---
格式版本: 2
标题: "NVIDIA Cosmos"
原文链接: "https://www.nvidia.com/en-us/ai/cosmos/"
发布日期: "2026-07-20"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "llm:scrape:original_meta_loose"
发布时间证据: "nv-pub-date: 2026-07-20T10:41:31.000Z"
发布时间校准原因: "该候选来自HTML元数据中的nv-pub-date，明确表示文章发布时间，且无其他更高优先级候选。"
发布时间校准置信度: "1"
发布时间候选数量: 6
发布时间严格候选数量: 0
发布时间原页读取状态: "原页面已读取"
发布时间未找到原因: ""
发布时间校准时间: "2026-07-26T21:11:12+08:00"
发布时间仲裁状态: "confirmed"
发布时间仲裁尝试次数: 1
发布时间仲裁耗时毫秒: 4976
发现时间: "2026-07-26T20:51:34+08:00"
入库时间: "2026-07-26T13:11:18.353Z"
来源平台: "NVIDIA 站内搜索"
搜索渠道: "source_template"
搜索词: "https://www.nvidia.com/en-us/search/?q=CPO&page=1"
匹配关键词:
  - "CPO"
  - "performance"
相关厂家:
  - "NVIDIA"
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 40
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "页面为NVIDIA Cosmos物理AI模型介绍，聚焦软件/模型，仅提及GB200作为运行平台，无超节点、机柜级架构、CPO等相关技术细节，纯属模型营销页，与项目主题明显无关。"
AI质检模型: "deepseek-v4-flash"
AI质检时间: "2026-07-27T11:07:57+08:00"
AI主题相关性: 2
AI来源权威性: 15
AI新颖性: 10
AI技术细节: 0
AI商业部署信号: 5
AI完整性: 8
采集批次: "2026年7月26日2点27分21秒"
采集批次ID: "20260726-022721-515"
去重键: "https://www.nvidia.com/en-us/ai/cosmos"
---

![](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNkYAAAAAYAAjCB0C8AAAAASUVORK5CYII=)

Develop physical AI faster with leading world foundation models and open data processing, training, and evaluation frameworks.

Cosmos 3

## The Open Physical AI Foundation Model

The first [omni-model](https://www.nvidia.com/en-us/glossary/omni-model/) with native reasoning, world and action generation. Built on [Mixture-of-Transformers](https://www.nvidia.com/en-us/glossary/mixture-of-transformers/).

### Power Vision AI Reasoning

Use as a vision language model (VLM) to reason over objects, interactions, and intent across complex real-world scenarios.

For real-time alerts and dense captioning across quality inspection, public safety, traffic monitoring, logistics, and autonomous driving.

### Build Policy Models

Accelerate robot policy learning with NVIDIA Cosmos™ 3 as the backbone for World Action Models (WAMs).

Post-train the generalized world foundation model on specialized camera and embodiment data. The policy model adapts pre-learned actions to specific tasks, domains, and behaviors at scale.

### Simulate Worlds

Run as a controllable, physics-grounded world simulator to predict multiple approaches, evaluate outcomes in a closed loop, and converge on the right behavior.

Scale the loop across environments, tasks, and conditions to continuously improve without real-world risk.

### Scale Synthetic Video Data

Generate infinite plausible futures from text, image, video, ambient sound and action input.

Use video generation as imagination to train physical AI without being constrained by what's been physically captured.

Video

## Introducing Cosmos

Meet NVIDIA Cosmos—the frontier of physical AI. Built on a new mixture-of-transformers architecture, this open omnimodel brings together pixels, actions, sound, and language to reason, plan, and generate what comes next.

Starting Options

## Get Started With NVIDIA Cosmos

#### 1

Ready to build? Access open models and code directly.

#### 2

Not ready to build yet? Try Cosmos models in our hosted catalog.

#### 3

Need help? Start quickly with our hands-on model recipes.

## Develop With Cosmos

Build on the same technology powering Cosmos 3. Open frameworks and skills so developers worldwide can customize, extend, and contribute to physical AI.

### Data Curation

Quickly filter, annotate, and deduplicate large amounts of sensor data with Cosmos Curator.

Review and score generative video outputs at scale using Cosmos Evaluator.

### Training and Acceleration

Quickly build, post-train or deploy world models using open post-training, evaluation, optimization frameworks, and inference scripts and skills.

### Agent Skills for Synthetic Data Generation

Turn coding agents into synthetic data experts for physical AI development.

