Gemini Omni 1.2 Flash

Gemini Omni 1.2 Flash is the new generation of Google video model with multimodal input: text, images, audio and video at once. Generation from scratch, from first and last frame. Native audio, 1–15 seconds. Faster and cheaper than Omni 1.1; unlike 1.1 it does not accept seed.

Параметры

Field	Description
model	google/gemini-omni-1.2-flash
prompt	Scene description, required; no vendor limit found (104,000 characters accepted), Mixen limit is 30,000
seconds	Integer 1 to 15; default 5
size	Size from the table below; default 1280x720
input_reference	Image-to-video: one image as the start frame, data: URL
input_references	Not supported by the model: references return 400
input_video	Input video for editing and extending
seed	Not supported by the model: passing it returns 400

Resolutions and frame formats (360p is for drafts; 1080p and 4K are upscaled):

Ratio	360p	720p	1080p	4K
16:9	640x360	1280x720	1920x1080	3840x2160
9:16	360x640	720x1280	1080x1920	2160x3840

Examples

Text-to-video

Python

import requests

job = requests.post(
    "https://api.mixen.ai/v1/videos",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "model": "google/gemini-omni-1.2-flash",
        "prompt": "Неоновый город под дождём, камера скользит вдоль мокрого асфальта, отражения вывесок",
        "seconds": "8",
        "size": "1280x720",
    },
).json()
print(job["id"], job["status"])

cURL

curl https://api.mixen.ai/v1/videos \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemini-omni-1.2-flash",
    "prompt": "Неоновый город под дождём, камера скользит вдоль мокрого асфальта, отражения вывесок",
    "seconds": "8",
    "size": "1280x720"
  }'

Polling and download

curl https://api.mixen.ai/v1/videos/video_... \
  -H "Authorization: Bearer $MIXEN_API_KEY"

curl -L https://api.mixen.ai/v1/videos/video_.../content \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -o omni12.mp4

Statuses: queued → in_progress → completed or failed. A sane poll interval is 5–10 seconds. See the video guide for the full cycle.

Price

Per-second billing: 360p — 3.28 ₽ per second, 720p — 11.95 ₽, 1080p — 17.52 ₽, 4K — 34.22 ₽. Current prices — catalog.

Strengths and limitations

• Multimodal input at once: text, images, audio and video in a single request.
• First and last frame (start_image_url + end_image_url).
• Editing and extending an existing clip via input_video.
• Faster and cheaper than Omni 1.1 at the same formats.
• No seed and no references; only two frame formats — 16:9 and 9:16.

All models — in the catalog. The general video workflow — video guide.