BindWeave

by ByteDance

About the model:

Video generation model using cross-modal integration via entity grounding and representation alignment.

Model type:

Video

Model performance class:

Full

Link to the model:

https://huggingface.co/ByteDance/BindWeave

Base models:

BindWeave

End model:

BindWeave

End model license:

Apache-2.0

About the organisation:

ByteDance, the parent company of TikTok.

Link to the organisation:

https://www.bytedance.com/

Model release date:

March 2024

Availability

Base Model Data

Preprint mentions curation of OpenS2V-5M dataset.

https://arxiv.org/pdf/2510.00438

End User Model Data

Preprint mentions curation of OpenS2V-5M dataset.

https://arxiv.org/pdf/2510.00438

Base Model Weights

Weights made available on HuggingFace.

https://huggingface.co/ByteDance/BindWeave

End User Model Weights

Weights made available on HuggingFace.

https://huggingface.co/ByteDance/BindWeave

Training Code

No training code found.

Documentation

Code Documentation

No training code, so no documentation.

Hardware Architecture

No information about hardware architecture found.

Preprint

Preprint available on arXiv.

https://arxiv.org/pdf/2510.00438

Paper

No peer-reviewed paper found.

Modelcard

Model card primarily contains information on model performance.

https://huggingface.co/ByteDance/BindWeave

Datasheet

Dataset has associated preprint.

https://arxiv.org/pdf/2505.20292

Access

Licenses

Apache-2.0, an OSI-approved license.

https://huggingface.co/ByteDance/BindWeave

Is this information not up to date?

Contribute here ->

Supported by the Centre for Language Studies and the Dutch Research Council. Website design & development © 2024 by BSTN. This version of the index generated 09 April 2026, website content last updated 11 March 2026.