Wan 3.0 is Alibaba's AI video model, released in August 2026. Turn your text prompts, photos, or reference clips into full 30-second videos with sound at up to 1080p. Sign up free to get 40 starter credits and make your first video.
Turn your prompts, photos and reference clips into complete 30-second videos with dialogue, music and sound effects at up to 1080p.
Sign in and write a line or two first — Enhance rewrites what you have.
Generate 30 fps MP4 videos up to 30 seconds long in a single try, doubling the 15 seconds of Wan 2.7. You can pick one continuous take or direct a storyboard of 4-to-6-second shots across frames like 16:9 and 9:16.
Spend 9 credits per second at 480p, 18 at 720p, and 36 at 1080p, about a third of the cost of Seedance 2.5 AI video maker. Test your prompts at 480p for 270 credits before you render in 1080p for 1,080 credits, and plan your runs with the Wan 3.0 cost calculator.
Get spoken dialogue, background music, and audio effects generated directly with your video frames. Put dialogue in double quotes, write "no music" to remove songs, or turn audio off while keeping the same credit rate.
Upload up to 10 images, 5 video clips, and 5 audio clips to guide your scenes as Image 1, Video 1, or Audio 1. You get stable consistency for characters, props, spaces, and style because Wan 3.0 follows your prompt closely.
Set one picture as your starting frame, or define both first and last frames so your video lands on your chosen end image. You can build your opening picture with the Nano Banana 2 image generator and animate it immediately.
Paste 1 public website link or upload 1 document up to 100 MB and 50 pages, including PDF, Word, and PowerPoint files. Wan 3.0 reads your material and turns it into a clear explainer video with a spoken voiceover.
Upload your product photos and write your lines in the prompt to get a 30-second 9:16 vertical ad with a voiceover and sound effects. You launch polished social campaigns faster without hiring a production crew.
Add your character images and type your storyboard to create 15-to-30-second scenes with spoken dialogue and ambient sound. You stretch your production budget because the same credits make about three times as many seconds as Seedance 2.5.
Provide character sheets or illustrations as references to generate stylized 2D, anime, or comic-book shots with matching audio. You keep your character look consistent across every action take with zero manual drawing.
Drop in a PDF, slide deck, or web link to turn written content into a short explainer video with a clear voiceover. You deliver engaging lessons and product walkthroughs to your students or customers in minutes.
Describe your scene, action and sound, and add your first frame, up to 20 reference files, or a document or web page link.
Choose your video length up to 30 seconds, select 480p, 720p or 1080p resolution, set your aspect ratio, and switch sound on or off.
Click generate to create your video and save your finished 30 fps MP4 file right away.
Wan 3.0 is Alibaba's newest AI video model, created by its Tongyi Lab team. Alibaba writes the name as Wan3.0. It is the 8th generation of the Wan family. The model launched as a public beta on August 6, 2026 and had its full release on August 24, 2026. One single model generates video and sound together from text prompts, photos, reference clips, documents, or web pages, creating up to 30 seconds of video at up to 1080p.
Yes. You can sign up for free on Seadanse with no credit card required. You get 40 starter credits right away, which is enough to make a 4-second 480p Wan 3.0 video for 36 credits. When you want more credits, you can buy credit packs or choose a plan on our Seadanse pricing page.
No. Alibaba keeps the Wan 3.0 model weights closed and runs the model directly in its cloud. Wan 2.2, released in July 2025, is Alibaba's newest video model with open weights. On Seadanse, you run Wan 3.0 right in your web browser with no GPU, no downloads, no API key, and no code.
Alibaba launched the public beta on August 6, 2026 and rolled out the full release on August 24, 2026. Wan 3.0 is generally available right now with no waitlist or application. You can create your first video today on Seadanse.
Seadanse bills Wan 3.0 by the second. You spend 9 credits per second at 480p, 18 credits at 720p, and 36 credits at 1080p. A 30-second video costs 270 credits at 480p, 540 credits at 720p, or 1,080 credits at 1080p. Turning sound on or off costs the exact same amount. The credit cost appears beside Generate before you start, and failed videos refund your credits automatically. You can check any length on our Wan 3.0 cost calculator.
