In Alibaba’s AI landscape, Tongyi serves as the unified brand umbrella: Tongyi Qianwen handles text-based conversations, while Tongyi Wanxiang manages image generation. The name is quite evocative—“Wanxiang” (Myriad Manifestations) is derived from the idiom “wan xiang geng xin” (all things renewed), and paired with “Tongyi” (Universal Meaning/Understanding), it conveys the idea of an AI that comprehends everything and generates myriad forms.
Among domestic AI painting tools, Tongyi Wanxiang follows a strategy of “big-tech backing + Chinese-first focus + multi-style coverage,” competing with other domestic AI art tools like Wenxin Yige and Miaohua to fill the market gap for Chinese users who don’t want to tinker with Midjourney or Stable Diffusion.
What is Tongyi Wanxiang?
Tongyi Wanxiang (tongyi.aliyun.com/wanxiang) is Alibaba’s AI image generation and processing platform, part of the broader Tongyi family, developed by the image generation team at Alibaba DAMO Academy. It supports various functions including text-to-image, image-to-image, style transfer, and image editing. Specifically optimized for Chinese input, it is one of the mainstream choices for domestic general users seeking AI art tools.
The platform allows direct login via Alibaba accounts (Alipay or Taobao accounts) and integrates with Alibaba Cloud’s API services, making it accessible to both individual users and developers.
Core Features
Text-to-Image: Chinese-Friendly as the Key Advantage
Input a Chinese description, select a style, and the AI generates the corresponding image. Tongyi Wanxiang’s biggest highlight is its Chinese comprehension capability: you don’t need to learn how to write English prompts; using normal Chinese descriptive sentences yields good results.
For example, describing “a lotus pond at sunset, an ancient pavilion in the distance, water reflections, a tranquil atmosphere, in the style of Chinese ink wash painting” will be accurately understood by the AI and rendered into an image matching the description. Compared to Midjourney, which requires precise English descriptions, or Stable Diffusion, which demands knowledge of tag syntax, Tongyi Wanxiang has a significantly lower learning curve.
Supported Style Directions:
Traditional Chinese Styles: Ink wash painting, Gongbi (meticulous brushwork), blue-green landscape, and freehand flower-and-bird painting—these are directions where Tongyi Wanxiang has invested specific optimization efforts. The quality for generating Guofeng (Chinese national style) content is assured, representing its most distinct differentiating advantage over overseas tools.
Realistic Photography Styles: High-realism portraits, landscapes, architecture, products, etc., suitable for creative scenarios requiring photo-like imagery.
Illustration and Commercial Design: Flat illustrations, commercial poster aesthetics, book covers, brand materials, and other visual styles suitable for commercial use.
Anime and ACG (Animation, Comic, Games): Japanese anime styles and Chinese comic styles, with specialized models for generating character designs.
Artist Styles: Imitations of Western art styles such as oil painting, watercolor, and Impressionism.
Image Style Transfer
Upload a photo, select a target style, and the AI transforms the photo into an artwork in that corresponding style. This feature requires no prompt writing and is the simplest to operate:
- Photo to Oil Painting: Transform travel photos into thick-paint oil painting textures.
- Photo to Ink Wash: Convert landscape photos into traditional Chinese painting styles.
- Portrait to Anime: Turn real-person photos into anime-style images.
- Modern to Ancient Style: Add classical elements and color tones.
Style transfer is suitable for quickly experiencing different artistic effects or creating uniformly styled images for specific thematic content.
Image Editing Features
Smart Erase: If there are passersby or unwanted clutter in a picture that you want to remove, circle the area, and the AI will automatically erase it and fill in the background seamlessly. This is very practical for cleaning up travel photos or product shots.
Local Redraw: Box the area of the image you want to modify, describe the desired changes, and the AI will only modify the selected area while leaving the rest unchanged. This can be used to change a person’s clothing, replace parts of the background, or add specific elements.
Image Super-Resolution: Upscale low-resolution images by having the AI fill in details to enhance clarity, suitable for enlarging old photos or low-quality assets.
ACG Character Generation
There is a dedicated function for designing original anime characters. You can describe the character’s physical features, clothing, and personality to generate corresponding original anime character images. This has practical value for self-media creators, storytellers, and game developers who need character design capabilities.
Alibaba Cloud API Integration
Through the Alibaba Cloud DashScope platform, Tongyi Wanxiang’s image generation capabilities can be invoked via API. Developers can integrate these into their own products and applications, paying based on the number of calls. Pricing is supported by Alibaba Cloud’s pricing structure, including bulk discount plans.
