UlazAI - AI Image & Video Tools
Now live in Video Studio
Gemini Omni Flash 1.1
Create 4–10 second videos from one prompt. Add images, a source video, saved voice IDs or reusable characters when the shot needs tighter control. Choose 360p, 720p, 1080p or 4K before generating.
See what Flash 1.1 actually made
Three unedited 8-second 1080p generations. Play a clip, then open the exact prompt.
Turn Google Flow's four new Omni showcases into working briefs →
Exact prompt
Single continuous cinematic shot, eight seconds. An elderly man sits alone at a small kitchen table at blue hour, one empty chair opposite him and an old music box between them. He gently opens the music box. Warm golden fireflies rise from it and form the soft, fleeting silhouette of his late dance partner beside the empty chair. He recognizes her, his eyes fill and he smiles. The silhouette touches his hand, then becomes fireflies again as dawn light enters the room. Intimate natural performance, realistic skin, restrained emotion, slow dolly in, shallow depth of field, subtle film grain, physically believable light, premium cinema, no dialogue, no text, no logo, no watermark, no cuts.
Exact prompt
Single continuous cinematic shot, eight seconds. A rain-soaked rooftop garden after a violent storm, broken pots and dark clouds. A tired young woman in a yellow raincoat kneels beside one tiny bent sunflower and shields it with both hands. The rain stops. One beam of sunrise breaks through the clouds, the sunflower slowly lifts, and hundreds of luminous birds emerge from the city windows and sweep across the sky. Her face changes from exhaustion to astonished hope. Camera starts close on her muddy hands, then cranes up into the vast glowing skyline, realistic motion and weather, emotional but restrained, premium cinema, no dialogue, no text, no logo, no watermark, no cuts.
Exact prompt
Single continuous cinematic shot, eight seconds. A quiet train platform at sunrise. An older firefighter steps off the train, still carrying a worn duffel bag. His grey rescue dog waits at the far end, sees him, freezes for one heartbeat, then sprints at full speed. The firefighter drops the bag and kneels. The dog leaps into his arms and both tumble into a joyful embrace as early sunlight floods the platform. Tears, laughter, tail whipping, real weight and contact, subtle handheld camera that runs with the dog then settles close on the reunion, natural faces and fur, premium cinematic realism, no dialogue, no text, no logo, no watermark, no cuts.
Gemini Omni Flash 1.2 makes the same shot for about half the credits
We ran one prompt through both versions on 10 November 2026. Same first frame (a cappuccino, a barista's hand with a milk pitcher), 720p, 4 seconds. Both poured the milk and drew a heart. Flash 1.1 cost 90 credits. Flash 1.2 cost 44.
The difference you can see: 1.2 finished the heart and pulled the pitcher out of frame by the third second, then held the shot. 1.1 used all four seconds for the pour. Neither is wrong. If you want the action to breathe, 1.1 is still there. If you want a finished shot with room to cut, 1.2 gets there sooner and costs less.
Exact prompt
The barista tilts the steel pitcher and pours steamed milk into the cappuccino, drawing a clean heart of latte art, then lifts the pitcher out of frame. Static camera, soft café ambience, quiet clink of the pitcher.
Exact prompt
The barista tilts the steel pitcher and pours steamed milk into the cappuccino, drawing a clean heart of latte art, then lifts the pitcher out of frame. Static camera, soft café ambience, quiet clink of the pitcher.
What changed in 1.2
| Flash 1.1 | Flash 1.2 | |
|---|---|---|
| Length | 4, 6, 8 or 10 s | any whole second from 3 to 15 s (3 to 10 s with a first and last frame) |
| 720p, 4 s | 90 credits | 44 credits |
| 1080p, 8 s | 150 credits | 96 credits |
| Billing | fixed steps | per second: 4.8 (360p), 10.8 (720p), 12 (1080p) or 26.4 (4K) credits a second, total rounded up |
| Aspect ratios | 16:9, 9:16, 1:1, 4:3, 3:4 | 16:9, 9:16 or auto |
| Saved voices and character IDs | yes | no |
| Edit an existing clip | yes | yes, clips up to 15 s, billed on the length of your clip |
One thing 1.2 drops: saved voices and character IDs. If your workflow depends on a recurring character, stay on 1.1 for now.
More control without a complicated workflow
Start with text when speed matters. Add references only when clothing, a face, a voice or camera movement has to remain recognisable. Flash 1.1 fits short ads, product shots and social variants that need a fast turnaround.
- Text and images for a fast art-direction route.
- One source video uses two of the seven reference units.
- Up to three character IDs in one generation.
- A first frame is used on its own, without other references.
Video input and 4K change the price
Without a source video, the price scales with duration. With video input, the price is fixed per generation. Studio shows the exact charge before you start.
| Resolution | 4s | 6s | 8s | 10s | With video |
|---|---|---|---|---|---|
| 360p / 720p / 1080p | 90 | 120 | 150 | 180 | 240 |
| 4K | 210 | 240 | 270 | 300 | 360 |
Prices are UlazAI credits per generation. 4K with video input costs 360 credits; all other video-input resolutions cost 240.