MiniMax H3 | BeatAPI API Documentation

curl -X POST https://api.beatapi.io/v1/videos/tasks \
     -H "Authorization: Bearer <token>" \
     -H "Content-Type: application/json" \
     -d '{
  "model": "minimax-h3",
  "prompt": "A slow cinematic push through a misty mountain village at dawn.",
  "duration": 5,
  "aspect_ratio": "16:9",
  "resolution": "768P"
}'

{
  "data": {
    "capability_id": "minimax-h3",
    "capability_version": null,
    "completed_at": null,
    "created_at": 1782210000,
    "error_code": null,
    "error_message": null,
    "id": "task_8K2qA",
    "media_type": "video",
    "model": "minimax-h3",
    "object": "task",
    "output": null,
    "request_id": "req_abc123",
    "stage": "queued",
    "status": "queued",
    "task_kind": "video",
    "updated_at": 1782210000,
    "usage": {
      "billable_duration_seconds": 5,
      "credits_charged": 0.2,
      "credits_refunded": 0,
      "credits_reserved": 0.2,
      "credits_settled": 0
    }
  }
}

Generate 4–15 second videos from text, first/last frames, or multimodal references, with output up to 2K.

Supported modes

Mode	Required input	Rules
Text to video	`model`, `prompt`	Omit all media arrays.
First frame	`model`, `prompt`, `images`	Provide one image URL.
First and last frames	`model`, `prompt`, `images`	Provide two image URLs in first-to-last order.
Multimodal references	`model`, `prompt`, one `reference_*` array	Do not combine `images` with `reference_*` inputs.

Async flow: Create the task, save data.id, then poll the Task with GET /v1/tasks/{task_id}. On succeeded, read data.output.media. On failed, read data.error_code and data.error_message, then follow the error and retry policy.

Using a local file? upload a local input first and pass the returned data.url.

Generate 4–15 second videos from text, first/last frames, or multimodal references, with output up to 2K. ### Supported modes | Mode | Required input | Rules | | --- | --- | --- | | Text to video | `model`, `prompt` | Omit all media arrays. | | First frame | `model`, `prompt`, `images` | Provide one image URL. | | First and last frames | `model`, `prompt`, `images` | Provide two image URLs in first-to-last order. | | Multimodal references | `model`, `prompt`, one `reference_*` array | Do not combine `images` with `reference_*` inputs. | **Async flow:** Create the task, save `data.id`, then [poll the Task](/quick-guide#get-task-status) with `GET /v1/tasks/{task_id}`. On `succeeded`, read `data.output.media`. On `failed`, read `data.error_code` and `data.error_message`, then follow the [error and retry policy](/quick-guide#errors-and-retry-policy). **Using a local file?** [upload a local input](/quick-guide#upload-an-input-file) first and pass the returned `data.url`.

Authentication

AuthorizationBearer

Send your API key as Authorization: Bearer <BEATAPI_API_KEY>.

Request

This endpoint expects an object.

model"minimax-h3"Required

Must be minimax-h3.

promptstringRequired1-5000 characters

Video generation instructions.

imageslist of stringsOptional

One first-frame image or first- and last-frame images as public HTTPS URLs.

reference_imageslist of stringsOptional

Public HTTPS image references for multimodal reference generation.

reference_videoslist of stringsOptional

Public HTTPS video references for multimodal reference generation.

reference_audioslist of stringsOptional

Public HTTPS audio references for multimodal reference generation. Audio also requires at least one reference image or video.

durationintegerOptional4-15Defaults to 5

Requested output duration in seconds.

aspect_ratioenumOptional

Text mode defaults to 16:9 and does not accept adaptive. Frame mode always uses adaptive. Reference mode defaults to adaptive and also accepts a concrete ratio.

resolutionenumOptionalDefaults to 768P

Output resolution tier.

Response

Video generation task accepted

Errors

400

Bad Request Error

401

Unauthorized Error

402

Payment Required Error

409

Conflict Error

429

Too Many Requests Error

Generate 4–15 second videos from text, first/last frames, or multimodal references, with output up to 2K.

Supported modes

Mode	Required input	Rules
Text to video	`model`, `prompt`	Omit all media arrays.
First frame	`model`, `prompt`, `images`	Provide one image URL.
First and last frames	`model`, `prompt`, `images`	Provide two image URLs in first-to-last order.
Multimodal references	`model`, `prompt`, one `reference_*` array	Do not combine `images` with `reference_*` inputs.

Using a local file? upload a local input first and pass the returned data.url.

Supported modes

Authentication

Headers

Request

Response

Errors

Supported modes