No GPU required
Run the model in the browser or via API. No 24GB VRAM, no ComfyUI setup, no model weights to manage.
Open source·1080p·Strong prompt adherence
Alibaba's open-source model family, hosted so you don't need a GPU. Excellent prompt adherence and a permissive license make it the developer and ComfyUI favourite.
No install. No account for the first clip. Watermark-free on paid plans.
Run the model in the browser or via API. No 24GB VRAM, no ComfyUI setup, no model weights to manage.
Follows dense, specific prompts — useful when you need a particular composition rather than a pretty picture.
Animate stills with controllable camera motion and consistent subject identity.
Prototype a video feature without hosting weights.
Generate specific shots that stock libraries don't have.
Compare open-source output against closed models on the same prompt.
| Spec | Wan 2.5 | Wan 2.2 |
|---|---|---|
| Max resolution | 1080p | 720p |
| Prompt adherence | Higher | Good |
| Audio | No | No |
| License | Permissive | Permissive |
| Price | $0.09 / second | $0.06 / second |
Sources: provider documentation, retrieved 2026-10-21. Prices are per second of generated video at the stated resolution.
Yes, the weights are open source. This studio hosts it so you can run it without a GPU.
No. Generation happens on our infrastructure; you only need a browser.
2.5 adds 1080p output and better prompt adherence at roughly 1.5× the price per second.
Yes on paid plans. Check the upstream license for self-hosted use.
$0.09 per second at 1080p — a 5-second clip is $0.45.
Yes. The same endpoint routes to Wan 2.5 and the other five models; see the API docs.
$0.09 per second. 20 free credits on sign-up.
Start freeNo credit card for the first clip.