Omni-Flash-Ext Video Generation

• Omni-Flash-Ext unified video generation model
• Supports Text-to-Video, single-image Image-to-Video, reference video, and 3-reference-image fusion
• Supports 720p/1080p/4k resolution and 4/6/8/10 second duration
• Asynchronous task API. Submit a task first, then query the result by task ID

POST /v1/videos/generations

cURL

curl --request POST \
  --url https://api.apimart.ai/v1/videos/generations \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "Omni-Flash-Ext",
    "prompt": "a girl is dancing happily in a sunny garden",
    "duration": 10,
    "resolution": "1080p",
    "aspect_ratio": "9:16"
  }'

200

{
  "code": 200,
  "data": [
    {
      "status": "submitted",
      "task_id": "task_01KS1H7ZYSJWH1N779S2FSHTKA"
    }
  ]
}

cURL

curl --request POST \
  --url https://api.apimart.ai/v1/videos/generations \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "Omni-Flash-Ext",
    "prompt": "a girl is dancing happily in a sunny garden",
    "duration": 10,
    "resolution": "1080p",
    "aspect_ratio": "9:16"
  }'

200

{
  "code": 200,
  "data": [
    {
      "status": "submitted",
      "task_id": "task_01KS1H7ZYSJWH1N779S2FSHTKA"
    }
  ]
}

Authentication

Authorization · string · required

All requests require Bearer Token authentication.

Get an API key:

Visit the API Key management page to get your API key.

Add the following header when making requests:

Authorization: Bearer YOUR_API_KEY

Request Parameters

model · string · required

Video generation model name. Must be Omni-Flash-Ext.

prompt · string · required

Video content description. We recommend describing the scene, subject, action, environment, camera movement, visual style, and audio cues in detail.

Example: "a girl is dancing happily in a sunny garden"

duration · integer · default: "6"

Video duration in seconds.

Supported values: 4, 6, 8, 10.

Other values such as 5 or 7 will return an invalid_duration error.

Do not pass duration when uploading a reference video. duration and video_urls cannot be passed at the same time.

resolution · string · default: "720p"

Video resolution. Values are case-insensitive.

Supported values:

• 720p
• 1080p
• 4k

Other resolutions will return an invalid_resolution error.

aspect_ratio · string · default: "16:9"

Video aspect ratio. Use it to control landscape or portrait output.

Common values:

• 16:9 - landscape
• 9:16 - portrait

Default: 16:9

size · string

Compatibility field. It has the same meaning as aspect_ratio. If both are provided, keep them consistent.

image_urls · array<url>

Reference image URL array. You can omit it, provide 1 image, or provide 3 images:

• Omitted or empty array: Text-to-Video
• 1 image: single-image Image-to-Video
• 3 images: reference image fusion

Only publicly accessible image URLs are supported.

The first-frame plus last-frame mode with 2 images is not supported. Passing 2 images returns an unsupported_image_count error. 4 or more images have not been fully verified and are not recommended.

video_urls · array<url>

Reference video URL array. You can omit it or provide 1 reference video.

Only publicly accessible HTTP/HTTPS video URLs are supported. You can pass it together with image_urls: images are used as identity or composition references, while the video is used as motion reference.

Omni-Flash-Ext supports only 0 or 1 reference video. Passing 2 or more videos returns an unsupported_video_count error.

Do not pass duration when passing video_urls. video_urls and duration cannot be passed at the same time.

Response

code · integer

Response status code. Successful requests return 200.

data · array

Returned task array.

Show Array item

status · string

Initial task status. It is submitted after successful submission.

task_id · string

Unique task ID for querying task status and result.

Query Task Result

Video generation is asynchronous. After submission, the API returns a task_id. Use the Get task status endpoint to query progress and results.

cURL

curl --request GET \
  --url https://api.apimart.ai/v1/tasks/task_01KS1H7ZYSJWH1N779S2FSHTKA \
  --header 'Authorization: Bearer <token>'

We recommend waiting 5-10 seconds after submission before the first query, then polling every 5-10 seconds. A single task usually completes in about 3-5 minutes.

Successful Result Example

{
  "code": 200,
  "data": {
    "id": "task_01KS1H7ZYSJWH1N779S2FSHTKA",
    "status": "completed",
    "progress": 100,
    "created": 1779246294,
    "completed": 1779246534,
    "actual_time": 240,
    "estimated_time": 600,
    "cost": 0.4,
    "result": {
      "videos": [
        {
          "url": ["https://cdn.example.com/videos/abc.mp4"],
          "expires_at": 1779332760
        }
      ]
    }
  }
}

Failed Result Example

{
  "code": 200,
  "data": {
    "id": "task_01KS1H7ZYSJWH1N779S2FSHTKA",
    "status": "failed",
    "progress": 100,
    "created": 1779246294,
    "completed": 1779246534,
    "actual_time": 240,
    "estimated_time": 600,
    "cost": 0,
    "error": {
      "message": "invalid duration 7, must be one of 4/6/8/10",
      "code": "task_failed"
    }
  }
}

Use Cases

Scenario 1: Text-to-Video

{
  "model": "Omni-Flash-Ext",
  "prompt": "a beautiful sunset over the ocean with seagulls flying",
  "duration": 6,
  "resolution": "720p",
  "aspect_ratio": "16:9"
}

Scenario 2: Single-Image Video

{
  "model": "Omni-Flash-Ext",
  "prompt": "make the character smile and slowly turn around, cinematic camera motion",
  "duration": 6,
  "resolution": "1080p",
  "aspect_ratio": "9:16",
  "image_urls": ["https://example.com/character.jpg"]
}

Scenario 3: 3-Reference-Image Fusion

{
  "model": "Omni-Flash-Ext",
  "prompt": "a creative scene combining these elements with smooth camera motion",
  "duration": 10,
  "resolution": "1080p",
  "aspect_ratio": "9:16",
  "image_urls": [
    "https://example.com/scene.jpg",
    "https://example.com/character.jpg",
    "https://example.com/product.jpg"
  ]
}

Scenario 4: 4K Short Video

{
  "model": "Omni-Flash-Ext",
  "prompt": "close-up of a hummingbird hovering in front of a red flower",
  "duration": 4,
  "resolution": "4k",
  "aspect_ratio": "16:9"
}

Scenario 5: Reference Video Generation

{
  "model": "Omni-Flash-Ext",
  "prompt": "the same scene but at night with neon lights",
  "resolution": "720p",
  "aspect_ratio": "16:9",
  "video_urls": ["https://example.com/reference.mp4"]
}

Error Codes

HTTP	Error type	Meaning	Suggested action
400	invalid_request_error	model is not Omni-Flash-Ext, prompt is empty, or the JSON format is invalid	Check the request body
400	invalid_duration	duration is not 4, 6, 8, or 10	Use a supported duration
400	invalid_resolution	resolution is not 720p, 1080p, or 4k	Use a supported resolution
400	unsupported_image_count	The number of image_urls is unsupported, commonly caused by passing 2 images	Use 0, 1, or 3 images
400	unsupported_video_count	The number of video_urls is unsupported, commonly caused by passing 2 or more videos	Use 0 or 1 reference video
401	authentication_error	Invalid token	Check the Bearer Token
402	payment_required	Insufficient balance	Recharge and try again
429	rate_limit_error	Rate limit exceeded	Reduce concurrency or try again later

When a task fails, the task status API returns the failure reason in data.error. Common causes include temporary upstream quota exhaustion, content moderation failure, or upstream timeout.