Use Cases

## How Cosmos Accelerates AI Across Industries

Use Cosmos WFMs to simulate, reason, and generate data for downstream pipelines in robotics, autonomous vehicles, and industrial vision systems.

Performance

## Runs Best On NVIDIA AI

Cosmos 3 is optimized for the best performance on NVIDIA hardware. NVIDIA RTX PRO™ 6000 Blackwell Series Servers accelerate physical AI development for robots, autonomous vehicles, and AI agents across training, synthetic data generation, simulation, and inference.

Unlock peak performance for Cosmos world foundation models on NVIDIA Blackwell GB200 for industrial post-training and inference workloads.

![NVIDIA GB200 Grace Blackwell Superchip](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_copy_co_650699617/nv_image.coreimg.jpeg/1784544089434/grace-gb200-1920x1080.jpeg "NVIDIA GB200 Grace Blackwell Superchip")

Ecosystem

## Adopted by Leading Physical AI Innovators

Model developers from the robotics, autonomous vehicles, and vision AI industries are using Cosmos to accelerate physical AI development.

![1X Technologies logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image.coreimg.svg/1784544089912/1x-logo.svg "1X Technologies logo") ![Agile Robots logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1418432116.coreimg.svg/1784544089921/agibot-logo.svg "Agile Robots logo") ![Agility Robotics logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_53093496_co.coreimg.svg/1784544089936/agility-robotics-logo.svg "Agility Robotics logo") ![Ambient AI](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_53093496_co_1932789907.coreimg.svg/1784544089945/ambient-ai-logo.svg "Ambient AI") ![Avathon](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_53093496.coreimg.svg/1784544089953/avathon-logo.svg "Avathon") ![Carla](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1754986427__1442917183.coreimg.svg/1784544089962/carla-logo.svg "Carla") ![Centific](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1754986427__1761464654.coreimg.svg/1784544089970/centific-logo.svg "Centific") ![Dyna](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1754986427__1576635871.coreimg.svg/1784544089979/dyna-robotics-logo.svg "Dyna") ![Field AI](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1754986427_.coreimg.svg/1784544089987/field-ai-logo.svg "Field AI") ![Figure AI logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1754986427.coreimg.svg/1784544089996/figure-ai-logo.svg "Figure AI logo") ![Foretellix logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_843066756.coreimg.svg/1784544090004/foretellix-logo.svg "Foretellix logo") ![Galbot logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1991113721.coreimg.svg/1784544090012/galbot-logo.svg "Galbot logo") ![Gatik](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868__45777570.coreimg.svg/1784544090020/gatik-logo.svg "Gatik") ![General Motors logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868__659702528.coreimg.svg/1784544090029/logo-general-motors.svg "General Motors logo") ![Hexagon](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868__1498989880.coreimg.svg/1784544090037/hexagon-logo.svg "Hexagon") ![Hitachi](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_947552108_c_1826531645.coreimg.svg/1784544090045/hitachi-logo-no-slogan.svg "Hitachi") ![IntBot logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_947552108_c.coreimg.svg/1784544090054/intbot-logo.svg "IntBot logo") ![Inverted AI](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_947552108.coreimg.svg/1784544090062/inverted-ai-logo.svg "Inverted AI") ![Li Auto](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1260815404.coreimg.svg/1784544090072/li-auto-logo.svg "Li Auto") ![Linker Vision](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_348982109_c_1651896451.coreimg.svg/1784544090081/linker-vision-logo.svg "Linker Vision") ![Magna](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_348982109_c_1101289669.coreimg.svg/1784544090089/magna-logo.svg "Magna") ![Mentee Robotics](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_348982109_c_1591669415.coreimg.svg/1784544090098/mentee-robotics-logo.svg "Mentee Robotics") ![Milestone Systems](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_348982109_c.coreimg.svg/1784544090106/milestone-logo-black.svg "Milestone Systems") ![Neura Robotics logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_2026556266.coreimg.svg/1784544090113/neura-robotics-logo.svg "Neura Robotics logo") ![Nexar](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1950570993__1903627401.coreimg.svg/1784544090123/logo-nexar.svg "Nexar") ![Oxa](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1950570993__803945144.coreimg.svg/1784544090131/logo-oxa.svg "Oxa") ![Parallel Domain](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1950570993__2055541261.coreimg.svg/1784544090139/logo-parallel-domain.svg "Parallel Domain") ![PlusAI](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1950570993_.coreimg.svg/1784544090147/plusai-logo.svg "PlusAI") ![Skild AI logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_235957032.coreimg.svg/1784544090156/skild-ai-logo.svg "Skild AI logo") ![Toyota Research Institute](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_828023784.coreimg.svg/1784544090163/toyota-research-institute-logo.svg "Toyota Research Institute") ![Tulip](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_2067393670__713820409.coreimg.svg/1784544090172/tulip-interfaces-logo.svg "Tulip") ![Uber logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_2067393670_.coreimg.svg/1784544090180/uber-logo.svg "Uber logo") ![VAST Data](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_2067393670.coreimg.svg/1784544090188/vast-logo.svg "VAST Data") ![Virtual Incision logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_881929779.coreimg.svg/1784544090197/virtual-incision-logo.svg "Virtual Incision logo") ![VorWerk logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_881929779_c_1603534746.coreimg.svg/1784544090205/vorwerk-logo.svg "VorWerk logo") ![Voxel51](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_881929779_c.coreimg.svg/1784544090213/gtc24-spring-partner-logo-voxel51-1250x703.svg "Voxel51") ![Wistron logo](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868_.coreimg.svg/1784544090225/logo-wistron.svg "Wistron logo") ![Xiaomi](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868__185879760.coreimg.svg/1784544090233/xiaomi-alt-logo.svg "Xiaomi") ![X-humanoid](https://www.nvidia.com/content/nvidiaGDC/us/en_US/ai/cosmos/_jcr_content/root/responsivegrid/nv_container_2060460585/nv_container/nv_image_1628383868.coreimg.svg/1784544090242/x-humanoid-logo.svg "X-humanoid")

