Text-to-Video

Doc version：v1.0.0 | Last updated：2026-07-22
This platform fully supports the official Tongyi Wanxiang (Wan) video generation API. Requests and responses are transparently proxied; parameter semantics are identical to the official API.

Generate video from a text prompt. The wan2.7/wan2.6 series support automatic audio dubbing (or pass in an audio file for audio-visual sync) and multi-shot storytelling, up to 1080P and up to 15 seconds.

The official API splits Text-to-Video into two request protocols. The submission endpoint is the same; the request body parameters differ:

New protocol (wan2.7-t2v series): output specs are controlled by resolution (resolution tier) + ratio (aspect ratio);

Legacy protocol (wan2.6 and earlier models): output specs are controlled by size (exact pixels, "width*height").

Model Comparison

Model	Protocol	Resolution control	Duration (seconds)	Audio	Multi-shot	Prompt limit
`wan2.7-t2v`, `wan2.7-t2v-2026-06-12`	New	`resolution` + `ratio` (720P/1080P)	Integer 2–15, default 5	With audio (auto dubbing or `audio_url`)	Supported, controlled via natural language in prompt	5000 characters
`wan2.6-t2v`	Legacy	`size` (720P/1080P tiers)	Integer 2–15, default 5	With audio (auto dubbing or `audio_url`)	Supported, requires `shot_type="multi"`	1500 characters
`wan2.6-t2v-us`	Legacy	`size` (720P/1080P tiers)	5 or 10, default 5	Same as wan2.6	Same as wan2.6	1500 characters
`wan2.5-t2v-preview`	Legacy	`size` (480P/720P/1080P tiers)	5 or 10, default 5	With audio (auto dubbing or `audio_url`)	Not supported	1500 characters
`wan2.2-t2v-plus`	Legacy	`size` (480P/1080P tiers)	Fixed 5, must not be passed	Silent	Not supported	800 characters
`wanx2.1-t2v-turbo`	Legacy	`size` (480P/720P tiers)	Fixed 5, must not be passed	Silent	Not supported	800 characters
`wanx2.1-t2v-plus`	Legacy	`size` (720P tier only)	Fixed 5, must not be passed	Silent	Not supported	800 characters

Create Task

POST https://platform.dataeyes.ai/ali/api/v1/services/aigc/video-generation/video-synthesis

Request Headers

Parameter	Type	Required	Default	Description
`Content-Type`	string	Yes	application/json	Data exchange format
`Authorization`	string	Yes		Authentication, `Bearer {API_KEY}`

The X-DashScope-Async: enable header required by the official API is added automatically by the platform; you do not need to include it.

Request Body — New Protocol (wan2.7 Series)

Parameter	Type	Required	Default	Description
`model`	string	Yes		`wan2.7-t2v`, `wan2.7-t2v-2026-06-12`
`input.prompt`	string	Yes		Text prompt, Chinese or English, ≤5000 characters; excess is automatically truncated
`input.negative_prompt`	string	No		Negative prompt describing content you do not want, ≤500 characters
`input.audio_url`	string	No	Auto dubbing if omitted	Audio file URL (publicly accessible). Format wav/mp3, duration 2–30 seconds, ≤15MB; audio longer than the video duration is truncated, and if shorter, the trailing segment is silent
`parameters.resolution`	string	No	1080P	Resolution tier: `720P`, `1080P`; directly affects cost
`parameters.ratio`	string	No	16:9	Aspect ratio: `16:9`, `9:16`, `1:1`, `4:3`, `3:4`
`parameters.duration`	integer	No	5	Video duration in seconds, integer in [2, 15]; directly affects cost
`parameters.prompt_extend`	boolean	No	true	Intelligent prompt rewriting; noticeably improves short prompts but increases processing time
`parameters.watermark`	boolean	No	false	When `true`, adds an "AI-generated" watermark at the bottom-right corner of the video
`parameters.seed`	integer	No	Random	Random seed, range [0, 2147483647]; identical seeds do not guarantee identical results

wan2.7 does not support the size and shot_type fields (they take no effect). Control single/multi-shot directly in the prompt with natural language, e.g. "generate a multi-shot video" or timestamped shot breakdowns like "Shot 1 [0–3s]…"; if unspecified, the model decides based on the prompt semantics.

Request Body — Legacy Protocol (wan2.6 / wan2.5 / wan2.2 / wanx2.1)

Parameter	Type	Required	Default	Description
`model`	string	Yes		See the model comparison table above, e.g. `wan2.6-t2v`
`input.prompt`	string	Yes		Text prompt; wan2.6/wan2.5 series ≤1500 characters, wan2.2/wanx2.1 series ≤800 characters, excess is automatically truncated
`input.negative_prompt`	string	No		Negative prompt, ≤500 characters
`input.audio_url`	string	No	Auto dubbing if omitted	Supported by wan2.6/wan2.5 series only. Format wav/mp3, duration 3–30 seconds, ≤15MB
`parameters.size`	string	No	Model-dependent (see table below)	Output resolution; must be an exact pixel value such as `1280720` — not* `1:1` or `480P`; valid values depend on the model and directly affect cost
`parameters.duration`	integer	No	5	Video duration in seconds. wan2.6-t2v: [2, 15]; wan2.6-t2v-us / wan2.5-t2v-preview: 5 or 10; wan2.2/wanx2.1 series are fixed at 5 seconds and must not be passed
`parameters.shot_type`	string	No	single	wan2.6 only. `single` (single shot) / `multi` (multi-shot); only takes effect when `prompt_extend=true`, and takes precedence over shot descriptions in the prompt
`parameters.prompt_extend`	boolean	No	true	Same as new protocol
`parameters.watermark`	boolean	No	false	Same as new protocol
`parameters.seed`	integer	No	Random	Same as new protocol

