Wan 2.7 AI

Wan 2.7 AI is a browser-based creative studio wrapped around the Wan 2.7 multimodal model. Its selling point is control: rather than hoping the model interprets a prompt well, you can steer composition at the structural level and iterate on the result without leaving the workflow. The headline capabilities are 1080p output, clips up to 15 seconds, and the 3x3 grid image-to-video mode that lets you feed a storyboard layout straight into the model.

Frame-level direction is where it stands out for anyone planning shots. You can set a first frame, a last frame, or both, which makes it possible to define how a clip opens and closes instead of regenerating until something fits. Combined with subject and voice references that lock facial features, clothing and vocal characteristics, this supports multi-shot sequences where a character has to stay recognisable — a problem that has historically derailed AI video projects.

The studio also covers editing, not just generation. Instruction-based recreation lets you modify footage with natural-language directions while holding the motion structure, and the image side adds 4K output, up to eight hex colour codes for palette-accurate brand work, text rendering up to 4,000 characters across 13 languages, and box-select local editing that changes only the region you choose. Note that this is an independent platform built around the Wan model — its own disclaimer states it is not affiliated with or endorsed by Alibaba or the official Wan AI team. New users start with a batch of free credits before moving to paid plans.