Resources

## The Latest From Cosmos Developers

## Frequently Asked Questions

### What’s new in NVIDIA Cosmos architecture?

Cosmos 3 is built on Mixture of Transformers architecture. Reasoning and generator modules use different transformers for efficient generation and performance. The model therefore reasons first and then generates, resulting in leading physics accuracy across capabilities. Learn more about the architecture [here](https://www.nvidia.com/en-us/glossary/mixture-of-transformers/).

### What is the licensing model for Cosmos world foundation models?

Cosmos WFMs are available under the OpenMDW1.1 license from Linux Foundation.

### How do I post-train Cosmos models for my downstream applications?

Cosmos 3 is openly available with post-training scripts on GitHub for each modality and module. In addition, NVIDIA TAO 7 provides a suite of agent skills and tools for fine-tuning vision AI models, including Cosmos 3, with coding agents and natural language prompts.

### Can I build a world model from scratch using tools from the Cosmos platform and my custom or in-house foundation model?

Yes, you can leverage Cosmos to build from scratch with your preferred foundation model or model architecture. You can start by using [Cosmos Curator](https://github.com/nvidia-cosmos/cosmos-curate) for video data preprocessing. Then compress and decode your data with [Cosmos tokenizer](https://developer.nvidia.com/blog/state-of-the-art-multimodal-generative-ai-model-development-with-nvidia-nemo/). Once you have processed the data, you can train or fine-tune your model.

Using [NVIDIA NIM™ microservices](https://build.nvidia.com/search?q=%22cosmos%22+-nemotron), you can easily integrate your physical AI models into your applications across cloud, data centers, and workstations.

You can also use [NVIDIA DGX Cloud](https://www.nvidia.com/en-us/data-center/dgx-cloud/?ncid=pa-srch-goog-542457-DGX-Brand-prsp&_bt=719595959246&_bk=nvidia%20dgx%20cloud&_bm=b&_bn=g&_bg=169380731029&gad_source=1&gclid=EAIaIQobChMI5ofZgLyvigMVlNXCBB0Vsgu8EAAYASAAEgKPbvD_BwE) to train AI models and deploy them at scale anywhere.

### What’s the difference between Cosmos 3 and previous Cosmos models?

Cosmos 3 is an omni-model it can generate across text, image, video, sound and action. While Cosmos 2.5 and Cosmos 2 kept perception and generation as separate models and modalities were limited to text, image and video.

### What is the difference between Cosmos and Omniverse?

Omniverse creates realistic 3D simulations of real-world tasks by using different generative APIs, SDKs, and NVIDIA RTX rendering technology.

Developers can input Omniverse simulations as instructional videos into Cosmos Transfer models to generate controllable, photorealistic synthetic data.

Together, Omniverse provides the simulation environment before and after training, while Cosmos provides the foundation models for generating video data and training physical AI models.

Learn more about [NVIDIA Omniverse](https://www.nvidia.com/en-us/omniverse/).