Wan 2.7 is the current top of Alibaba’s Wan (Tongyi Wanxiang) video line, a family that earned its reputation by pairing strong visual quality with an open, developer-friendly ecosystem. Where earlier releases focused on generating clips from scratch, 2.7 broadens the toolkit: it can also rewrite footage you already have and build scenes around the people, products, or styles you show it.
Four modes run on Flux 3 AI. Text-to-video and image-to-video cover classic generation, video edit applies a prompt directly to an uploaded clip, and reference-to-video accepts up to nine reference images so a character or product stays recognizable in the output. Style presets are available across all four, and the edit mode includes an audio option for the source clip.
Every mode renders 5 or 10 seconds at 720p or 1080p; generation modes add square 1:1 alongside 16:9 and 9:16, while edit sticks to widescreen and vertical. A practical loop: lock a look with reference-to-video, then use video edit to revise the take instead of re-rolling from zero.
See what's possible on Flux 3 AI
Sample generations from the platform — write your own prompt to see what Wan 2.7 does with it.
Wan 2.7 vs similar AI video models
Specs from the live catalog — every model below is available on the same account.
| Spec | Wan 2.7 | Wan 2.6 | Wan 2.5 |
|---|---|---|---|
| Max clip length | 10s | 15s | 10s |
| Durations | 5s, 10s | 5s, 10s, 15s | 5s, 10s |
| Resolutions | 720p · 1080p | 720p · 1080p | 720p · 1080p |
| Aspect ratios | 3 | 3 | 3 |
| Audio | |||
| Image to video | |||
| Reference images | up to 9 | up to 1 | up to 1 |
| Keyframes | |||
| Negative prompt |
Capabilities reflect what each model supports on Flux 3 AI today.
What makes Wan 2.7 stand out
Four ways in
Start from a prompt, a still image, an existing video, or a stack of reference images — one model covers the whole pipeline.
Nine reference images
Reference-to-video mode accepts up to 9 images, letting you pin characters, wardrobe, products, and style in a single generation.
Prompt-based video editing
Upload a clip and describe the change — Wan 2.7 re-renders the footage instead of forcing you to regenerate the scene.
Full HD output
Every 2.7 mode renders at 720p or 1080p, with square 1:1 framing available on the generation modes for feed-native posts.
Style presets
Apply a consistent visual treatment from the Create studio without wrestling long style descriptions into your prompt.
Popular use cases
Why creators run Wan 2.7 on Flux 3 AI
- Run all four Wan 2.7 modes plus 40+ other models from a single account.
- One shared credit pool — you pay only for the renders you actually generate.
- Test the same brief on Wan 2.7, Kling, and Veo side by side before committing.
Wan 2.7 specs & capabilities
Output
- Clip duration5, 10 seconds
- Resolutions720p · 1080p
- Aspect ratios16:9 · 9:16 · 1:1
- Generation modesText to Video · Image to Video · Video Editing · Reference to Video
Controls
- Image to video (start frame)
- Reference images
- First & last frame keyframes
- Audio generation
- Negative prompt
- Video input / editing
- Extend video
Which Wan 2.7 version should you use?
Every version below is selectable in the Create studio on the same credits.
Wan 2.7 (T2V)
The from-scratch mode: describe the shot and Wan 2.7 generates it with nothing uploaded. Pick it when no source material exists yet. Renders 5- or 10-second clips at 720p or 1080p in 16:9, 9:16, or 1:1, with style presets available. It renders silent video and takes no reference images — that job belongs to R2V.
Wan 2.7 (I2V)
Animates an uploaded still, treating it as the anchor of the scene, and accepts one reference image to keep the subject faithful. Choose it when you already have the exact frame — a product photo, a portrait — the video should grow from. Output matches T2V: 5 or 10 seconds, 720p or 1080p, in 16:9, 9:16, or 1:1.
Wan 2.7 Video Edit
The revision tool: upload existing footage, describe the change, and 2.7 re-renders the clip. Pick it over regenerating when the take is right but details need altering. It supports one reference image and an audio setting for the source clip’s sound, outputting 5 or 10 seconds at 720p or 1080p — 16:9 or 9:16 only, no square.
