Choose a category
3 models available
8 models available
3 models available
5 models available
4 models available
8 models available
Wan 2.6 Text to Video
Wan's multi-shot storytelling in one prompt, up to 15 seconds.
Creative Vision
Describe the scene, the camera and the motion.
Video Parameters
Cinema Preview
Generation History0 records
Wan 2.6 Text to Video Generator
Wan 2.6 is Alibaba's December 2025 video model, and its distinguishing trick is multi-shot: one prompt can come back as a sequence of cuts that hold the same subject, not a single unbroken take. Five, ten or fifteen seconds at 720p or 1080p.
Multi-shot is off by default — turn it on only when your prompt actually describes more than one shot
Why Choose Wan 2.6 Text to Video
Alibaba unveiled the Wan 2.6 series on 16 December 2025 with storytelling as the stated goal rather than raw clip quality. The practical result is a model that will cut between shots on its own and keep the subject recognisable across the cut.
Multi-Shot in a Single Prompt
Switch on Multi Shots and Wan 2.6 breaks the clip into several shots with transitions between them, keeping the subject consistent. It is the difference between a moving image and something that reads as a scene.
Up to Fifteen Seconds
Three fixed lengths — 5, 10 and 15 seconds. Fifteen is the point at which a multi-shot prompt has room to breathe; below ten, extra cuts tend to feel rushed rather than deliberate.
720p or 1080p
A flat per-clip price at each tier rather than per-second billing, which makes budgeting simple: you are choosing between two numbers, not computing a rate.
Room for a Long Prompt
The prompt field holds 5,000 characters — enough to write an ordered shot list with camera, subject and lighting for each beat, which is exactly what the multi-shot mode consumes.
Length or Clarity, Same Price
Fifteen seconds at 720p and ten seconds at 1080p both cost 50 credits. That is a genuine fork in the road: the same spend buys you either a longer story or a sharper one.
Failed Runs Are Refunded
Credits are held when you submit and released back automatically if the generation fails or the provider never calls back. Nothing is charged for a clip you never receive.
Start at 720p and Five Seconds
Twenty credits is the cheapest useful run on this page, and it already shows you whether the model understood the scene. Scale up length and resolution once the prompt is right.
How to Use Wan 2.6 Text to Video
The workflow here splits by whether you want one shot or several. Single-shot prompts behave like any other text-to-video model; multi-shot prompts want to be written like a shot list, in order.
Write the Scene
Subject, setting, lighting and camera. Wan responds well to concrete cinematography: shallow depth of field, handheld follow, backlit at dusk. Vague adjectives like beautiful or cinematic do less work than a stated lens or light source.
Decide Single or Multi Shot
Leave Multi Shots off for one continuous take. Turn it on and write the beats in order — wide establishing, then close on the hands, then a reaction — and the model will cut between them.
Draft at 720p, Five Seconds
Twenty credits. Composition, subject and motion style are all legible at 720p, and those are what most prompts get wrong on the first attempt. Resolution is the last thing to fix, not the first.
Render the Full Version
Move to 1080p and the length your story needs — 75 credits for the full fifteen seconds. Generation takes a few minutes, you can leave the page, and the clip lands in your history when it finishes.
No Seed, No Negative Prompt
Wan 2.6 exposes prompt, duration, resolution and the multi-shot switch, and nothing else. If your work depends on reproducing an exact result or excluding something by name, Wan 2.7 has both a seed and a negative prompt.
Wan 2.6 Text to Video
FAQ
Multi-shot mode, resolutions, clip length, audio and credit costs for the Wan 2.6 text-to-video model.
Wan 2.6 is a video generation model from Alibaba, announced as part of the Wan 2.6 series on 16 December 2025. The series was built around multi-shot storytelling and longer outputs — up to fifteen seconds — rather than incremental gains on single-clip fidelity. This page runs its text-to-video endpoint; we handle generation, credit accounting and result storage. FlowVeo3 is an independent platform and is not affiliated with Alibaba.
No. The endpoint behind this page returns silent video and exposes no audio setting. Alibaba's wider 2.6 series does include audio-visual work, but that is not what this integration provides — so plan on adding sound yourself, or use Kling 2.6 or HappyHorse 1.1 on this site, both of which generate audio with the picture.
Between 20 and 75 credits, priced per clip rather than per second. At 720p: 20 credits for 5 seconds, 35 for 10, 50 for 15. At 1080p: 25 for 5 seconds, 50 for 10, 75 for 15. The cost for your current settings is shown before you submit.
With it off, you get one continuous take. With it on, the model divides the clip into several shots with transitions and works to keep the subject consistent across them. It costs no extra credits, but it only helps if the prompt describes more than one beat — enabling it on a single-action prompt usually just produces unmotivated cuts.
Different parameter set, not just a newer number. Wan 2.6 gives you three fixed durations and the multi-shot switch. Wan 2.7 gives you any length from 2 to 15 seconds, a seed for repeatable output, a negative prompt, and first/last frame control on its image endpoint — and it bills per second, starting at 8 credits. Pick 2.6 for multi-shot storytelling, 2.7 for fine control.
At 50 credits you can have 15 seconds at 720p or 10 seconds at 1080p. For social video that will be watched on a phone, the longer 720p clip almost always wins. For anything that will be shown on a large screen or cut into other 1080p footage, take the resolution.
A few minutes typically, and longer for fifteen-second 1080p runs. You can leave the page while it works and the clip appears in your history when it is ready. Finished videos stay available for seven days, so download anything you want to keep. Failed runs are refunded automatically.
Try Wan 2.6 Text to Video
Five seconds at 720p costs 20 credits and shows you how the model reads your scene.