Wan 通义万相
Alibaba's video models: open Wan 2.x weights and the closed Wan 3.0
Overview
Wan (通义万相, Tongyi Wanxiang) is Alibaba's image and video generation family: Apache-2.0 open weights for Wan 2.1/2.2 that run locally in ComfyUI, and the closed Wan 3.0 flagship (Aug 2026) with clips up to 30 seconds at 1080p. Available on wan.video, the Wan app and Alibaba Cloud Model Studio.
---
通义万相是阿里巴巴旗下的图像与视频生成模型系列：Wan 2.1/2.2 以 Apache-2.0 协议开源，可在 ComfyUI 本地运行；2026 年 8 月正式上线的 Wan 3.0 旗舰版为闭源，单段视频最长 30 秒、最高 1080P。可通过 wan.video、万相 App 与阿里云百炼使用。
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 1 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- 1 video and 1 image job at a time
- 6 image styles
- 300 credits/month, up to 60 videos
- Watermark-free 1080p downloads
- Videos of 10-30 seconds
- 3 video and 3 image jobs at a time
- 1,200 credits/month, up to 240 videos
- 8 video and 5 image jobs at a time
- Same output features as Pro
- $0.05 / $0.10 / $0.20 per second at 480P / 720P / 1080P
- wan3.0-video-prime up to $0.28 per second
Use cases
What you can produce with Wan 通义万相
- Open-weight Wan 2.1 and 2.2 video models (Apache-2.0)
- Wan 3.0 video generation up to 30 seconds at 1080p
- Text, image, video and audio inputs
- Local generation in ComfyUI
- wan.video website and Wan mobile app
- Wan API on Alibaba Cloud Model Studio
ASEAN Perspective
Wan 通义万相 in Southeast Asia
阿里同时开源 Wan 2.1 权重，与混元视频共同构成国产开源视频模型双极。
Wan remains the most useful open video model family for anyone willing to run it locally, and its Chinese-prompt handling is a genuine strength. The picture has split, though: the best model, Wan 3.0, is available only through Alibaba's website and API, so "open-source" now describes the older 2.x line rather than the flagship. Hosted plans are cheap, and Model Studio's Singapore region publishes clear per-second prices. Strong for ComfyUI users and ASEAN developers; casual creators wanting a polished editor may prefer packaged rivals.
What we found
Wan, known in Chinese as Tongyi Wanxiang, is Alibaba's family of image and video generation models, built by its Tongyi Lab. It began as an image model in July 2023, and its video line became the leading open-weight option after Alibaba released Wan 2.1 in February 2025 and Wan 2.2 that July under the Apache 2.0 licence. Wan 2.2 uses a mixture-of-experts design in a 14-billion-parameter version, with a 5-billion-parameter model that fits consumer GPUs, and runs in ComfyUI with official workflows. Alibaba has since added open motion and character models, Wan-Dancer and Wan2.2-Animate-2, in July and August 2026.
The newest models are not open. Wan 3.0, generally available since 24 August 2026, makes clips of up to 30 seconds, twice Wan 2.7's limit, at up to 1080p, and accepts text, images, video, audio and reference documents such as PDFs and slides, with up to 20 reference assets. Its GitHub repository holds only a description and licence, and its weights are not on Hugging Face. Prompts in Chinese are a long-standing strength.
There are three ways to use it. The open 2.x weights are free to run on your own hardware. The wan.video website and the Wan mobile app have a free plan with daily check-in credits and one image and one video job at a time; Pro costs $5 a month billed yearly or $6.50 monthly for 300 credits, enough for up to 60 videos, with 1080p, watermark-free downloads and clips of 10 to 30 seconds; and Premium costs $20 a month billed yearly or $26 monthly for 1,200 credits. Those plans cover the website only. Developers pay separately through Alibaba Cloud Model Studio, whose Singapore region lists Wan 3.0 at $0.05 a second at 480p up to $0.20 at 1080p, with a higher-quality Prime version up to $0.28.
Wan suits ComfyUI users wanting free local video generation and developers needing an API in Southeast Asia.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Wan 通义万相's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Wan 通义万相 unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Wan 通义万相 directly →
Spotted something out of date? Suggest an update →
Alternatives to Wan 通义万相
More in Video & Audio