Comparison with Competitors
Vs. Wenxin Yige (Baidu): As another major Chinese AI painting platform from a big tech company, it is the most direct competitor. Their functional architectures are similar; the main differences lie in: Wenxin Yige has more prominent Guofeng characteristics, with more specialized optimizations and operations for traditional styles like ink wash and Gongbi; Tongyi Wanxiang offers more diverse overall style coverage and deeper integration with the Alibaba Cloud ecosystem. Both are worth trying, and you can choose based on specific needs.
Vs. Midjourney: MJ is the international benchmark for AI painting quality, undeniably leading in visual impact and artistic aesthetics. However, MJ’s barriers to entry include: requiring Discord, English prompts, a paid subscription, and the need for scientific internet access (VPN) within China—these are real obstacles for many users. Tongyi Wanxiang has an extremely low barrier to entry: direct Chinese input, domestic connectivity, and free quotas, making it suitable for users who don’t want to deal with complications.
Vs. Stable Diffusion (including various wrapped platforms): The biggest advantage of the SD ecosystem is flexibility—massive community models, LoRAs, and ControlNet allow almost any visual style to be realized. However, SD has a steep learning curve, requiring knowledge of basic parameters and model selection. Tongyi Wanxiang is more like a point-and-shoot camera: ready to use out of the box, at the cost of flexibility and controllability compared to the SD ecosystem.
Vs. DALL-E 3 (ChatGPT): DALL-E 3 excels at understanding complex, abstract descriptions, particularly in integrating text within images (though Chinese text rendering still has limitations); it requires a ChatGPT Plus subscription and faces access barriers in China. Tongyi Wanxiang offers direct domestic connectivity and better Chinese description comprehension, making it suitable for daily use by domestic users.
Vs. Adobe Firefly: Firefly is Adobe’s AI image generation tool, ensuring content safety (clearly licensed training data) and integrating into the Adobe Creative Suite; it is suitable for professional users already working within the Adobe ecosystem, while those outside it have little reason to choose it.
Unique Advantages in Guofeng Content
This deserves separate emphasis: In generating Guofeng content, Tongyi Wanxiang is one of the few AI painting platforms that truly takes this direction seriously. Overseas tools (Midjourney, DALL-E) show inconsistent results in traditional styles like ink wash, Gongbi, and blue-green landscapes because such content makes up a relatively small portion of their training data.
The Alibaba team behind Tongyi Wanxiang has performed specialized model optimizations for these traditional styles. The generated ink-wash landscapes and Gongbi flower-and-bird paintings are far more accurate in style than those from overseas tools. For creators needing to produce Chinese traditional-style content—such as for WeChat Official Accounts, Xiaohongshu, or brand materials—this is a tangible advantage.
Who Should Use Tongyi Wanxiang?
General users who don’t want to learn English prompts: Say what you mean in Chinese and get what you want; this is the lowest-barrier AI painting experience, perfectly suitable for users who have never touched AI art before.
Creators needing Guofeng content: For traditional Chinese painting styles, cultural elements, and Chinese aesthetics, Tongyi Wanxiang is one of the best domestically optimized platforms currently available.
Alibaba Ecosystem Users: Those with existing Alibaba/Taobao accounts can use it without registration; additionally, if development needs arise, the integration path for Alibaba Cloud APIs is smooth.
Self-Media Content Creators: Forillustration (illustrations) needs on WeChat Official Accounts, Xiaohongshu, and Zhihu, Tongyi Wanxiang’s free quotas and multi-style coverage can meet daily creative illustration needs.
Education and Presentation Purposes: For illustrations needed in school courseware, lecture PPTs, or display materials, the no-registration quick generation satisfies ad-hoc needs.
Limitations
Compared to top-tier tools like Midjourney and DALL-E 3, there is a gap in image quality, particularly in detail refinement, lighting texture, and handling complex scenes.
Its self-developed models are incompatible with the Stable Diffusion ecosystem—unable to call community LoRA models or ControlNet controls for advanced features, limiting customizability and flexibility.
The stability of generated results is not high enough in certain styles; repeated generations from the same description can vary significantly, making it difficult to precisely control specific details.
Pricing
Tongyi Wanxiang offers a daily free generation quota, so basic usage is free. Once you exceed the free quota, you’ll need to purchase credits, which are billed per image; high-resolution images and certain high-quality models may incur additional fees. For exact details, refer to the official website, as Alibaba regularly adjusts its free quotas and pricing policies.
For everyday needs of general users, the free quota is usually sufficient. However, for professional users or developers with high-volume generation requirements, paid plans or API access are more suitable.
Tongyi Wanxiang is Alibaba’s serious entry into AI image generation. With a Chinese-friendly user experience and specialized optimization for traditional Chinese artistic styles, it occupies a clear niche in China’s AI art market. For users who want to generate images directly from Chinese descriptions, it is one of the most accessible and seamless options available.