Wan

Alibaba's video generation family, and the reason AI video isn't only for people with API budgets. The open Wan 2.x models run locally on your own hardware: no per-generation fees, no sending your work to someone else's servers, and the community has built optimized versions and plugins for ComfyUI and other popular tools. Wan 3.0 (August 2026) is the big step up: clips from 2 to 30 seconds in up to 1080p with native audio, from text, images, video, audio or even a PDF or slide deck. For now Wan 3.0 is an application-based beta on Alibaba's Model Studio and Qwen Cloud, with a members' site at wan.video planned. Perfect for experimenting with AI video, building prototypes or producing content at scale without the API bill making you cry.

Something here out of date? Tell me