Creators making talking or singing character videos who can run an open model themselves; free under Apache 2.0.
Kenning's take
What it does
Wan2.2-S2V is a 14-billion-parameter speech-to-video model from Alibaba's Tongyi Lab. Given a single image and an audio track, it generates video of the character speaking, singing or performing, with facial expressions, body movement and camera work, in full-body or half-body shots at 480p or 720p. The weights and code are free to download under Apache 2.0, and there are online demos on Hugging Face and ModelScope.
- Image plus audio to video
- Speaking, singing and performance scenes
- Full-body and half-body characters
- 480p and 720p output
- Consistent subjects in longer videos
- Open weights and inference code
Who it suits. Developers and creators making talking or singing character videos who can run or try an open model.
Key facts
- Category
- AI models
- Made by
- Alibaba (Tongyi Lab)
- Based in
- China
- Released
- 26 August 2025
- How to get it
- open weights on Hugging Face and ModelScope, demos on Hugging Face Spaces and ModelScope Studio
- Works on
- Hugging Face, ModelScope, self-hosted
- In the UK
- Yes
- Free plan
- free to download and run under Apache 2.0
FAQ
When was Wan2.2-S2V released?
Alibaba released Wan2.2-S2V-14B on 26 August 2025.
Is Wan2.2-S2V free?
Yes. The model weights and code are free under the Apache 2.0 licence.
What does Wan2.2-S2V need as input?
A single image and an audio clip, plus an optional text prompt.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools

ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.

Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.

Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.