Resolution Tiers and Pixel Mapping

New protocol (wan2.7): resolution × ratio → output pixels

Tier	16:9	9:16	1:1	4:3	3:4
720P	1280*720	720*1280	960*960	1104*832	832*1104
1080P	1920*1080	1080*1920	1440*1440	1648*1248	1248*1648

Legacy protocol: valid size values

Tier	Valid size values (width*height: aspect ratio)
480P	832480 (16:9), 480832 (9:16), 624*624 (1:1)
720P	1280720 (16:9), 7201280 (9:16), 960960 (1:1), 1088832 (4:3), 832*1088 (3:4)
1080P	19201080 (16:9), 10801920 (9:16), 14401440 (1:1), 16321248 (4:3), 1248*1632 (3:4)

Available tiers and defaults per model

Model	Available tiers	Default
`wan2.7-t2v(-2026-06-12)`	720P, 1080P (resolution)	1080P / 16:9
`wan2.6-t2v`, `wan2.6-t2v-us`	All sizes in the 720P and 1080P tiers	1920*1080
`wan2.5-t2v-preview`	All sizes in the 480P, 720P and 1080P tiers	1920*1080
`wan2.2-t2v-plus`	All sizes in the 480P and 1080P tiers	1920*1080
`wanx2.1-t2v-turbo`	All sizes in the 480P and 720P tiers	1280*720
`wanx2.1-t2v-plus`	All sizes in the 720P tier only	1280*720

Note: in the legacy 720P/1080P tiers, the 4:3 and 3:4 pixel values (1088*832, 1632*1248, etc.) differ from the new-protocol wan2.7 values (1104*832, 1648*1248, etc.); do not mix them up.

Request Example (wan2.7-t2v, New Protocol)

curl --request POST \
  --url 'https://platform.dataeyes.ai/ali/api/v1/services/aigc/video-generation/video-synthesis' \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "wan2.7-t2v",
    "input": {
      "prompt": "At sunset, waves gently lap against the rocks while a sailboat drifts by in the distance"
    },
    "parameters": {
      "resolution": "720P",
      "ratio": "16:9",
      "duration": 5
    }
  }'

Request Example (wan2.6-t2v, Legacy Protocol)

curl --request POST \
  --url 'https://platform.dataeyes.ai/ali/api/v1/services/aigc/video-generation/video-synthesis' \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "wan2.6-t2v",
    "input": {
      "prompt": "On a spring day, cherry blossoms drift down over a classical courtyard as a gentle breeze blows"
    },
    "parameters": {
      "size": "1280*720",
      "duration": 5
    }
  }'

Response Example

{
  "output": {
    "task_status": "PENDING",
    "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx"
  },
  "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx"
}

Response Fields

Field	Type	Description
`output.task_id`	string	Task ID used for polling; valid for 24 hours
`output.task_status`	string	Task status; `PENDING` after successful creation
`request_id`	string	Unique request identifier, useful for troubleshooting
`code` / `message`	string	Error code and details, returned only when creation fails

Notes

Fixed-duration models: wan2.2-t2v-plus, wanx2.1-t2v-turbo, and wanx2.1-t2v-plus always generate 5-second videos. Do not pass duration — doing so returns an error (duration customization is not supported).

Prompt over-length is auto-truncated: see the model comparison table for per-model prompt limits (5000/1500/800 characters); the excess is truncated silently without an error. negative_prompt is limited to 500 characters.

Multi-shot: wan2.7 removed the shot_type parameter — describe the shot structure directly in the prompt with natural language or timestamped shot breakdowns; wan2.6 requires shot_type: "multi" with prompt_extend: true.

Audio requirements: audio_url is supported by the wan2.7/wan2.6/wan2.5 series only; wav/mp3 format, duration 2–30 seconds for wan2.7 and 3–30 seconds for legacy models, size ≤15MB. If omitted, the model dubs the video automatically (wan2.2/wanx2.1 series generate silent videos only).

Billing: charged by the number of seconds of successfully generated video; resolution and duration directly affect cost. Inputs are not billed, and failed tasks are not billed.

Result validity: both the task_id and the generated video URL are valid for 24 hours; download or archive the video promptly.

Query Task

After creating a task, poll the task status and results with the returned output.task_id:

GET https://platform.dataeyes.ai/ali/api/v1/tasks/{task_id}

See 09-Query-Task for details.

01-Text-to-Video

Text-to-Video#

Model Comparison#

Create Task#

Request Headers#

Request Body — New Protocol (wan2.7 Series)#

Request Body — Legacy Protocol (wan2.6 / wan2.5 / wan2.2 / wanx2.1)#

Resolution Tiers and Pixel Mapping#

Request Example (wan2.7-t2v, New Protocol)#

Request Example (wan2.6-t2v, Legacy Protocol)#

Response Example#

Response Fields#

Notes#

Query Task#