Wan AI video generator
Alibaba's open and commercial video model family, with strong community adoption and variants for text, image, speech, and animation-driven video.
Should You Use Wan?
Use it for: open-weight benchmarking, developer workflows, image-to-video, animation and character replacement, speech-to-video experiments
Skip or compare first if: Hardware requirements can be high. Newer catalogue versions beyond Wan2.2 require official confirmation. Consumer experience depends on wrapper quality.
Positioning: Alibaba's open-weight and commercial video model ecosystem for T2V, I2V, TI2V, speech-to-video, and animation.
Top Capabilities
- Open model ecosystemWan2.2 provides public code, weights, model cards, and integration into community tooling.Source
- MoE video diffusionWan2.2 introduces a Mixture-of-Experts architecture to increase model capacity at similar compute cost.Source
- T2V/I2V/TI2V coverageThe repository supports text-to-video, image-to-video, and text-image-to-video workflows.Source
- Animation and speech variantsWan2.2 added Animate and S2V variants for character animation/replacement and audio-driven cinematic video.Source
Company Snapshot
Built by Alibaba Group
Founded / HQ: 1999 / China
What it does: Alibaba Group is a global technology company focused on AI, cloud computing, and consumption, with major businesses spanning e-commerce, logistics, local services, entertainment, and enterprise cloud.
Main products/ecosystem: Wan, Tongyi Wanxiang, HappyHorse, EasyAnimate, I2VGen-XL, DAMO Text-to-Video
Founders/leadership: partial/evidence pending
Scale/funding/status: Status: Public Company
Relationship to Wan: Alibaba is the company behind Wan/Tongyi Wanxiang and related open/commercial video generation efforts, with GitHub, Hugging Face, ModelScope, and Alibaba Cloud-style distribution routes.
Price, Access, Open Status
Access: Open Weights, Closed Api, Aggregator
Open/closed: Open.
Open weights can be downloaded, but practical cost depends on GPU hardware or hosted inference. Commercial Alibaba/partner endpoints should be priced separately.
Price, regional access, and commercial terms need a live check before cost-sensitive recommendations.
Latest Version
Wan 2.5 / 2.6 / 2.7
2026 / partial
Names appear in third-party catalogues and should not be treated as confirmed official releases yet.
Latest News
Wan2.2-Animate-14B has been integrated into Diffusers
2025-11-13 / Wan-Video GitHub
External source link for this model.
Open originalWhy It Ranks Here
Rank: #4 87
Score source: Wan appears in Artificial Analysis and Arena snapshots, and the open repository/paper provide reproducible public evidence.. Evidence confidence is high.
Key Score Drivers
These are the scoring dimensions most responsible for the current ranking position.
- Model Quality Proxy: 92Wan appears in Artificial Analysis and Arena snapshots, and the open repository/paper provide reproducible public evidence.
- Capability Depth: 76Supported workflow tags: t2v, i2v, ti2v, speech_to_video, character_animation, open_weights. Depth rewards T2V/I2V/V2V, references, native audio, editing, API, and open weights where presen...
- Version Maturity: 995 version/history entries in the profile; latest public date is 2026.
- Access And Pricing: 88Access types: open_weights, closed_api, aggregator. Pricing note: Open weights can be self-hosted, while hosted APIs and apps charge separately. Hardware cost is a major practical pricing d...
Alternatives And Comparisons
Use these when price, access, output style, or workflow fit is uncertain.
- HunyuanVideoTencent's open-weight video generation family, important for developers, researchers, and self-hosted benchmarking.View modelCompare
- LTX VideoLightricks' open-weight and productized video model family, notable for synchronized audio-video generation and production-oriented open deployment.View modelCompare
- CogVideoXAn influential open video generation model family from the THUDM/Zhipu ecosystem, useful as a research and self-hosted baseline.View modelCompare
- Runway GenA mature creative video generation platform and model family for production-style image-to-video, text-to-video, editing, and workflow automation.View modelCompare
What It Is
Primary content is rendered into static HTML.
Wan is Alibaba's large-scale video generative model family. It is important because it bridges commercial distribution with open weights, GitHub code, Hugging Face/ModelScope hosting, and developer workflows.
Wan2.2 is a major upgrade that introduced a Mixture-of-Experts approach for video diffusion and released inference code and model weights for text-to-video, image-to-video, and TI2V tasks. Later Wan2.2 variants expanded into speech-to-video and character animation.
For consumers and developers, Wan is less of a single app and more of an ecosystem. It should be evaluated for open deployment, ComfyUI/Diffusers support, model size, hardware requirements, community fine-tunes, and whether hosted services expose the latest versions.
Version Progress
- Wan 2.5 / 2.6 / 2.7Names appear in third-party catalogues and should not be treated as confirmed official releases yet.
- Wan2.2Released inference code and model weights. Supports T2V, I2V, and TI2V. Integrated with Diffusers and ComfyUI.
- Wan technical reportPublished the Wan open large-scale video generative model paper line.
Open / Source Evidence
Status: Open
Wan2.2 publishes code and model weights with Apache-2.0 repository licensing shown by GitHub repository metadata.
Latest News
Model-specific news entries link directly to the original publisher; article bodies stay off-site.
- Wan2.2-Animate-14B has been integrated into Diffusers External source link for this model. Open original
- Wan: Open and Advanced Large-Scale Video Generative Models Technical paper source for the model family. Open original
Full Score Breakdown
Weighted evidence dimensions separate model quality from access, web authority, ecosystem, and source confidence. See methodology and rankings.
| Dimension | Score | Weight | Evidence | Confidence |
|---|---|---|---|---|
| Model Quality Proxy | 92 | 24% | Wan appears in Artificial Analysis and Arena snapshots, and the open repository/paper provide reproducible public evidence. | high |
| Capability Depth | 76 | 16% | Supported workflow tags: t2v, i2v, ti2v, speech_to_video, character_animation, open_weights. Depth rewards T2V/I2V/V2V, references, native audio, editing, API, and open weights where present. | high |
| Access And Pricing | 88 | 12% | Access types: open_weights, closed_api, aggregator. Pricing note: Open weights can be self-hosted, while hosted APIs and apps charge separately. Hardware cost is a major practical pricing dimension for open deployment. | high |
| Version Maturity | 99 | 12% | 5 version/history entries in the profile; latest public date is 2026. | high |
| Ecosystem Popularity | 100 | 10% | GitHub snapshot includes 1 repo(s), 16,923 stars, 2,118 forks, 285 open issues, latest pushed 2026-03-17T10:48:41Z. | high |
| Web Authority | 59 | 10% | SEO snapshot found product domain wan.video and company domain home.alibabagroup.com; sampled sitemap URL count is 0. | medium |
| Company Distribution | 88 | 10% | Alibaba Group distribution profile: Wan and related Alibaba video models matter because they combine serious cloud/API distribution with open-weight releases, making Alibaba one of the most important vendors for both cr... | medium |
| Source Confidence | 100 | 6% | Profile source confidence is confirmed with 2 official URL(s), 4 feature evidence item(s), SEO present, GitHub present. | high |
Limitations And Pricing
Open weights can be self-hosted, while hosted APIs and apps charge separately. Hardware cost is a major practical pricing dimension for open deployment.