
Alibaba Models
Alibaba Wan, Qwen-Image and Z-Image via API: text-to-image, image editing, inpainting and ControlNet-guided image generation models.
Alibaba Group develops the Wan, Qwen-Image and Z-Image text-to-image and image editing models through Tongyi Lab, which builds Wan and Z-Image, and the Qwen team, which builds Qwen-Image. The company was founded in Hangzhou, China, in 1999 by 18 people led by Jack Ma, and is still headquartered there.
Available models
Z-Image
alibaba/z-image-base/text-to-imageText to Image$0.01 / megapixel
alibaba/z-image-turbo/text-to-imageText to Image$0.005 / megapixel
alibaba/z-image-turbo/image-to-imageImage to Image$0.005 / megapixel
alibaba/z-image-turbo-controlnet/image-to-imageImage to Image$0.0065 / megapixel
alibaba/z-image-turbo-inpaint/image-to-imageImage to Image$0.01 / megapixel
alibaba/z-image-turbo-tiling/text-to-imageText to Image$0.02 / megapixel
Wan 2.7
Wan 2.2
About
Alibaba builds language, image and video models and publishes many of them with open weights. Its image models cover text-to-image generation, instruction-based image editing, inpainting and rendering text inside images.
- Wan: Text-to-image and image editing from Tongyi Lab's Wan line, which also covers video; Wan 2.2 has Apache 2.0 weights and its A14B model uses mixture-of-experts.
- Qwen-Image: Text-to-image on a 20B-parameter MMDiT model from the Qwen team; renders complex text in images, including multi-line English and Chinese.
- Z-Image: Text-to-image on 6B-parameter open-weight models; Base for fine-tuning, Turbo for 8-step generation, and Turbo variants for ControlNet, inpainting and tiling.
Build with the API
One MuseSpan API key calls every Alibaba model on this page: copy a model ID and send it as the model field of your request.








