行后训练以定制机器人策略，或蒸馏 Cosmos 3 Super 以实现针对您的应用的快速图像和视频生成。","Cosmos 3 Edge 通过共享的世界表示连接理解、预测、模拟和行动。开发者可以将其部署在更靠近传感器的设备上。该模型可以用作推理器或动作生成器。","使用开放 Cosmos 框架自定义和专业化模型。","查看 Cosmos 3：https://huggingface.co/collections/nvidia/cosmos3：https://huggingface.co/collections/nvidia/cosmos3","我们将继续推进 Cosmos 3 在物理 AI 领域的发展，未来将在互动世界生成、驾驶场景模拟和机器人策略方面进行改进。","我们还在投资于更快的推理和在更广泛硬件上的更高效的后训练，从而减少构建下游模型所需的时间和计算量。","这包括使用开放推理和优化框架（如 vLLM）优化 Cosmos 3 检查点，同时更多优化和开发工具即将推出。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"NVIDIA 在 Hugging Face 发布 Cosmos 3 Edge。这是一款四十亿参数的开放世界模型，面向机器人和视觉 AI 智能体，可在边缘设备上理解环境、实时推理并生成机器人动作。","background":"公开材料称，物理 AI 系统需要理解场景变化、预判后续状态并评估动作影响。工厂、仓库和医院等边缘场景又受内存限制，因此需要兼顾推理性能与资源效率的模型。","viewpoint":"Aioga 判断，Cosmos 3 Edge 的关注点不只在参数规模，而在将视觉理解、世界建模和动作生成置于边缘计算环境。其实际价值仍需结合具体硬件、任务和独立测试评估。","implications":"材料称，该模型支持多类 NVIDIA 边缘计算设备，并在 Jetson Thor 上以六百四十乘三百六十观察分辨率运行，每次推理生成三十二个动作，实现十五赫兹实时控制。","nextStep":"值得关注后续是否出现可复现的部署报告与独立基准验证，包括不同设备上的内存占用、吞吐表现、任务准确性，以及机器人策略学习在真实场景中的稳定性。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-07-23T00:29:33.644Z","sourceHash":"6a6c3a335df462e6","review":{"approved":true,"groundedness":96,"clarity":92,"duplicationRisk":18,"blockingIssues":[],"notes":["候选内容准确区分了来源事实、Aioga 判断和待验证事项，未将观点冒充事实。","“支持多类 NVIDIA 边缘计算设备”是对来源所列 RTX PRO、DGX、GeForce RTX、Jetson 等硬件的概括；如需更精确，可直接列举代表性设备，但不影响批准。","可选补充来源中“在 VANTAGE-Bench 同规模模型中排名第一”等基准声明，并明确归因于发布方；当前未提及不构成事实错误。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":0,"checks":["schema","length","source-attribution","low-source-overlap","no-html","independent-ai-review"]}},"tags":["模型更新","Hugging Face：Blog（RSS）"],"translations":{"zh-CN":{"title":"NVIDIA 发布 Cosmos 3 Edge：4B 参数开源世界模型，为机器人及边缘 AI 提供实时推理与动作生成","summary":"NVIDIA 在 Hugging Face 上开源了 Cosmos 3 Edge，一个 40 亿参数的世界模型，旨在帮助机器人和视觉 AI 智能体在边缘设备上理解环境、实时推理并生成动作。","category":"模型更新","source":"huggingface.co","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA 发布 Cosmos 3 Edge：4B 参数开源世界模型，为机器人及边缘 AI 提供实时推理与动作生成 - Aioga AI资讯","description":"NVIDIA 在 Hugging Face 上开源了 Cosmos 3 Edge，一个 40 亿参数的世界模型，旨在帮助机器人和视觉 AI 智能体在边缘设备上理解环境、实时推理并生成动作。","url":"https://www.aioga.com/news/cmrtgmuyf3366bitl3gnpc8eo/"},"en":{"title":"NVIDIA Releases Cosmos 3 Edge: 4B Parameter Open-Source World Model, Providing Real-Time Inference and Action Generation for Robots and Edge AI","summary":"NVIDIA has open-sourced Cosmos 3 Edge on Hugging Face, a 4-billion-parameter world model designed to help robots and visual AI agents understand the environment, reason in real time, and generate actions on edge devices.","category":"Models","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA Releases Cosmos 3 Edge: 4B Parameter Open-Source World Model, Providing Real-Time Inference and Action Generation for Robots and Edge AI - Aioga AI News","description":"NVIDIA has open-sourced Cosmos 3 Edge on Hugging Face, a 4-billion-parameter world model designed to help robots and visual AI agents understand the environment, reason in real tim...","url":"https://www.aioga.com/en/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:25:14.139Z"},"ja":{"title":"NVIDIA、Cosmos 3 Edge を発表：40億パラメータのオープンワールドモデルが、ロボットおよびエッジ AI にリアルタイム推論と動作生成を提供","summary":"NVIDIAはHugging FaceでCosmos 3 Edgeをオープンソース化しました。