Wan 3
Text & image to video · 2026-08
Alibaba's newest video model (August 2026). Up to 30 seconds at up to 1080p, with optional audio and references from images, clips and sound. Wan 3.0 Prime is its speed-optimised sibling.
AI video model · By Alibaba
Alibaba's Wan 3, Wan 2.7, 2.6, 2.5 and 2.2 in your Qten account: text, image and reference to video, plus Wan image models.
Free plan · No credit card · Updated 2026-11-18
Wan
By Alibaba
10 versions of Wan in one Qten account, from fast drafts to the newest release.
Wan is Alibaba's AI video model family. It turns text, images or reference clips into video, adds sound from Wan 2.5 on, and Wan 3 reaches 30 seconds at up to 1080p. Qten offers Wan from 2.2 to 3.0 Prime, in text-to-video, image-to-video and reference-to-video modes, plus Wan 2.7 Image.
Wan comes from Alibaba's Tongyi Lab. Wan 2.1 and 2.2 were released as open-weight models in 2025 and became popular with creators. Wan 2.5 added audio, Wan 2.6 brought longer, multi-shot clips with lip-sync, Wan 2.7 added reference-to-video, and Wan 3, launched in public beta in August 2026, makes clips of up to 30 seconds from text, images, video and audio references.
| Made by | Alibaba |
|---|---|
| Newest on Qten | Wan 3 |
| Versions on Qten | 10 |
| Max resolution | Up to 1080p |
| Max length | Up to 30 s (Wan 3) |
| Native audio | Yes (Wan 2.5+) |
| Inputs | Text prompt · Start image · Reference images · Audio |
Limits are the maker's published figures, checked 2026-11-18. What you can use may depend on your plan.
10 versions of Wan in one Qten account, from fast drafts to the newest release.
Text & image to video · 2026-08
Alibaba's newest video model (August 2026). Up to 30 seconds at up to 1080p, with optional audio and references from images, clips and sound. Wan 3.0 Prime is its speed-optimised sibling.
Image to video · Text to video · Reference to video
Text to video and image to video up to 15 seconds, plus reference-to-video to keep a character or product consistent.
Image to video · Text to video · 2025-12
Released in December 2025: longer, multi-shot clips with synced audio and lip movement. Wan 2.6 Flash is the fast tier.
Full Wan 2.6 pageImage to video · Text to video · 2025-07
A quick, low-cost Wan for drafts, in text-to-video and image-to-video modes.
| Version | What it makes | Tier | Released |
|---|---|---|---|
| Wan 2.2 Fast | Image to video · Text to video | Fast / lower cost | 2025-07 |
| Wan 2.5 | Image to video · Text to video | Standard | 2025-09 |
| Wan 2.5 Fast | Image to video · Text to video | Fast / lower cost | — |
| Wan 2.6 | Image to video · Text to video | Standard | 2025-12 |
| Wan 2.6 Flash | Text & image to video | Fast / lower cost | — |
| Wan 2.7 | Image to video · Text to video · Reference to video | Standard | — |
| Wan 3 | Text & image to video | Standard | 2026-08 |
| Wan 3.0 Prime | Text & image to video | Pro tier | — |
| Wan 2.7 Image | Text to image & editing | Standard | — |
| Wan 2.7 Image Pro | Text to image & editing | Pro tier | — |
Text to video, image to video or reference to video, depending on the version.
Wan 3 makes long clips in one generation, at up to 1080p.
From Wan 2.5, clips can include sound; Wan 2.6 and later sync speech to lips.
Wan 2.7 Image and Image Pro create stills you can animate with Wan video.
Animate this portrait: the woman turns toward the camera, hair moving in a light breeze, soft smile, shallow depth of field, 9:16.A lighthouse on a cliff during a storm, huge waves crashing, the beam sweeping through heavy rain, thunder rumbles, slow push-in, 16:9.Isometric 3D animation of a tiny city where packages travel on glowing conveyor lines between buildings, clean pastel colors, upbeat, 16:9.Animate product photos, art and thumbnails.
Lip-synced clips with Wan 2.6 and later.
Up to 30 seconds per clip with Wan 3.
Wan 2.2 Fast and 2.6 Flash for quick tests.
The other video models in the same Qten account. Pick per scene, or let Qten choose.
| Wan | Kling | Seedance | Veo | MiniMax Hailuo | |
|---|---|---|---|---|---|
| Made by | Alibaba | Kuaishou | ByteDance | Google DeepMind | MiniMax |
| Newest on Qten | Wan 3 | Kling 4 | Seedance 2.5 | Veo 3.1 | MiniMax H3 |
| Versions on Qten | 10 | 6 | 7 | 2 | 7 |
| Max resolution | Up to 1080p | Up to 1080p | Up to 1080p | Up to 1080p | Up to 1080p |
| Max length | Up to 30 s (Wan 3) | Up to 30 s (Kling 4) | Up to 30 s (Seedance 2.5) | Up to 8 s | Up to 15 s (MiniMax H3) |
| Native audio | Yes (Wan 2.5+) | Yes | Yes (Seedance 2.0 / 2.5) | Yes | Yes (MiniMax H3) |
| Inputs | Text prompt · Start image · Reference images · Audio | Text prompt · Start image · First & last frame · Reference images | Text prompt · Start image · Reference images · Audio | Text prompt · Start image · First & last frame · Reference images | Text prompt · Start image · First & last frame · Reference images |
Limits are the maker's published figures, checked 2026-11-18. What you can use may depend on your plan.
Sign up on the Free plan. No credit card, and you get credits to start.
Write your prompt or script, then choose Wan as the model, or let Qten pick the best model for each step.
Review the result, edit it in Qten, and post or schedule it to YouTube, TikTok, Instagram and more.
FAQ
Wan is a family of AI video models from Alibaba's Tongyi Lab. It creates video from text, images and reference clips; earlier versions such as Wan 2.2 were released as open-weight models.
Qten offers Wan 2.2 Fast, Wan 2.5, Wan 2.5 Fast, Wan 2.6, Wan 2.6 Flash, Wan 2.7, Wan 3 and Wan 3.0 Prime for video, in text-to-video, image-to-video and reference-to-video modes, plus Wan 2.7 Image and Wan 2.7 Image Pro.
Qten has a Free plan with 100 free credits and no credit card. Generating with Wan uses credits, depending on the version, resolution and length. See the Pricing page for details.
Wan 3 makes up to 30 seconds per generation. Wan 2.7 makes up to 15 seconds for text and image to video.
Wan is a flexible, budget-friendly choice with separate image-to-video and reference modes. Kling is strong on multi-shot scenes. Both are in Qten.
No. Qten runs Wan for you in the cloud. You use it from your Qten account with Qten credits, with no setup, GPU or API key.
Model names and logos are trademarks of their owners. Qten is not affiliated with or endorsed by them.
Wan and 130+ other AI models in one account. Start free.
Start creating free