Wan 2.7 R2V
The consistency specialist: reference-to-video composes a new scene around up to 9 reference images, the highest count in the Wan family. Choose it whenever a character, product, or style must stay recognizable across a series of clips. It renders the standard 5- or 10-second, 720p or 1080p output in 16:9, 9:16, or 1:1.
Wan 2.7 prompt examples
Copy one as a starting point, or send it straight to the Create studio.
“A ceramic artist shaping a bowl on a spinning wheel in a sunlit studio, slow orbit around the wheel, wet clay glistening, dust motes drifting in golden light”
Tests fine hand motion and rotating-object coherence in text-to-video
“Using the referenced sneaker and model, a low-angle hero shot on a wet rooftop at dusk, slow 360 orbit, neon reflections rippling in puddles”
Shows reference-to-video holding product and person consistent together
“Change the season to deep winter: add falling snow, frost the windows, shift the grade to cool blue, keep the actors’ movement untouched”
Demonstrates prompt-based video editing applied to uploaded footage
Create with Wan 2.7 in three steps
Describe your idea
Write a prompt or upload a starting image — the more specific the scene, motion, and style, the better Wan 2.7 performs.
Pick Wan 2.7 & settings
Choose duration, resolution, and aspect ratio. The studio shows the exact credit cost before you generate.
Generate & download
Wan 2.7 renders in the cloud — track progress in your library, then download watermark-free or share with a link.
Wan 2.7 questions, answered
What is Wan 2.7?
Wan 2.7 is the latest release in Alibaba’s Wan video-generation family. Beyond text-to-video and image-to-video, it adds prompt-based video editing and a reference-to-video mode that composes scenes from up to nine reference images. All four modes run in the browser on Flux 3 AI.
What does Wan 2.7 video edit do?
You upload an existing clip and describe the change in a prompt; Wan 2.7 re-renders the footage accordingly. Edits output 5 or 10 seconds at 720p or 1080p in 16:9 or 9:16, and the mode includes an audio option for handling the source clip’s sound.
How does reference-to-video work in Wan 2.7?
Reference-to-video takes up to 9 reference images along with your prompt and generates a new clip that keeps those subjects consistent. It’s the strongest option in the Wan family for recurring characters or exact product appearances across a series of videos.
What durations and resolutions does Wan 2.7 support?
All Wan 2.7 modes render 5- or 10-second clips at 720p or 1080p. Text-to-video, image-to-video, and reference-to-video offer 16:9, 9:16, and 1:1 aspect ratios; video edit supports 16:9 and 9:16.
Wan 2.7 vs Wan 2.6 — which should I pick?
Pick Wan 2.7 for editing existing footage or multi-reference consistency — 2.6 has neither. Pick Wan 2.6 when you want 15-second text-to-video clips or its generate-audio toggle, since 2.7’s generation modes top out at 10 seconds and render silent video.
Does Wan 2.7 generate audio?
No — Wan 2.7’s generation modes produce video without a generated soundtrack. If you want audio created with the clip, use Wan 2.6 or Wan 2.5 text-to-video on Flux 3 AI, which include a generate-audio toggle, or score the 2.7 output in your editor.
Is Wan 2.7 free to try?
Sign-up on Flux 3 AI comes with free credits that work on any listed model, Wan 2.7 included. Once those run out, the pricing page lists plans and credit packs, and everything you render downloads without a watermark.
Do I need an Alibaba or Tongyi account to use Wan 2.7?
No. Wan 2.7 runs directly on Flux 3 AI, so a single Flux 3 AI account covers it along with every other model on the platform — Kling, Sora, Veo, and more. There is no separate Alibaba Cloud or Tongyi Wanxiang sign-up, subscription, or region requirement.
Can I access Wan 2.7 through an API?
Yes — Flux 3 AI offers API access for developers who want to build Wan 2.7 generation into their own apps and pipelines. Details are outlined in the API section on the homepage, and the same account and credit balance you use in the browser applies.