これは40億パラメータの世界モデルで、ロボットやビジュアルAIエージェントがエッジデバイス上で環境を理解し、リアルタイムで推論し、行動を生成するのを支援することを目的としています。","category":"モデル更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA、Cosmos 3 Edge を発表：40億パラメータのオープンワールドモデルが、ロボットおよびエッジ AI にリアルタイム推論と動作生成を提供 - Aioga AIニュース","description":"NVIDIAはHugging FaceでCosmos 3 Edgeをオープンソース化しました。これは40億パラメータの世界モデルで、ロボットやビジュアルAIエージェントがエッジデバイス上で環境を理解し、リアルタイムで推論し、行動を生成するのを支援することを目的としています。","url":"https://www.aioga.com/ja/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:25:34.173Z"},"ko":{"title":"NVIDIA가 Cosmos 3 Edge를 발표: 40억 매개변수 오픈 소스 세계 모델로 로봇 및 엣지 AI에 실시간 추론과 행동 생성을 제공","summary":"NVIDIA는 Hugging Face에서 Cosmos 3 Edge를 오픈소스로 공개했으며, 이는 40억 개의 매개변수를 가진 세계 모델로, 로봇과 시각 AI 에이전트가 엣지 장치에서 환경을 이해하고, 실시간 추론하며, 행동을 생성할 수 있도록 돕기 위해 설계되었습니다.","category":"모델 업데이트","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA가 Cosmos 3 Edge를 발표: 40억 매개변수 오픈 소스 세계 모델로 로봇 및 엣지 AI에 실시간 추론과 행동 생성을 제공 - Aioga AI 뉴스","description":"NVIDIA는 Hugging Face에서 Cosmos 3 Edge를 오픈소스로 공개했으며, 이는 40억 개의 매개변수를 가진 세계 모델로, 로봇과 시각 AI 에이전트가 엣지 장치에서 환경을 이해하고, 실시간 추론하며, 행동을 생성할 수 있도록 돕기 위해 설계되었습니다.","url":"https://www.aioga.com/ko/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:26:18.913Z"},"es":{"title":"NVIDIA lanza Cosmos 3 Edge: modelo mundial de código abierto con 4 mil millones de parámetros, proporcionando inferencia en tiempo real y generación de acciones para robots e IA en el borde","summary":"NVIDIA ha abierto el código de Cosmos 3 Edge en Hugging Face, un modelo mundial de 40 mil millones de parámetros, diseñado para ayudar a los robots y agentes de IA visual a comprender el entorno, razonar en tiempo real y generar acciones en dispositivos periféricos.","category":"Modelos","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA lanza Cosmos 3 Edge: modelo mundial de código abierto con 4 mil millones de parámetros, proporcionando inferencia en tiempo real y generación de acciones para robots e IA en el borde - Aioga Noticias de IA","description":"NVIDIA ha abierto el código de Cosmos 3 Edge en Hugging Face, un modelo mundial de 40 mil millones de parámetros, diseñado para ayudar a los robots y agentes de IA visual a compren...","url":"https://www.aioga.com/es/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:26:15.634Z"},"fr":{"title":"NVIDIA lance Cosmos 3 Edge : modèle mondial open source de 4 milliards de paramètres, offrant une inférence en temps réel et la génération d'actions pour les robots et l'IA en périphérie","summary":"NVIDIA a open source Cosmos 3 Edge sur Hugging Face, un modèle mondial de 4 milliards de paramètres, conçu pour aider les robots et les agents d'IA visuelle à comprendre l'environnement, à raisonner en temps réel et à générer des actions sur les appareils périphériques.","category":"Modèles","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA lance Cosmos 3 Edge : modèle mondial open source de 4 milliards de