Gemini Omni 1.2 Flash AI Video Generator
Fifteen seconds. Edit in conversation.
Google made Gemini Omni 1.2 Flash generally available on 5 Nov 2026. Generate or edit a clip from 3 to 15 seconds, then refine the same clip in later turns. The model ID is gemini-omni-1.2-flash. Generate it on Topview.
- 3–15s clips
- Multi-turn edits
- 16:9 and 9:16
- Available on Topview
Trusted by leading companies
One take. Then the next edit.
Single case clips from the video library, in 9:16, 16:9, and 21:9. Comparison boards are not included. These are not Gemini Omni 1.2 Flash outputs.
What's new in Gemini Omni 1.2 Flash
- Duration01
Fifteen seconds in one take
Generate and edit natively up to 15 seconds (3–15s, 1s steps), up from 10s on 1.1. First–last frame and per-turn extension stay on a 3–10s ceiling.
Google blog and Cloud model page, Nov 2026.
- Edit02
Edit the same clip, turn by turn
Conversational multi-turn edits via the Interactions API. Swap a product, the lighting, the weather, or the aspect ratio without regenerating the whole clip.
Google blog and Gemini API Omni guide.
- References03
Keep the product. Keep the face.
Up to 6 reference images on the Gemini API (Cloud lists 10). Up to 3 reference videos for 3–10s clips; only 1 video reference, about 3 seconds, for 11–15s clips. Audio inputs are not open on the public Gemini API.
Google blog, Gemini API Omni guide, and Cloud model page.
- Draft to 4K04
Draft at 720p. Upscale what you keep.
Iterate at 360p or 720p, then upscale selected clips to 1080p or 4K. Do not treat every clip as native 4K.
Google and Cloud blogs, Nov 2026.
Model parameters
Gemini Omni 1.2 Flash settings from Google's Cloud model page, checked 6 Nov 2026.
How to use Gemini Omni 1.2 Flash
Step 1: Set the shot
Pick 16:9 or 9:16, a duration from 3–15s, and draft at 360p or 720p. Add product or character references.
Step 2: Generate or mark the edit
For a new clip, prompt the action, camera, and dialogue. For a change, keep the same clip and describe only what should move.
Step 3: Upscale and take it into Topview
Upscale keepers to 1080p or 4K, then continue on Board with captions, Avatar, or a longer assembly when you need more than 15 seconds. Generate 1.2 on Topview.
Where Gemini Omni 1.2 Flash fits
Short clips with sound that continue in Topview video workflows.
Presenter stays. Scene changes. Keep the performance and change the place in a later turn.
Then on Topview: Product Avatar.
One hero clip. Many market cuts. Restage the background and framing for each channel on the same clip.
Then on Topview: Voiceover and captions.
A full scene beat in fifteen seconds Hold a short scene with dialogue in one generation, instead of stitching ten-second pieces.
Then on Topview: Drama Studio.
Catalog stills → 15s product video Turn packaging photos into one product clip. Keep the label and colors steady across the turn.
Then on Topview: review versions on Board.
First frame in. Last frame out. Bridge a start frame and an end frame. That path stops at 10 seconds.
Then on Topview: Storyboard.
Client markups become the next turn Describe only the requested change. Leave the rest of the clip alone.
Then on Topview: keep each version on Board.
References lock the SKU Use the real product photos so the edit changes the asked detail and not the package.
Then on Topview: Product Photography.
Callouts that track the shot A longer clip can carry on-screen labels. Google's example is a macro explainer with motion-tracked callouts.
Then on Topview: AI Ads Video.
Video models
Other video models
Create with Gemini Omni 1.2 Flash.
Fifteen-second clips with native audio and conversational edits. Generate them on Topview.
Google lists about $0.07 per second at 720p on the Gemini API pricing page. Audio reference uploads are not open on the public Gemini API. Cloud audio references are allowlisted. Topview credits for 1.2 are not set.