Wan 2.7
Wan 2.7 is the newest Wan model family currently being integrated on WanVideoFree. The hosted endpoints cover text-to-video, first-frame or first-and-last-frame image-to-video, video continuation, and short video editing, with 720p or 1080p output and native audio options. WanVideoFree is independently operated and is not the official Wan website.
What is new in the hosted Wan 2.7 endpoints?
| Workflow | Inputs | Current verified schema |
|---|---|---|
| Text to video | prompt; optional driving audio | 2–15 seconds, 720p/1080p, five aspect ratios |
| Image to video | first frame; optional last frame, prompt, audio | 2–15 seconds, 720p/1080p; can continue a 2–10 second clip |
| Edit video | 2–10 second MP4/MOV, edit instruction; optional reference image | 720p/1080p; output can match or truncate input duration |
The text endpoint can create background audio when no driving audio is supplied. The image endpoint accepts either an image-driven start or a video continuation input. The edit endpoint is for changing an existing clip rather than inventing a shot from nothing.
Which Wan 2.7 workflow should you choose?
- Choose text-to-video when composition can be invented from a description.
- Choose image-to-video when the subject, product, character, or first frame must remain recognizable.
- Choose video continuation when an existing short clip should carry on.
- Choose edit-video when timing should stay similar but the visual treatment or scene details must change.
Cost for a controlled first test
WanVideoFree locks initial Wan 2.7 text and image tests to five seconds at 720p for 50 credits. The first edit-video tier is three seconds at 720p for 30 credits so it cannot request an output longer than a short input. A 1080p five-second text or image generation is 75 credits. These credits follow the current hosted provider rate and are rechecked before a model moves from verification into the public selector.
Start with 720p. Use 1080p only after motion and composition are acceptable; higher resolution cannot rescue a poor instruction.
Is Wan 2.7 open source?
As of the fact-check date below, the official Wan-Video GitHub organization publicly exposes Wan2.1 and Wan2.2 repositories. The Wan 2.7 capability described here is available through a hosted provider endpoint, but that is not the same as an official open-weight release. We therefore do not describe Wan 2.7 as open source. This answer will change only when an official model source supports it.
A prompt that uses the new model well
A porcelain robot finishes folding a blue paper crane and lifts it toward the morning window. One continuous slow clockwise orbit, warm natural light, subtle paper sounds and quiet room ambience, physically plausible hands, no captions, no logo.
For edit-video, describe only the transformation and what must remain: Restyle the entire clip as soft watercolor on textured paper. Preserve the original subject, timing, camera path, and composition.
Sources
- Wan 2.7 text-to-video API schema
- Wan 2.7 image-to-video API schema
- Wan 2.7 edit-video API schema
- Official Wan-Video GitHub organization
Last fact check: August 12, 2026.
Frequently asked questions
Does Wan 2.7 support image to video?
Yes. The hosted schema accepts a first-frame image, an optional last-frame image, and an optional prompt. It also accepts a short input video for continuation, but a video cannot be combined with the first-frame image in the same request.
How long can a Wan 2.7 video be?
Text-to-video and image-to-video expose durations from 2 to 15 seconds. Edit-video accepts a 2–10 second input and can match that input duration or truncate it to a selected 2–10 second length.
Does Wan 2.7 generate audio?
The text and image endpoints include audio controls. Text-to-video can create matching background audio when no driving-audio file is provided; image-to-video can accept a driving-audio URL.
Is Wan 2.7 available on WanVideoFree now?
A model appears in the public selector only after its paid canary produces a playable output and its locked credits match provider cost. A page can be published for factual evaluation while a particular endpoint remains marked as verifying.