paramètres, offrant une inférence en temps réel et la génération d'actions pour les robots et l'IA en périphérie - Aioga Actualités IA","description":"NVIDIA a open source Cosmos 3 Edge sur Hugging Face, un modèle mondial de 4 milliards de paramètres, conçu pour aider les robots et les agents d'IA visuelle à comprendre l'environn...","url":"https://www.aioga.com/fr/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:27:08.476Z"},"de":{"title":"NVIDIA veröffentlicht Cosmos 3 Edge: 4B Parameter Open-Source-Weltmodell, bietet Echtzeit-Inferenz und Aktionsgenerierung für Roboter und Edge KI","summary":"NVIDIA hat auf Hugging Face Cosmos 3 Edge veröffentlicht, ein Weltmodell mit 4 Milliarden Parametern, das Robotern und visuellen KI-Agenten helfen soll, die Umgebung auf Edge-Geräten zu verstehen, in Echtzeit zu schließen und Aktionen zu generieren.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA veröffentlicht Cosmos 3 Edge: 4B Parameter Open-Source-Weltmodell, bietet Echtzeit-Inferenz und Aktionsgenerierung für Roboter und Edge KI - Aioga KI-News","description":"NVIDIA hat auf Hugging Face Cosmos 3 Edge veröffentlicht, ein Weltmodell mit 4 Milliarden Parametern, das Robotern und visuellen KI-Agenten helfen soll, die Umgebung auf Edge-Gerät...","url":"https://www.aioga.com/de/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:27:00.484Z"},"pt-BR":{"title":"NVIDIA lança Cosmos 3 Edge: modelo mundial de código aberto com 4 bilhões de parâmetros, fornecendo inferência em tempo real e geração de ações para robôs e IA de borda","summary":"A NVIDIA lançou de forma aberta no Hugging Face o Cosmos 3 Edge, um modelo mundial com 4 bilhões de parâmetros, destinado a ajudar robôs e agentes de IA visual a compreender o ambiente, realizar inferências em tempo real e gerar ações em dispositivos de borda.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA lança Cosmos 3 Edge: modelo mundial de código aberto com 4 bilhões de parâmetros, fornecendo inferência em tempo real e geração de ações para robôs e IA de borda - Aioga Notícias de IA","description":"A NVIDIA lançou de forma aberta no Hugging Face o Cosmos 3 Edge, um modelo mundial com 4 bilhões de parâmetros, destinado a ajudar robôs e agentes de IA visual a compreender o ambi...","url":"https://www.aioga.com/pt-BR/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:27:50.337Z"},"ru":{"title":"NVIDIA выпустила Cosmos 3 Edge: открытая мировая модель с 4 миллиардами параметров, предоставляющая роботам и пограничному ИИ возможности для реального времени вывода и генерации действий","summary":"NVIDIA на Hugging Face открыла исходный код Cosmos 3 Edge, мировой модели с 4 миллиардами параметров, предназначенной для помощи роботам и визуальным ИИ-агентам в понимании окружающей среды на периферийных устройствах, выполнении расчетов в реальном времени и генерации действий.