wan2.7

Wan2.7 Image is a unified image generation and editing model from Alibaba, released in April 2026. Rather than splitting creation and correction across separate tools, it puts both in one workflow. Portrait control goes down to bone structure, eye shape, eyebrow arc, and nose bridge, which is the difference between a face and a generic face. Colour can be specified with exact hexadecimal codes and proportions, so brand palettes survive into the output. Long text renders at print quality in 12 languages with prompts up to 3000 tokens, and a single session can produce up to 12 coherent images that hold style, lighting, and palette together. Editing is interactive and pixel level. The site states the model is currently free to use.

- Generate and edit images in the same session

- Control facial features at a detailed level

- Set exact colours with hexadecimal codes

- Render long text in 12 languages at print quality

- Designers working to a strict brand palette

- Marketers who need long copy inside an image

- Content creators producing coherent multi image sets

- Anyone who wants editing without switching tools

Free use is described as the current position rather than a permanent policy, so plan for change. Detailed control requires detailed prompts; asking for a specific nose bridge in one vague sentence will not deliver it. Long text rendering is a strength, but check spelling and line breaks in the final image, since a single misplaced character ruins a poster. Multi image coherence helps a series hold together, yet lighting and proportion can still drift between frames, so review the set as a whole before it ships.