Wan 2.2
Description
Wan 2.2 is a family of open video generation models from Alibaba (Wan-AI). It has three main models: Wan2.2-T2V-A14B creates video from text, Wan2.2-I2V-A14B animates an image, and the compact Wan2.2-TI2V-5B does both at 720p and 24 frames per second. The author also released separate weights for audio-driven video and character animation. The base text-to-video and image-to-video models do not generate sound, and clip length is not stated in the author's model card. The weights are published on Hugging Face under the Apache 2.0 license, which allows commercial use; the author claims no rights over generated content. A demo without installation is available from the author on Hugging Face (a Space for the 5B model). The online Wan service, where you can try the models without running them yourself, is in our catalog: Wan AI — that is a separate service card, while this page describes the open Wan 2.2 weights.
Pros and cons
Pros
- Apache 2.0 license, commercial use allowed
- Text-to-video, image-to-video and a hybrid 5B model
- The compact 5B runs on a consumer GPU
- Extra weights for audio-driven video and character animation
- Official demo on Hugging Face
Cons
- The base models do not generate sound
- The 14B models need at least 24 GB VRAM with offloading, 80 GB without
- Released in 2025; the online Wan service may run a newer model than the open weights
Key facts
- License
- Apache 2.0
- Commercial use
- Allowed
- Hardware requirements
- The 5B model runs on a consumer card such as an RTX 4090; 14B with offloading needs at least 24 GB VRAM (per the author)
- Developer
- Alibaba (Wan-AI)
- Interface languages
- English
- Checked on
- Oct 17, 2026