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA выпустила Cosmos 3 Edge: открытая мировая модель с 4 миллиардами параметров, предоставляющая роботам и пограничному ИИ возможности для реального времени вывода и генерации действий - Aioga Новости ИИ","description":"NVIDIA на Hugging Face открыла исходный код Cosmos 3 Edge, мировой модели с 4 миллиардами параметров, предназначенной для помощи роботам и визуальным ИИ-агентам в понимании окружаю...","url":"https://www.aioga.com/ru/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:27:56.103Z"},"ar":{"title":"أصدرت NVIDIA نظام Cosmos 3 Edge: نموذج عالمي مفتوح المصدر يحتوي على 4 مليارات معلمة، لتوفير الاستدلال الفوري وتوليد الحركة للروبوتات وذكاء الحافة","summary":"أطلقت NVIDIA على Hugging Face نموذج Cosmos 3 Edge مفتوح المصدر، وهو نموذج عالمي يحتوي على 4 مليارات معلم، يهدف إلى مساعدة الروبوتات ووكلاء الذكاء الاصطناعي البصري على فهم البيئة على الأجهزة الطرفية، والاستدلال في الوقت الفعلي، وتوليد الإجراءات.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"أصدرت NVIDIA نظام Cosmos 3 Edge: نموذج عالمي مفتوح المصدر يحتوي على 4 مليارات معلمة، لتوفير الاستدلال الفوري وتوليد الحركة للروبوتات وذكاء الحافة - Aioga أخبار الذكاء الاصطناعي","description":"أطلقت NVIDIA على Hugging Face نموذج Cosmos 3 Edge مفتوح المصدر، وهو نموذج عالمي يحتوي على 4 مليارات معلم، يهدف إلى مساعدة الروبوتات ووكلاء الذكاء الاصطناعي البصري على فهم البيئة عل...","url":"https://www.aioga.com/ar/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:28:43.216Z"},"hi":{"title":"NVIDIA ने Cosmos 3 Edge जारी किया: 4B पैरामीटर वाला ओपन-सोर्स विश्व मॉडल, रोबोट और एज AI के लिए रियल-टाइम इनफेरेंस और एक्शन जनरेशन प्रदान करता है","summary":"NVIDIA ने Hugging Face पर Cosmos 3 Edge को ओपन-सोर्स किया है, जो एक 40 बिलियन पैरामीटर वाला विश्व मॉडल है, जिसका उद्देश्य रोबोट और विज़ुअल एआई एजेंटों को एज डिवाइस पर वातावरण को समझने, रीयल-टाइम निष्कर्ष निकालने और क्रियाएं उत्पन्न करने में मदद करना है।","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA ने Cosmos 3 Edge जारी किया: 4B पैरामीटर वाला ओपन-सोर्स विश्व मॉडल, रोबोट और एज AI के लिए रियल-टाइम इनफेरेंस और एक्शन जनरेशन प्रदान करता है - Aioga AI समाचार","description":"NVIDIA ने Hugging Face पर Cosmos 3 Edge को ओपन-सोर्स किया है, जो एक 40 बिलियन पैरामीटर वाला विश्व मॉडल है, जिसका उद्देश्य रोबोट और विज़ुअल एआई एजेंटों को एज डिवाइस पर वातावरण को सम...","url":"https://www.aioga.com/hi/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:28:53.528Z"},"it":{"title":"NVIDIA lancia Cosmos 3 Edge: modello mondiale open source con 4 miliardi di parametri, per fornire inferenza in tempo reale e generazione di azioni a robot e AI edge","summary":"NVIDIA ha reso open source su Hugging Face Cosmos 3 Edge, un modello mondiale da 4 miliardi di parametri, progettato per aiutare robot e agenti di intelligenza artificiale visiva a comprendere l'ambiente, fare inferenze in tempo reale e generare azioni sui dispositivi edge.