Cosmos Embed1 NIM is an NVIDIA Inference Microservice (NIM) that provides access to the **Cosmos-Embed1** joint video-text embedding model via a standard HTTP API.
Cosmos Embed1 NIM Overview
Description:
Cosmos Embed1 NIM is an NVIDIA Inference Microservice (NIM) that provides access to the Cosmos-Embed1 joint video-text embedding model via a standard HTTP API. It generates aligned vector embeddings for both text and short-form videos in a unified vector space, enabling a variety of applications including:
- Text-to-video retrieval
- Inverse video search
- Semantic deduplication of content
- Zero-shot and k-NN classification
- Automated video curation pipelines
- Content clustering and classification
- Embeddings for other downstream tasks
The container is purpose-built by NVIDIA for Physical AI applications such as robotics and autonomous vehicles, and encapsulates the model along with all necessary pre/post-processing steps into a GPU-accelerated microservice.
The NIM leverages CV-CUDA, PyNvVideoCodec, Triton Inference Server, and TensorRT for optimized preprocessing and inference performance.
The container components are ready for commercial/non-commercial use.
NIM Documentation
Please visit the Cosmos-Embed1 NIM Documentation for full details on all NIM features and how to get started.
License/Terms of Use
The NIM container is governed by the NVIDIA Software License Agreement (found at https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-software-license-agreement/) and the Product-Specific Terms for NVIDIA AI Products (found at https://www.nvidia.com/en-us/agreements/enterprise-software/product-specific-terms-for-ai-products/), and use of this model is governed by the NVIDIA Community Model License. Additional Information: Apache License 2.0; MIT.
Deployment Geography:
Global
Release Date:
NGC: 09/10/2025
Program Classes:
The Cosmos Embed1 NIM container includes the following model:
| Model Name & Link | Use Case | How to Pull the Model |
|---|---|---|
| Cosmos-Embed1 | Implementation of Cosmos-Embed1-224p HF model that includes tritonserver, CV-Cuda, PyNvVideoCodec and TensorRT optimizations. | Automatic |
Deployment Details:
To deploy the container:
- Pull the NIM container
docker pull nvcr.io/nim/nvidia/cosmos-embed1:1.0.0
- Run the NIM container
docker run -it --rm --runtime=nvidia --name=cosmos-embed1 --gpus device=0 -p 8000:8000 -e NGC_API_KEY=$NGC_API_KEY nvcr.io/nim/nvidia/cosmos-embed1:1.0.0
- Explore the API
Once running, the API documentation is available at:
http://localhost:8000/docs
Our AI models are designed and optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA’s hardware (GPU cores) and software frameworks (CUDA, PyNVCodec, CV-CUDA, TensorRT), the model achieves significantly faster inference compared to CPU-only solutions.
Reference(s):
- Cosmos-Embed1: A Joint Video-Text Embedder for Physical AI Link
- Hugging Face: nvidia/Cosmos-Embed1-224p
- Radford, A. et al. “Learning transferable visual models from natural language supervision.” ICML 2021. Paper Link
- Li, J. et al. “BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models.” Paper Link
- CV-CUDA
- TensorRT
- PyNvVideoCodec
- Triton Inference Server
Container Version(s):
- v1.0.0: Initial release of the Cosmos-Embed1 NIM.
Ethical Considerations:
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their internal developer team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.
- Users are responsible for model inputs and outputs. Users are responsible for ensuring safe integration of this model, including implementing guardrails as well as other safety mechanisms, prior to deployment
Please report security vulnerabilities or NVIDIA AI Concerns here.
Enterprise Support:
This NIM is supported as part of the NVIDIA AI Enterprise software platform, which includes enterprise-grade support, security, and management features.