Documents and spreadsheets
docx, doc, xlsx, xls, pptx, ppt, pdf, txt and md. One file per generation.
Alibaba's newest video model reads a document, a spreadsheet or a web page and generates the video it describes — up to 30 seconds in one run, with the audio built in.
Input formats no other model takes
Wan 3.0 reads the document, not a picture of it. Hand it a spreadsheet and it cites your cell values; hand it a deck and it makes the film the deck describes. One file or one URL per generation.
docx, doc, xlsx, xls, pptx, ppt, pdf, txt and md. One file per generation.
Pass a URL instead of a file and Wan 3.0 works from the page's content.
Read the full prompt guide: six official formulas, from one line to a multi-shot script →
Edit what you generated
Add, replace, remove, relight, restage, restyle, redub, edit by reference, or re-plot the scene entirely — each one an instruction against a clip you already have.
Specifications
How it works
Sign up at pixo.video and pick Wan 3.0 from the model list. New accounts start with free credits.
Type a prompt, or attach one document or web URL — or up to 10 images, 5 clips and 5 audio tracks as references.
Take the result up to 30 seconds with audio, then change any part of it with an edit instruction instead of re-rolling.
FAQ
Wan 3.0 is Alibaba's newest video generation model, released in August 2026 by the Tongyi Lab. It generates up to 30 seconds of video in a single run with audio produced alongside the picture, and it accepts documents and web pages as input in addition to text, images, video and audio. You can run it at pixo.video.
Anywhere from 2 to 30 seconds in one generation, at 480p, 720p or 1080p. If you feed it a video as input, the input and output durations together stay within 30 seconds.
docx, doc, xlsx, xls, pptx, ppt, pdf, txt and md — or a web URL instead. It is one or the other, and one item per generation: a single file, or a single URL.
Up to 10 images, up to 5 video clips totalling 15 seconds, and up to 5 audio tracks totalling 15 seconds. The prompt itself can run to 20,000 characters.
Yes, and it is a separate mode: a generation either pins first and last frames, or uses reference assets. The two cannot be combined in one run.
Yes. Pass the clip and describe the change: add or remove an element, replace a subject, relight the scene, restage where things stand, restyle it, change a spoken line, edit by reference to an image, or re-plot the scene. The rest of the shot stays as it was.
Yes. Pixo runs Wan 3.0 next to Seedance, Veo, Kling and MiniMax H3, and you can combine their shots in one project at pixo.video.
Sign up at pixo.video, open the Playground and pick Wan 3.0. New users get 200 free credits on sign-up, and plans are currently up to 55% off.
Wan 3.0 is live in Pixo's Playground, alongside Seedance, Veo, Kling and MiniMax H3.