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA lancia Cosmos 3 Edge: modello mondiale open source con 4 miliardi di parametri, per fornire inferenza in tempo reale e generazione di azioni a robot e AI edge - Aioga Notizie IA","description":"NVIDIA ha reso open source su Hugging Face Cosmos 3 Edge, un modello mondiale da 4 miliardi di parametri, progettato per aiutare robot e agenti di intelligenza artificiale visiva a...","url":"https://www.aioga.com/it/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:29:48.839Z"},"nl":{"title":"NVIDIA lanceert Cosmos 3 Edge: 4 miljard parameters open source wereldmodel, biedt realtime inferentie en actiegeneratie voor robots en edge AI","summary":"NVIDIA heeft op Hugging Face Cosmos 3 Edge open source gemaakt, een wereldmodel met 4 miljard parameters, bedoeld om robots en visuele AI-agents te helpen om de omgeving op randapparaten te begrijpen, in realtime redeneervaardigheden toe te passen en acties te genereren.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA lanceert Cosmos 3 Edge: 4 miljard parameters open source wereldmodel, biedt realtime inferentie en actiegeneratie voor robots en edge AI - Aioga AI-nieuws","description":"NVIDIA heeft op Hugging Face Cosmos 3 Edge open source gemaakt, een wereldmodel met 4 miljard parameters, bedoeld om robots en visuele AI-agents te helpen om de omgeving op randapp...","url":"https://www.aioga.com/nl/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:29:38.264Z"},"tr":{"title":"NVIDIA Cosmos 3 Edge'i Duyurdu: 4B Parametreli Açık Kaynak Dünya Modeli, Robotlar ve Kenar AI için Gerçek Zamanlı Çıkarım ve Hareket Üretimi Sağlıyor","summary":"NVIDIA, Hugging Face üzerinde Cosmos 3 Edge adlı, 40 milyar parametreli bir dünya modeli açık kaynak yaptı; bu model, robotların ve görsel yapay zeka ajanlarının çevreyi kenar cihazlarda anlamalarına, gerçek zamanlı olarak akıl yürütmelerine ve eylemler üretmelerine yardımcı olmayı amaçlıyor.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA Cosmos 3 Edge'i Duyurdu: 4B Parametreli Açık Kaynak Dünya Modeli, Robotlar ve Kenar AI için Gerçek Zamanlı Çıkarım ve Hareket Üretimi Sağlıyor - Aioga AI Haberleri","description":"NVIDIA, Hugging Face üzerinde Cosmos 3 Edge adlı, 40 milyar parametreli bir dünya modeli açık kaynak yaptı; bu model, robotların ve görsel yapay zeka ajanlarının çevreyi kenar ciha...","url":"https://www.aioga.com/tr/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:30:44.781Z"},"vi":{"title":"NVIDIA phát hành Cosmos 3 Edge: mô hình thế giới 4 tỷ tham số mã nguồn mở, cung cấp suy luận thời gian thực và tạo hành động cho robot và AI biên","summary":"NVIDIA đã mở mã nguồn Cosmos 3 Edge trên Hugging Face, một mô hình thế giới với 4 tỉ tham số, nhằm giúp robot và các thực thể AI thị giác hiểu môi trường trên các thiết bị biên, suy luận thời gian thực và tạo ra hành động.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA phát hành Cosmos 3 Edge: mô hình thế giới 4 tỷ tham số mã nguồn mở, cung cấp suy luận thời gian thực và tạo hành động cho robot và AI biên - Tin tức AI Aioga","description":"NVIDIA đã mở mã nguồn Cosmos 3 Edge trên Hugging Face, một mô hình thế giới với 4 tỉ tham số, nhằm giúp robot và các thực thể AI thị giác hiểu môi trường trên các thiết bị biên, su...","url":"https://www.aioga.com/vi/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:30:41.462Z"},"id":{"title":"NVIDIA