Storytellers
Plan short character scenes with dialogue and a clear beginning and ending.
Gemini Omni 1.2 Flash generates video with reference-guided scenes and dialogue. It supports longer clips and follow-up edits that refine the same scene instead of starting again.
Explore these video examples, then use a starting prompt to create your own version.
Gemini Omni 1.2 Flash
Fashion in motion
Gemini Omni 1.2 Flash
One scene, different worlds
Gemini Omni 1.2 Flash
Animated conversations
Gemini Omni 1.2 Flash
A flight through the Pantheon
Gemini Omni 1.2 Flash
Inside a living beehive
Gemini Omni 1.2 Flash
From shoreline to island
Google DeepMind's video model for creating scenes and revising them through conversation.
Google announced Gemini Omni 1.2 Flash on November 5, 2026. The release extends native video creation to 15 seconds and improves reference adherence, scene consistency and spoken dialogue.
The model brings creation and revision into the same workflow:
Fast sports action, abrupt script changes and complex physics remain areas Google is working to improve.
Six capabilities for creating and refining video.
Build a short sequence in one generation, with durations from 3 to 15 seconds in one-second steps.
Revise a background, product or lighting through conversation while working with the same clip.
Combine images, short motion references and audio cues to guide the result.
The release aims to reduce changes in characters, objects and environments between frames.
Improved speech and lip-sync help direct conversations involving multiple characters.
Guide continuous camera paths, zoom-outs and orbits with fewer unwanted cuts.
Explore stories, explain ideas and refine visual concepts with clear creative direction.
Plan short character scenes with dialogue and a clear beginning and ending.
Develop brief visual explainers from a scene description and references.
Explore how a visual concept looks in motion, then revise the scene.
Create variations around a product reference without rewriting the entire brief.
Turn a clear scene idea into a video in three steps.
Write the subject, action and camera direction. Add an image reference to guide the visual style or composition.
Select a model in the generator, then choose the aspect ratio, duration and resolution for your video.
Generate your video, review the result and refine your prompt for the next version. Check faces, text and motion before sharing.
Key capabilities and technical specifications of Gemini Omni 1.2 Flash.
| Developer | Google DeepMind |
| Announced | November 5, 2026 |
| Model type | Video generation and conversational editing |
| Native duration | 3–15 seconds, in one-second increments |
| Draft resolution | 360p or 720p |
| Upscaling | 1080p or 4K |
| Reference inputs | Images, motion clips up to 3 seconds, and audio cues |
| Audio | Dialogue with multi-speaker conditioning |
| Provenance | SynthID watermark |
Last updated: November 10, 2026
A focused comparison of the changes described in Google's 1.2 launch announcement.
| Capability | Gemini Omni 1.1 Flash | Gemini Omni 1.2 Flash |
|---|---|---|
| Maximum native clip length | 10 seconds | 15 seconds |
| Reference adherence | Earlier version used in Google's comparison | Improved adherence in Google's comparison |
| Editing across a 15-second clip | Not listed | Multi-turn conversational edits |
Gemini Omni 1.1 Flash Advantages: A previous option for shorter scenes; the launch announcement does not establish an advantage over 1.2 for every task.
Gemini Omni 1.2 Flash Advantages: Longer clips and stronger reference following provide more room to develop and revise a scene.
Consider 1.2 when clip length and iterative editing matter. Compare actual results for your scene rather than assuming either version will always produce the better video.
Choose a plan for your video projects. The generator shows the credit cost for your selected model and settings before you create.
Answers about video creation, references, editing, output quality and credits.
Open the Gemini Omni 1.2 Flash AI Video Generator page on Picir AI. Enter your scene description, choose your model and settings in the generator, and start creating.
Google DeepMind developed Gemini Omni 1.2 Flash. Google announced this version on November 5, 2026.
Google announces native creation and editing up to 15 seconds, with duration selectable from 3 to 15 seconds in one-second increments.
The model supports follow-up conversational edits. You can ask for changes to details such as the setting or lighting rather than beginning with a new brief.
Google describes drafting at 360p or 720p and upscaling selected results to 1080p or 4K. Upscaling should not be confused with native 4K generation.
The credit cost depends on the model and settings selected in the generator, including duration and resolution. Review the displayed cost before generating and choose a plan that fits your projects.
No. Review generated speech, text, identity and motion. Google identifies fast action and complex physics as ongoing areas for improvement, and outputs carry a SynthID watermark.
Describe the subject, action, setting and camera movement in a clear sequence. Use a reference image when visual consistency matters, and change one part of the prompt at a time when refining a result.