Alibaba's Wan 3.0 video model is now available on ImagineToVideo. Since October 21, 2026 you can pick it in our generator for text to video and image to video, with an optional end frame, at up to 1080p and up to 30 seconds with sound. It is also the cheapest model on the site: a 5-second 720p clip costs 20 credits.
This post covers what Wan 3.0 is, what you can set on ImagineToVideo, what it costs, how it compares with the other models here, and what is not wired in yet. Upstream facts come from Alibaba's QwenCloud documentation and fal's model page; on-site limits are the ones our generator enforces.
TL;DR
- What: Alibaba's Wan 3.0, a single model for text-to-video and image-to-video with native audio.
- Where: live in the ImagineToVideo generator since October 21, 2026, via fal's public API.
- Specs here: 480p, 720p or 1080p; 3 to 15, 20 or 30 seconds; 16:9, 9:16, 1:1, 4:3 or 3:4; start frame plus optional end frame.
- Price: from 10 credits for a 5-second 480p clip. The $0.99 Starter Pack covers a 5-second 720p clip.
- Not here yet: reference-to-video, 21:9 and 2-second clips.
Release Timeline
| Date | What happened | Source |
|---|---|---|
| August 6, 2026 | wan3.0-video appears in QwenCloud's model release log as an invite-only test | QwenCloud release log |
| August 24, 2026 | QwenCloud marks Wan 3.0 generally available and adds wan3.0-video-prime, a faster variant | QwenCloud release log |
| October 21, 2026 | Wan 3.0 goes live on ImagineToVideo through fal's public API | ImagineToVideo |
Alibaba has not published Wan 3.0's weights. As of October 21, 2026 there is no Wan 3.0 repository on Hugging Face or GitHub, unlike earlier Wan releases such as Wan2.2. Our Wan 3.0 page tracks that and the other open claims with a source for every line.
What You Can Set on ImagineToVideo
| Setting | Options |
|---|---|
| Modes | Text to video; image to video from a start frame, with an optional end frame |
| Resolution | 480p, 720p (default) or 1080p |
| Length | 3 to 15 seconds, 20 seconds or 30 seconds (default 5) |
| Aspect ratio | 16:9, 9:16, 1:1, 4:3, 3:4 for text to video; image to video follows your source image |
| Audio | Generated with the video, on by default |
Each clip is a single generation. A 30-second clip is not stitched from shorter pieces.
QwenCloud's own API offers a little more than we expose today. Its parameter list includes an adaptive ratio and 21:9, 2-second clips, and reference inputs for images, video, audio, files and links. Its resolution list stops at 1080P, so 4K is not an option anywhere.
Pricing
fal publishes Wan 3.0 at $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p, and notes the rates can change. On ImagineToVideo that becomes credits:
| Length | 480p | 720p | 1080p |
|---|---|---|---|
| 5 seconds | 10 credits | 20 credits | 40 credits |
| 10 seconds | 25 credits | 45 credits | 85 credits |
| 15 seconds | 35 credits | 70 credits | 135 credits |
| 30 seconds | 70 credits | 135 credits | 270 credits |
The exact cost appears next to the Generate button before you start. Failed generations refund their credits. New accounts can start with the $0.99 Starter Pack: 30 credits, enough for one 5-second 720p Wan 3.0 clip. It is a one-time purchase, not a subscription.
⚠️ QwenCloud's own API reference does not state a price. Per-second figures quoted for direct QwenCloud access come from third-party pages, so we do not repeat them here.
Wan 3.0 or Another Model?
Wan 3.0 is the default answer when you want long clips with sound at the lowest cost. Another model is the better pick when your shot needs something Wan 3.0 does not do here:
| You need | Pick | Why |
|---|---|---|
| Long clips on a budget | Wan 3.0 | Up to 30 seconds at up to 1080p, cheapest per second |
| 4K | VEO 3.1 | The only model here with 4K, at 4, 6 or 8 seconds |
| Reference images or video | Seedance 2.5 or Seedance 2.0 | Up to nine reference images plus reference videos |
| Spoken lines with lip sync | Seedance 2.5 | Lines in double quotes come back as lip-synced dialogue |
| 21:9 framing | Seedance 2.5 or Seedance 2.0 | Both support 21:9 |
For a side-by-side with the other 30-second model on the site, see Wan 3.0 vs Seedance 2.5. The Wan 3.0 alternatives page covers the rest.
How to Use Wan 3.0
- Open text to video or image to video. Wan 3.0 is in the model list with a Cheapest badge.
- Write your prompt. Describe the subject, the action, the camera and the sound you want. For image to video, upload a start frame.
- Optionally add an end frame. Wan 3.0 then animates from the first image to the second, which is useful for transitions and before-and-after shots.
- Choose length and resolution, check the credit cost, and generate. Clips arrive with audio and can be downloaded as MP4.
A practical tip: draft at 480p or 720p and short lengths, then rerun the prompt you like at 1080p or 30 seconds. A 5-second 480p test costs 10 credits.
What Is Not Available Yet
- Reference-to-video. Wan 3.0's API has a separate reference-to-video mode. We have not wired it in yet.
- 21:9 and 2-second clips. Both exist upstream but are not offered in our generator.
- wan3.0-video-prime. Alibaba's faster variant is not available on ImagineToVideo.
FAQ
Is Wan 3.0 available on ImagineToVideo? Yes. Since October 21, 2026 Wan 3.0 runs in the ImagineToVideo generator for text to video and image to video, with an optional end frame, at 480p, 720p or 1080p and up to 30 seconds with audio.
How much does Wan 3.0 cost? fal publishes Wan 3.0 at $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p. On ImagineToVideo a 5-second clip costs 10 credits at 480p, 20 at 720p and 40 at 1080p, and a 30-second 1080p clip costs 270 credits.
Is Wan 3.0 open source? No weights have been published. As of October 21, 2026 there is no Wan 3.0 repository on the Wan-AI organisation on Hugging Face or the Wan-Video organisation on GitHub, although earlier Wan models such as Wan2.2 were released openly.
Does Wan 3.0 support 4K? No. QwenCloud's API reference lists 480P, 720P and 1080P as the only resolutions. If you need 4K on ImagineToVideo, use VEO 3.1.
When was Wan 3.0 released? QwenCloud's release log lists Wan 3.0 as an invite-only test on August 6, 2026 and marks it generally available on August 24, 2026. ImagineToVideo added it on October 21, 2026.
Try it now in the Wan 3.0 text to video generator.
Changelog
- 2026-10-21: published on the day Wan 3.0 went live on ImagineToVideo