merilis Cosmos 3 Edge: Model dunia open-source dengan 4 miliar parameter, menyediakan inferensi waktu nyata dan generasi aksi bagi robot dan AI tepi","summary":"NVIDIA telah merilis Cosmos 3 Edge secara open source di Hugging Face, sebuah model dunia dengan 4 miliar parameter, yang dirancang untuk membantu robot dan agen AI visual memahami lingkungan, melakukan penalaran secara real-time, dan menghasilkan tindakan di perangkat edge.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA merilis Cosmos 3 Edge: Model dunia open-source dengan 4 miliar parameter, menyediakan inferensi waktu nyata dan generasi aksi bagi robot dan AI tepi - Berita AI Aioga","description":"NVIDIA telah merilis Cosmos 3 Edge secara open source di Hugging Face, sebuah model dunia dengan 4 miliar parameter, yang dirancang untuk membantu robot dan agen AI visual memahami...","url":"https://www.aioga.com/id/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:31:30.127Z"},"th":{"title":"NVIDIA เปิดตัว Cosmos 3 Edge: โมเดลโลกโอเพนซอร์ส 4B พารามิเตอร์ สำหรับหุ่นยนต์และ AI ที่ขอบเครือข่าย ให้การอนุมานและการสร้างการกระทำแบบเรียลไทม์","summary":"NVIDIA เปิดเผย Cosmos 3 Edge บน Hugging Face ซึ่งเป็นโมเดลโลกที่มีพารามิเตอร์ 4 พันล้านตัว โดยมีจุดประสงค์เพื่อช่วยให้หุ่นยนต์และเอเจนต์ AI ด้านการมองเห็นเข้าใจสภาพแวดล้อมบนอุปกรณ์ขอบ เผยเหตุผลแบบเรียลไทม์ และสร้างการกระทำ","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA เปิดตัว Cosmos 3 Edge: โมเดลโลกโอเพนซอร์ส 4B พารามิเตอร์ สำหรับหุ่นยนต์และ AI ที่ขอบเครือข่าย ให้การอนุมานและการสร้างการกระทำแบบเรียลไทม์ - ข่าว AI Aioga","description":"NVIDIA เปิดเผย Cosmos 3 Edge บน Hugging Face ซึ่งเป็นโมเดลโลกที่มีพารามิเตอร์ 4 พันล้านตัว โดยมีจุดประสงค์เพื่อช่วยให้หุ่นยนต์และเอเจนต์ AI ด้านการมองเห็นเข้าใจสภาพแวดล้อมบนอุปกรณ์...","url":"https://www.aioga.com/th/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:31:37.257Z"},"pl":{"title":"NVIDIA wprowadza Cosmos 3 Edge: otwarty na 4 miliardy parametrów model świata, zapewniający robotom i AI brzegowemu w czasie rzeczywistym wnioskowanie i generowanie akcji","summary":"NVIDIA udostępniła na Hugging Face Cosmos 3 Edge, światowy model o 4 miliardach parametrów, mający na celu pomoc robotom i inteligentnym agentom wizualnym w rozumieniu środowiska, wykonywaniu wnioskowań w czasie rzeczywistym i generowaniu działań na urządzeniach brzegowych.","category":"模型更新","source":"Hugging Face：Blog（RSS）","aggregationSource":"Hugging Face：Blog（RSS）","pageTitle":"NVIDIA wprowadza Cosmos 3 Edge: otwarty na 4 miliardy parametrów model świata, zapewniający robotom i AI brzegowemu w czasie rzeczywistym wnioskowanie i generowanie akcji - Aioga Wiadomości AI","description":"NVIDIA udostępniła na Hugging Face Cosmos 3 Edge, światowy model o 4 miliardach parametrów, mający na celu pomoc robotom i inteligentnym agentom wizualnym w rozumieniu środowiska,...","url":"https://www.aioga.com/pl/news/cmrtgmuyf3366bitl3gnpc8eo/","contentTranslated":true,"sourceHash":"0c588c0a0d9a12f2","translatedAt":"2026-07-22T17:32:23.425Z"}}}}