Wan 2.1

Wan 2.1 is an AI model from Alibaba for generating videos, images, and music from text prompts. Released in February 2025 as open source under the Apache 2.0 license, it’s available for free download on GitHub and HuggingFace and can run both online and locally on your computer.

What Wan 2.1 can do

• Generate video from a text description
• Edit existing video clips
• Generate audio for created videos

Users note strong video quality, including realistic physical effects (for example, water motion simulation). A 5-second 480p video reportedly takes about 4 minutes to generate on an NVIDIA RTX 4090.

Availability, languages, and setup

Wan 2.1 supports Chinese and English UI, and while a Russian interface isn’t available, it can understand Russian-language prompts. The platform provides 50 free credits per day.

For local use, you can download Wan 2.1 from GitHub or HuggingFace. There is also a lighter T2V-1.3B version that requires 8.19 GB of VRAM and can generate up to 5 seconds of video at 480p.

Key details

• Open-source license: Apache 2.0
• Includes a video-VAE architecture component
• Code can be used, studied, modified, and redistributed