Yes. In the Artificial Analysis Video Arena blind test rankings from September 2026, Wan 3.0 ranks #1 for text to video without audio with an Elo score of 1336. It ranks #2 for text to video with audio with an Elo score of 1229. Users praise Wan 3.0 for following prompt details closely, handling scenes with many reference files, and producing crisp 2D, anime, and comic-book styles.
Both models run on Seadanse from a single credit balance and make up to 30 seconds of video with sound. Wan 3.0 uses about a third of the credits per second that Seedance 2.5 uses: 9, 18, or 36 credits versus 25, 55, or 98 credits at 480p, 720p, or 1080p. Choose Wan 3.0 to follow prompts closely, manage multiple reference assets, generate 2D and anime art, and create more total footage on your budget. Pick the Seedance 2.5 AI video maker when you want lifelike live-action people, complex physics, and precise lip sync.
Wan 3.0 doubles the maximum video length to 30 seconds in a single generation, compared to 15 seconds on Wan 2.7. It unifies text, image, reference, and document inputs into one model. You can now supply up to 20 reference files, including up to 10 images, 5 video clips, and 5 audio clips. It also accepts documents and web links for explainer videos. In blind votes for text to video with sound, Wan 3.0 scores 1229, up from 1149 for Wan 2.7.
Wan 3.0 generates videos up to 30 seconds long in one take at 30 fps in MP4 format. You can select 480p, 720p, or 1080p resolution, with aspect ratios of 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, or Auto. You can stitch two 30-second clips to build a one-minute video, or use Script to Video to link multiple Wan 3.0 scenes into a longer project.
Yes. Wan 3.0 builds speech, background music, and sound effects right along with the video in one generation. Put spoken lines inside double quotation marks in your prompt to generate character speech. Type "no music" if you want dialogue and effects without background music. You can also turn audio off to use your own soundtrack, and the credit cost remains the same.
You can start with just a text prompt, or upload one photo as the starting frame. You can also set a first frame and a last frame to control how the video begins and ends. Wan 3.0 accepts up to 10 images (JPG, PNG, BMP, or WEBP up to 20 MB each), 5 video clips (MP4 or MOV, up to 15 seconds total), and 5 audio clips (WAV or MP3, up to 15 seconds total). You reference these in your prompt as Image 1, Video 1, and Audio 1. You can also upload 1 document up to 50 pages and 100 MB, or provide 1 public web link to make an explainer video.
Structure your prompt in five parts: camera shot and movement, subject details, action, lighting, and sound. Add "one continuous take, no cuts" for a smooth, single shot. For multiple scenes, outline a storyboard using 4-to-6-second shots with time marks. Place character dialogue inside double quotation marks. Alibaba recommends testing your prompt at 480p first, then generating your final take at 1080p. Find tested examples in our Wan 3.0 prompt guide.
Yes. Videos that you produce using any paid plan or credit pack download without watermarks and include a full commercial use license. You can view all options on our Seadanse pricing page.
Wan 3.0 generates clips up to 30 seconds long, while the Hailuo 3 AI video maker creates 4 to 15 seconds and the MiniMax H3 Max AI video maker creates 5 to 15 seconds. In blind Arena testing for text to video with sound, Wan 3.0 scores 1229, MiniMax H3 Max scores 1227, and MiniMax H3 scores 1220. Without audio, Wan 3.0 takes the #1 spot at 1336, ahead of MiniMax H3 at 1302. You can access all three models on Seadanse from a single credit balance.
Every page here runs the real models in your browser — jump to the one that matches the clip you have in mind.
Seedance, Veo, Kling, Hailuo and more, each on its own page with real renders.
Upload a photo, pick an effect, get a short video. No prompt to write.
Articles written for this model: how to prompt it, where it runs and what its limits are.
Wan 3.0 prompts work best when built in five parts. Here are four exact prompts, their reference rules, and the clips they generated.
Sign up with no credit card to get 40 starter credits and make your first Wan 3.0 video on Seadanse right now.
