Cosmos Predict2 2B Video2World is an open-weight video generation model hosted on Hugging Face. It belongs to the class of video world models that produce predictive simulations from input video sequences.
The model is provided by NVIDIA under a dedicated license agreement. This agreement permits commercial use of the model, allows users to create and distribute derivative models, and states that NVIDIA does not claim ownership of any outputs generated by the model or its derivatives. Acceptance of the agreement occurs through any use, reproduction, modification, distribution, performance, or display of the model or its derivatives.
The license is designed to support permissive use and to enable further development of AI technologies. It is a legal contract between the user or their represented entity and NVIDIA Corporation together with its affiliates. The model page on Hugging Face serves as the distribution point for the weights and associated documentation.
No additional technical specifications, training details, inference requirements, or example applications appear in the provided evidence.
In the Video generation space, Cosmos Predict2 2B Video2World takes a focused approach. It focuses on generating predictive world simulations and video continuations from input video clips. It is built as an open-source project for AI researchers and developers. Cosmos Predict2 2B Video2World is open source under the Apache-2.0 license. It runs on the web, the command line, and API.
It is developed by NVIDIA (United States), and the product first shipped in 2025. The GitHub repository has been archived. Key capabilities include Video Generation, World Model, and 2B Parameters.
Latest indexed changes and source events
nvidia/Cosmos-Predict2-2B-Video2World verified by the PulseGate indexer
⚠ Archived
Other apps tracked under the same category.
