Omni Video Models Overview

A model-by-model guide to omni-fast and omni-fast-v2v — video generation, task polling, and content download.

The Omni video models all use the OpenAI-compatible asynchronous /v1/videos task API. After submitting a task, keep the returned id or task_id and poll the task-query endpoint for status. Once a task completes you can read video_url, or fetch the MP4 through the content-download endpoint.

Capability	Endpoint
Create task	`POST /v1/videos`
Query task	`GET /v1/videos/{task_id}`
Download video	`GET /v1/videos/{task_id}/content`

Choosing a model

Model	Use case	Key inputs	Limits
`omni-fast`	Text-to-video, first-frame image-to-video, first/last frame, multi-reference-image video	`prompt`, `first_image_url`, `last_image_url`, `images`	`images` up to 5, no reference video
`omni-fast-v2v`	Reference-video editing, extension, regeneration	`video`, `video_url`, `input_video`	Reference video up to 15MB

Common request

Every Omni video task uses a JSON request body — no multipart upload required. The base fields are:

Field	Type	Description
`model`	string	Required, pass `omni-fast` or `omni-fast-v2v`
`prompt`	string	Required, the generation or edit prompt
`seconds`	string	Video duration, pass as a string, e.g. `"8"`
`aspect_ratio`	string	Aspect ratio, e.g. `"16:9"`, `"9:16"`
`resolution`	string	Resolution, e.g. `"720p"`, `"1080p"`, `"2k"`, `"4k"`

curl -X POST "$MODELSOK_BASE_URL/v1/videos" \
  -H "Authorization: Bearer $MODELSOK_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "omni-fast",
    "prompt": "Ocean waves gently rolling onto a sandy beach at golden hour, cinematic",
    "seconds": "8",
    "aspect_ratio": "16:9",
    "resolution": "720p"
  }'

A successful submission returns an async task object:

{
  "id": "video_abc123",
  "task_id": "video_abc123",
  "object": "video",
  "model": "omni-fast",
  "status": "queued",
  "progress": 0,
  "created_at": 1735689600,
  "video_url": ""
}

Query and download

curl "$MODELSOK_BASE_URL/v1/videos/video_abc123" \
  -H "Authorization: Bearer $MODELSOK_API_KEY"

Once complete, the response usually includes video_url:

{
  "id": "video_abc123",
  "task_id": "video_abc123",
  "object": "video",
  "model": "omni-fast",
  "status": "completed",
  "progress": 100,
  "created_at": 1735689600,
  "completed_at": 1735689900,
  "video_url": "https://example.com/results/omni-fast.mp4"
}

To download the MP4 directly, use the content endpoint:

curl -L "$MODELSOK_BASE_URL/v1/videos/video_abc123/content" \
  -H "Authorization: Bearer $MODELSOK_API_KEY" \
  -o result.mp4

Status and errors

Common status values include queued, in_progress, completed, and failed. When a task fails, read error.message and error.code to diagnose the cause:

{
  "id": "video_abc123",
  "task_id": "video_abc123",
  "object": "video",
  "model": "omni-fast",
  "status": "failed",
  "progress": 100,
  "error": {
    "message": "upstream task failed",
    "code": "task_failed"
  }
}

Poll task status at a fixed interval. After a task succeeds, download or re-host video_url promptly to avoid the temporary link expiring.

Sub-pages

Page	Content
omni-fast video generation	Text-to-video, image-to-video, first/last frame, multi-reference images
omni-fast-v2v reference-video generation	Reference-video editing, extension, regeneration

Choosing a model

Common request

Query and download

Status and errors

Sub-pages

On this page