Features
- Text-to-video and image-to-video generation
- Open-source models, training code and inference code
- Data processing tools including scene segmentation, filtering and text-video pairing
- Pre-trained checkpoints on Hugging Face and ModelScope
- Open-Sora 2.0 with an 11B model
- Videos in various sizes and aspect ratios, up to 720p depending on version
- Runs locally and can be modified for custom solutions
- Apache-2.0 license
Pricing Details
Open-Sora is available for free and as open source, with no need for a monthly subscription or paid plans; the project makes its code and models openly available on GitHub. Users can access the project and use it for research, development, experimentation, and building applications based on AI-generated video. Access to the tool costs $0, and its use involves downloading the code and models available within the project, with the option to run them locally depending on the device’s capabilities. Running large models may require powerful computing resources, such as graphics processing units (GPUs), to ensure adequate performance. The project is licensed under the Apache-2.0 license, which allows for its use and development under the terms of the open-source license.