Wan 2.2
Wan 2.2 brings Mixture-of-Experts to open video, at about 27B total and 14B active per step, plus a 5B model running 720p on consumer GPUs.
A two-expert design for the denoising process (high-noise and low-noise experts), more training data (+65.6% images, +83.2% videos vs 2.1), Apache 2.0. Later additions were S2V (2025-08) and Animate (2025-09-19). Later flagship versions (2.5 to 3.0) do not appear on Hugging Face and look API-only.
- Date
- Monday 28 July 2025
- Lab
- Alibaba (Tongyi)
- Kind
- model
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| Training data growth vs Wan2.1 | +65.6% images, +83.2% videos Per the Wan2.2 README (company) | company |
| Wan2.2-VAE compression | 16x16x4 5B model does text/image-to-video at 720P, 24fps and runs on consumer GPUs such as the 4090 | company |
Weights appear on Hugging Face from 2025-07-18 to 2025-07-28; GitHub lists 2025-07-28. No Wan 2.5-3.0 weights are on Hugging Face (only Wan2.2-Animate-2 and Wan-Dancer in 2026), which suggests later flagships are closed.
Sources
- github.com/Wan-Video/Wan2.2
- huggingface.co/Wan-AI/Wan2.2-T2V-A14B
- huggingface.co/Wan-AI/Wan2.2-TI2V-5B
This record was checked against its sources on 6 October 2026. How we check