Gemini Omni
Video
Gemini Omni Video turns text, images, and clips into cinematic AI video — multimodal references, omni control, and fast renders up to 1080p.
Key Features
Mix up to 50 references
Blend images, video snippets, and audio in one generation. Gemini Omni 1.2 Flash understands what each reference contributes — characters, style, camera motion, or sound — and composes them into a coherent shot.
Draft to final in seconds
The Flash tier is tuned for iteration speed without giving up fidelity. Explore variations at 480p, lock the look, then render the keeper at 720p or 1080p.
Direct every element
Aspect ratios from 16:9 to 21:9, durations from 4 to 30 seconds, first and last frame pinning, and omni-reference inputs give you frame-level creative control.
Try NowHow to Use
Describe your shot
Write a prompt describing the scene, subject, motion, and mood. Attach reference images, video clips, or audio for omni-reference control.
Pick your format
Choose aspect ratio, resolution, duration, and how many variants to generate. Credits are estimated before you commit.
Generate and iterate
Hit generate and watch the task finish in your workspace. Refine the prompt or make a similar variation from any result.
Creative Inspiration Gallery
FAQ
Everything you need to know about Gemini Omni 1.2 Flash
Gemini Omni 1.2 Flash is our flagship omni-reference video model. It generates short videos from text prompts, or guided by images, video clips, and audio you attach — up to 50 multimodal references at once.
Videos render at 480p, 720p, or 1080p, and run 4 to 30 seconds depending on the model settings you pick in the composer.
Choose Omni Reference to blend multiple images, clips, and audio into one shot. First & Last Frame pins the start and end of the video, and Reference Image steers style or subject from a single picture.
The credit cost is estimated live in the composer — it depends on resolution, duration, and how many variants you request. Higher resolutions and longer durations cost more.
Yes — videos you generate on a paid plan can be used commercially. Free-tier outputs are for personal use and evaluation.
Keep exploring
More AI Models
A full studio of frontier video and image models, one composer.
Gemini Omni 1.2 Flash
Omni-reference video generation — up to 50 multimodal references, 480p–1080p, 4–30s.
Seedance 2.5
Cinematic motion with first & last frame control.
Seedance 2.0
Balanced quality and speed for everyday clips.
Seedance 2.0 Fast
Draft-fast renders for rapid iteration.
GPT Image 2.6
Flagship image editing — multi-reference, native 2K.
GPT Image 2.5
Reliable everyday image generation.
GPT Image 2
Fast text-to-image drafts.
Seedream 5 Pro
High-fidelity stylized image output.