Model Details
Wan 2.7 turns a written prompt into a 2-15 second video clip that arrives with its own soundtrack — background music or sound effects generated to match what is happening on screen, in the same pass as the picture. Shot structure is driven entirely by prompt language: ask for a multi-shot video, or describe two or three shots with timestamps, and the cuts are rendered without any separate shot-type control. Pick 720P for drafts or 1080P for delivery, set a clip length anywhere from 2 to 15 seconds, and choose a 16:9, 9:16, 1:1, 4:3 or 3:4 frame to match landscape, vertical or square delivery.
## Best for - Short scenes that need sound as well as picture, with background music or sound effects generated in the same pass - Multi-shot sequences described in a single prompt — "generate a multi-shot video", or shots called out with timestamps - Vertical 9:16 and square 1:1 social clips, or 16:9 landscape b-roll, from text alone - Single takes up to 15 seconds long, without stitching shorter clips together - Concepting and storyboarding, where you iterate at 720P and re-run the keeper at 1080P
## Choose another model when - You want to animate an existing photo or starting frame — this model takes text only, with no reference image, video or audio input, so use an image-to-video model - You need a silent clip — audio is always generated and there is no documented way to switch it off - You need clips longer than 15 seconds, or frame-accurate timeline control - Your prompt is not in Chinese or English, or runs past 5,000 characters, beyond which it is truncated
## Tips - Write shot structure into the prompt itself — "generate a multi-shot video", or "0-3s: …, 3-6s: …" — rather than looking for a shot-type control - Concrete motion verbs ("slowly pans", "rushes forward") translate well to on-screen movement - `resolution` affects cost: 720P bills at $0.10 per second of finished video and 1080P (the default) at $0.15, so iterate at 720P - `enable_prompt_expansion` is on by default and rewrites your prompt for richer detail — it suits short prompts but adds some processing time; disable it to follow your exact wording
To run via the ModelRunner JavaScript client: ```js import { modelrunner } from "@modelrunner/client";
const result = await modelrunner.subscribe("wan-video/wan/v2.7/text-to-video", { input: { prompt: "Generate a multi-shot video: a lighthouse beam sweeping over dark water, then a close-up of waves breaking on rocks", resolution: "1080P", duration: 8, aspect_ratio: "16:9", }, }); ```



