Skip to content
API Reference

Gemini Omni

Gemini Omni is Google's multimodal video generation model — the successor direction for Veo. One model covers text-to-video, image-to-video, reference-guided generation, editing, and extension, and it generates an audio track (voice, music, sound effects) by default — you control it through the prompt.

POST https://llm.siraya.ai/v1/videos/generations

Supported Models

Model ID Description
gemini-omni-1.1-flash-preview Gemini Omni 1.1 Flash, video generation with audio
Authorization string Required
Your API Key (e.g., `Bearer `).

Body

model string Required
The model ID: `gemini-omni-1.1-flash-preview`.
prompt string Required
A text description of the video to generate. For editing or extension, describe the change ("remove the violin, keep everything else") or the continuation.
seconds integer Default: 10
Duration of the generated video: an integer between 3 and 10. Values outside the range are rejected with a `400`. When omitted, the model generates a 10-second video. Extension requests accumulate up to 40 seconds total.
resolution string Default: "720p"
Output resolution: `360p`, `720p`, `1080p`, or `4k`.
aspect_ratio string
Aspect ratio: `16:9` (landscape) or `9:16` (portrait).
size string
Pixel dimensions, e.g. 1280x720. A standard size resolves to resolution + aspect_ratio; if you send both, they must agree with size or the request returns a 400.
image_url string
(Optional) Input image URL for image-to-video generation.
frame_images array
First-frame and/or last-frame control. Each item has a `frame_type` (`first_frame` or `last_frame`) and an `image_url`.
input_references array
Reference materials: up to 10 images, or one video of up to 3 seconds (a video reference's audio track is ignored). Each item has a `type` (`image` or `video`), a `url`, and an optional `role`. Audio references are not supported by this model. For extension, pass the source video here (`type: "video"`, 1–30s).
task string
Omni-specific. Names the workflow: `text_to_video`, `image_to_video`, `reference_to_video`, `edit`, or `extend`. When omitted, the model picks the workflow from the materials you attach — send `task` when the intent is ambiguous, which is mainly edit vs extend (both take a source video). Vendor parameters are always top-level fields on this model; `extra_body` is not accepted.
async boolean Default: false
When false, the request waits for generation to finish and returns the completed video URL. When true, it returns immediately with a video id you poll for status. See Asynchronous generation.
callbackUrl string
Only used when async is true. An http(s) URL the router will POST to once generation completes (or fails), so you don't have to poll.

Not supported

seed, generate_audio (audio is always generated — steer it with the prompt), negative_prompt, and output_format have no effect on this model and are dropped; extra_body is refused with a 400 — send vendor parameters like task at the top level instead.

curl https://llm.siraya.ai/v1/videos/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "model": "gemini-omni-1.1-flash-preview",
    "prompt": "A cute baby sea otter",
    "seconds": 5,
    "resolution": "720p"
  }'

Edit a source video (explicit task):

curl https://llm.siraya.ai/v1/videos/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "model": "gemini-omni-1.1-flash-preview",
    "prompt": "Remove the violin from the scene, keep everything else unchanged",
    "task": "edit",
    "input_references": [
      { "type": "video", "url": "https://example.com/source.mp4" }
    ]
  }'
import requests

url = "https://llm.siraya.ai/v1/videos/generations"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer <API_KEY>"
}
data = {
    "model": "gemini-omni-1.1-flash-preview",
    "prompt": "A cute baby sea otter",
    "seconds": 5,
    "resolution": "720p"
}

response = requests.post(url, headers=headers, json=data)
print(response.json())

Example Response

{
    "created": 1760347750,
    "data": [
        {
            "url": "https://resources.siraya.ai/v1/videos/video_n6Sn-1aIT3z8x.../content"
        }
    ],
    "error": {}
}

With "async": true the status response has no output_url; download the result from GET /v1/videos/{video_id}/content. See Asynchronous generation.