Create Video Generation Task

Version: v1.0.0 | Last Updated: 2026-08-13
This platform fully supports the MiniMax Video Generation V2 (MiniMax-H3) official API. Requests and responses are transparently proxied, and all parameter semantics remain consistent with the official API.

The model generates a video from multimodal inputs — text, images, video, and audio — with native 2K output. This is an asynchronous endpoint: on success it returns a task_id, which you poll via the Query Task endpoint.

POST https://platform.dataeyes.ai/hailuo/v2/video_generation

Request Parameters

Parameter	Type	Required	Default	Description
`model`	string (enum)	Yes	—	Model name. Currently: `MiniMax-H3`
`content`	array	Yes	—	Multimodal input array (see below). Must include one non-empty `text` item
`resolution`	string (enum)	Yes	—	Video resolution: `768P` or `2K`
`duration`	integer	Yes	—	Output video duration in seconds, integer. Allowed values: `4`-`15`
`ratio`	string (enum)	Conditional	`adaptive`	Aspect ratio; rules vary by scenario (see below)
`callback_url`	string	No	—	Webhook URL for task status changes (see below)
`aigc_watermark`	boolean	No	`false`	Whether to add an AIGC watermark to the generated video

content Input Combinations

Each element of the content array is typed via type (text / image_url / video_url / audio_url) and its purpose is marked via role. Different combinations map to different generation scenarios:

Scenario	content combination
Text-to-video (t2va)	A single `text` item
Image-to-video, first frame	`text` + 1 `image_url` (`role=first_frame`; with a single image, `role` may be omitted and defaults to first frame)
Image-to-video, last frame	`text` + 1 `image_url` (`role=last_frame`)
Image-to-video, first + last frame	`text` + 2 `image_url` items (`role` set to `first_frame` and `last_frame` respectively)
Multimodal-reference-to-video (r2va)	`text` + any combination of reference images (`role=reference_image`), reference videos (`role=reference_video`), and reference audio (`role=reference_audio`)

Image-to-video and multimodal reference are mutually exclusive: once any reference_image / reference_video / reference_audio role appears in content, first_frame / last_frame must not appear (and vice versa).

content Element Structure

Field	Type	Description
`type`	string (enum)	`text` / `image_url` / `video_url` / `audio_url`
`text`	string	Prompt text. Every request must include one non-empty text item. Up to 7,000 characters per `text`
`image_url.url`	string	Image address (required when `type=image_url`)
`video_url.url`	string	Video address (required when `type=video_url`; multimodal reference only)
`audio_url.url`	string	Audio address (required when `type=audio_url`; multimodal reference only)
`role`	string (enum)	Purpose of the item, conditionally required: `first_frame` / `last_frame` / `reference_image` / `reference_video` / `reference_audio`

Each url accepts three forms:

A public URL;

mm_file://{file_id} (references an existing platform file, e.g. the file_id of a previous output);

A data:<mime>;base64,<Base64> data URI (lowercase mime subtype, e.g. data:image/png;base64,...).

The total request body must be ≤ 64 MB, and Base64 encoding inflates size by ~33%. Use public URLs or mm_file:// for large files instead of Base64.

Input Media Limits

Images (image_url):

Item	Limit
Formats	JPG, JPEG, PNG, WEBP, HEIC, HEIF
Single file size	≤ 30 MB
Width / height	[256, 5760] px
Aspect ratio (w/h)	[0.4, 2.5]
Count	first frame ≤ 1, last frame ≤ 1, reference images ≤ 9

Videos (video_url, multimodal reference only):

Item	Limit
Container / format	MP4 (`.mp4`), MOV (`.mov`)
Codecs	Video: H.264/AVC, H.265/HEVC; audio: AAC, MP3
Single file size	≤ 50 MB
Count	≤ 3
Duration	[2, 15] s per clip; total ≤ 15 s
Width / height	[256, 5760] px
Aspect ratio (w/h)	[0.4, 2.5]
Frame rate	[23.976, 60]

Audio (audio_url, multimodal reference only):

Item	Limit
Formats	WAV, MP3
Single file size	≤ 15 MB
Count	≤ 3
Duration	[2, 15] s per clip; total ≤ 15 s

ratio Rules

Defaults to adaptive (the most suitable aspect ratio is chosen automatically from the inputs; the actual ratio is available in the query endpoint's ratio field). Allowed values: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16.

Scenario	Rule
Text-to-video	Required, and must not be `adaptive`
Image-to-video	Determined by the input image; always `adaptive`. Other values are ignored without error
Multimodal reference	Optional; defaults to `adaptive`, or specify any concrete ratio

callback_url Webhook

When configured, the MiniMax server first sends a verification request containing a challenge field (echo the challenge back within 3 seconds to complete verification). After verification, every task status change is POSTed to the URL; the payload structure matches the Query Task response. Callback status values: queued / running / succeeded / failed / cancelled.

Request Examples

Text-to-video:

Image-to-video (first frame):

Multimodal-reference-to-video:

Response Example

{
  "task_id": "424010985738629"
}

Field	Type	Description
`task_id`	string	Task ID; use it with Query Task to retrieve status and results

Billing

Billed per second on usage.total_seconds (input reference video seconds + output seconds) at the resolution's unit price; the first 5 input images per task are free, additional images are billed per image. An estimated fee is pre-charged at submission and settled against actual usage on completion (difference charged or refunded); failed tasks are fully refunded. See Overview · Billing & Settlement.

01-Create-Video-Generation

Create Video Generation Task#

Request Parameters#

content Input Combinations#

content Element Structure#

Input Media Limits#

ratio Rules#

callback_url Webhook#

Request Examples#

Response Example#

Billing#