Wan in a visual AI workflow
Run Alibaba's Wan in a visual AI workflow — open-weights text- and image-to-video with a deep toolkit of control modes, BYOK at provider prices.
Fresh Wan examples are on the way. Meanwhile, here's the NodeTool canvas where you'll run it.
Wan is Alibaba's open-weights video family. Beyond plain text- and image-to-video it ships an unusually deep set of control modes — pose, depth, inpainting, outpainting, reframing — which makes it the video model to reach for when you need to steer motion, not just prompt it.
NodeTool exposes those modes as distinct nodes, so a Wan graph can be as simple as prompt-to-clip or as involved as a pose-guided, inpainted, reframed pipeline. Being open weights, Wan also runs on infrastructure you control when a provider isn't the right fit.
Wan is BYOK through the hosted providers below, or self-hosted. The provider table is generated from the manifests.
Where to run Wan
NodeTool is bring-your-own-key: you call Wan through a provider you already have access to and pay their list price — no credits, no markup. These providers serve Wan today.
Use Wan in a ready-made workflow
Drop Wan into a template and you have a repeatable pipeline — prompt in, asset out — that you can edit, re-run, and share as a single file. Open it in NodeTool and swap in your own key.
Open the templateFrequently asked
- Is Wan open source?
- Wan ships open weights, so you can run it on your own hardware in addition to calling it through a hosted provider. NodeTool supports both.
- What control modes does Wan support?
- Pose, depth, inpainting, outpainting, and reframing, among others — each available as its own node in NodeTool for precise, steerable motion.
- Can I run Wan locally?
- Yes. Because the weights are open, Wan can run self-hosted; in NodeTool you point the node at your endpoint instead of a hosted provider.
Run Wan your way.
Download NodeTool Studio and build across image, video, audio, and text with your own keys.