Choose a category
3 models available
8 models available
3 models available
2 models available
4 models available
8 models available
Wan 3.0 Prime Image to Video
Start from a still on Wan 3.0's separately priced Prime tier, and optionally pin the ending.
Frames
First frame is required; add a last frame to pin where the shot ends.
Creative Vision
Describe the motion, not the picture — the model can already see it.
Video Parameters
Priced per second — 2 seconds at 480P is 6 credits, 30 seconds at 1080P is 345.
Each step up doubles the price. Draft at 480P, finish at 1080P.
Auto keeps your first frame's shape — usually what you want.
AI audio synced to the video. Does not change the price.
Cinema Preview
Generation History0 records
Wan 3.0 Prime Image to Video Generator
Wan 3.0 Prime is a separately priced tier of Alibaba's Wan 3.0. Upload a first frame, optionally pin a last frame, describe the motion, and get 2 to 30 seconds at up to 1080P with a native soundtrack. The public controls match standard and Prime costs about 1.6 times as much. We have not published a controlled quality or speed benchmark, so use identical frames and settings when comparing the tiers.
Keep both frames, the prompt, seed, length and resolution unchanged when comparing Prime with standard
What Changes With Wan 3.0 Prime Image to Video
Prime accepts exactly what standard Wan 3.0 image to video accepts: a first frame, an optional last frame, a motion prompt and an optional audio reference. The upstream model ID and price change, with Prime costing about 1.6 times as much. Until a controlled benchmark is available, compare the tiers with identical inputs instead of assuming a particular advantage.
A Separate Tier, Tested With the Same Frames
Prime is a separate upstream model with the same public parameter schema as standard Wan 3.0. Reuse the first frame, last frame, prompt, seed, duration, resolution and ratio so any output difference comes from the tier rather than a changed setup.
First Frame Required, Last Frame Optional
One JPEG, PNG or WebP up to 20MB starts the clip. Add a second image and the shot is pinned to end on it, which is how you get a product turn that lands squarely or a transition that arrives at an exact composition.
Any Length From 2 to 30 Seconds
A continuous slider billed by the second. Two seconds is a blink or a reaction; thirty is a full move across a set. You pay for the length you pick, not the next tier up.
480P, 720P or 1080P, Auto Ratio
Each resolution step roughly doubles the price. Leave the aspect ratio on Auto and the clip keeps the shape of your first frame; pick 16:9, 9:16, 1:1, 4:3 or 3:4 when the output has to fit a specific placement.
Native Audio, Plus an Audio Reference
A generated soundtrack is on by default and does not change the price. Add one mp3 or wav of up to 15 seconds as @Audio1 and the motion follows its rhythm, music or voice; switch audio off if you will score the clip yourself.
Failed Runs Are Refunded
Credits are held when you submit and returned automatically if the generation fails or never comes back. A 345-credit master that does not arrive costs you nothing.
Make It a Controlled Comparison
Find the motion prompt on standard, then change nothing but the model. Compare both files at the same size and playback conditions before deciding whether Prime is worth the extra credits for your image.
How to Use Wan 3.0 Prime Image to Video
Four steps that keep the tier as the only variable. Settle the frames and motion on standard, save the exact settings, then pay for a like-for-like Prime comparison.
Upload the First Frame
A clean, well-lit image with the subject clearly visible, up to 20MB. This is frame one of the clip, so its framing and aspect ratio are the clip's unless you override the ratio. Add a last frame only if the ending genuinely has to land on a specific picture.
Describe the Motion, Not the Image
The model can already see the picture. Write what changes: "she turns from the window and walks toward the camera, which pulls back slowly; rain on the glass behind her." Chinese and English both work, with room for 20,000 characters.
Create a Standard Baseline
Run the frames and prompt on Wan 3.0 standard image to video at 480P for a few seconds. Save the exact seed and settings. This shows whether the motion reads and the last frame is reachable before the more expensive comparison.
Run the Prime Comparison
Bring every input and setting over unchanged and switch only the model tier. Compare the outputs before increasing length or resolution. Expect a few minutes; you can leave the page and the clip waits in your history for seven days.
One Frame In, Not Ten References
This page starts from a specific image. If you want a character or product to appear without dictating the opening frame, the Wan 3.0 Prime Video page takes up to 10 reference images instead; the two modes cannot be combined in one run.
Wan 3.0 Prime Image to Video
FAQ
What Prime changes, first and last frames, aspect ratio, audio and credit costs for Wan 3.0 Prime image to video.
It is a separately priced upstream tier of Alibaba's Wan 3.0 image-to-video mode. Upload a first frame, optionally a last frame, describe the motion, and get a 2 to 30 second clip at 480P, 720P or 1080P. We handle generation, credit accounting and result storage. FlowVeo3 is an independent platform and is not affiliated with Alibaba.
There is no honest universal answer without a controlled comparison. Prime uses a separate upstream model ID and costs about 1.6 times as much, but we have not published a benchmark proving a specific quality or speed advantage. Start with standard; then repeat the same frames, prompt, seed, duration and resolution here and decide from the two outputs.
Between 6 and 345 credits, set by length and resolution. Two seconds at 480P is 6 credits. Five seconds costs 15 at 480P, 30 at 720P and 60 at 1080P. A full 30 seconds is 175 credits at 720P and 345 at 1080P. Adding a last frame or an audio reference does not change the price. The exact cost is shown before you submit.
It pins where the shot ends. The model generates motion that starts on your first frame and arrives at your last, which is how you get a turn that ends square-on or a reveal that lands on an exact composition. Leave it empty and the model decides the ending from the prompt. Keep the two images close in style and framing; a large jump between them forces an awkward transition.
Auto keeps the shape of your first frame, so a 4:5 portrait stays 4:5 and nothing is cropped. Choosing 16:9, 9:16, 1:1, 4:3 or 3:4 instead reframes the clip to that ratio, which is useful when the output has to fit a specific placement but means parts of your image may be cut or extended.
A native audio track is generated with the picture, on by default and free; switch it off if you will add your own music in an editor. To steer it, add one mp3 or wav of up to 15 seconds and refer to it as @Audio1 in the prompt. It sets the rhythm, music or voice; it is not lip-sync compositing, so expect the model to interpret it rather than copy it.
Usually a few minutes, and noticeably longer for a 30-second clip at 1080P. You can leave the page while it works; the clip appears in your history and stays available for seven days, so download anything you want to keep. If a generation fails, the held credits are returned to your balance automatically.
Ready for a Prime Comparison?
Build a standard baseline, repeat it here with the same inputs, and decide from your own two outputs whether Prime earns its higher price.