Gemini Omni Flash is Google's fast AI video generator and the first model in the Gemini Omni family, announced at Google I/O on May 19, 2026. It takes any mix of text, images, audio and video as input and produces a short video with sound, then lets you edit that video by chatting with it. Each generation runs 3 to 10 seconds, and the current release, Gemini Omni 1.1 Flash, adds scene extension up to 40 seconds, first/last frame control and output up to 4K through the API.

Gemini Omni Flash at a glance

Gemini Omni Flash
MakerGoogle DeepMind
AnnouncedGoogle I/O, May 19, 2026
FamilyGemini Omni (Flash is the first and fastest tier)
InputsText, images, audio and video
OutputVideo with native audio (speech, sound effects, ambience)
Clip length3 to 10 seconds per generation; extension up to 40 seconds on 1.1
Resolution (API)360p draft, 720p default, 1080p and 4K on 1.1
Aspect ratios16:9 (default) and 9:16
Frame rate24 fps
EditingMulti-turn conversational editing
WatermarkInvisible SynthID on every video, plus C2PA Content Credentials
API model IDsgemini-omni-1.1-flash (stable), gemini-omni-flash-preview (preview)

Specs come from the Gemini API model page and the Gemini Omni overview on Google DeepMind. Consumer apps may expose fewer options than the API.

A glowing translucent sphere absorbing streams of light shaped like photographs, sound waves and film strips, and emitting a single beam of cinematic footage, dark studio background
A glowing translucent sphere absorbing streams of light shaped like photographs, sound waves and film strips, and emitting a single beam of cinematic footage, dark studio background

What makes Gemini Omni Flash different

One model for every input

Most video generators accept a prompt and maybe one image. Google built Gemini Omni to be natively multimodal: text, images, audio and video are processed together, which Google says produces more cohesive and controllable results. In practice, you can hand it a product photo, a short video reference and a written direction in one request.

Editing by conversation

The headline feature is chat-based editing. You generate a clip, then ask for a change, such as a new background, a different camera angle or an added object, and the model applies it while keeping the rest of the scene. Google DeepMind describes each edit as building on the one before, so the scene stays consistent across turns. In the Gemini app, every turn can also include new image or video references.

World knowledge in the frame

Because it comes from the Gemini line, Omni Flash draws on an understanding of physics plus knowledge of history, science and culture. The aim is shorter prompts that still land: you describe what you want, and the model works out plausible details like how water splashes or how a period setting should look.

Built for speed

"Flash" is Google's label for its faster, lighter tier. Google positions Omni Flash for quick, high-quality generation and iteration, and the 1.1 release adds a 360p draft mode that Google says renders faster than the 720p default, so you can test ideas before committing to a full render.

What Gemini Omni Flash can do

  • Text to video: describe a scene and get a clip with matching sound.
  • Image to video: animate a still, or use images as subject references for characters and objects. The Gemini app accepts up to five images per prompt.
  • First and last frame: give two keyframes and the model generates the motion between them, useful for transitions and seamless loops (added in 1.1).
  • Video editing: upload a clip of up to 10 seconds, or edit one you just generated, using plain-language instructions.
  • Restyling: keep the motion of a clip and change its look, for example into claymation or watercolor.
  • Scene extension: continue a clip in 10-second steps up to 40 seconds, with the model reading up to 10 seconds of prior context (added in 1.1).
  • Upscaling: render at 1080p or 4K for final delivery (API, 1.1).
  • Avatars in the Gemini app: create a digital version of yourself and use it in videos, according to Google's launch post.

Try Gemini Omni Flash in the Video Studio

From preview to 1.1: a short timeline

DateWhat happened
May 19, 2026Gemini Omni announced at Google I/O; Omni Flash rolls out in the Gemini app for Google AI Plus, Pro and Ultra subscribers, plus YouTube Shorts and the YouTube Create app
June 30, 2026Developer preview (gemini-omni-flash-preview) in the Gemini API, Google AI Studio and the Gemini Enterprise Agent Platform
August 27, 2026Gemini Omni 1.1 Flash: scene extension to 40 seconds, first/last frames, 360p drafts, 1080p and 4K output, short video references
September 23, 2026Gemini Omni 1.1 Flash arrives in Google Vids for Google and Workspace accounts, with 1080p generation

Google's Gemini API video page now presents Omni Flash as the default model for video generation, keeping Veo 3.1 for specific use cases.

A long glowing timeline ribbon floating in a dark space, with four illuminated nodes along it, each node holding a small hovering film frame, cinematic blue and amber light
A long glowing timeline ribbon floating in a dark space, with four illuminated nodes along it, each node holding a small hovering film frame, cinematic blue and amber light

Limits and known weaknesses

Omni Flash is fast and flexible, but it is not a long-form or pixel-perfect tool. Google is upfront about several limits:

  • Short clips. Each generation tops out at 10 seconds. Longer pieces need extension or editing together several shots.
  • Consistency across edits. Google's model card lists keeping things consistent through edits as a remaining challenge, along with complex motion scenes.
  • Text on screen. Accurate text rendering is also on the model card's list of hard cases. Keep on-screen words short and check them.
  • Audio control. Speech-changing is restricted for now, and the API does not support voice editing or audio reference uploads.
  • Regional limits. Editing or extending uploaded videos is unavailable in the EEA, Switzerland and the UK, and the Gemini app help page adds some US states for video-to-video editing.
  • Language. The API docs say English is fully supported and other languages have not been evaluated.
  • Age and access. In the Gemini app, video creation is for adults 18+ with a Google AI plan or a qualifying Workspace license.
  • Real people. Certain recognizable people are not supported, and Google applies safety filters to prompts and outputs.

Gemini Omni Flash vs Veo 3.1 and other video models

Gemini Omni Flash and Veo 3.1 both come from Google and both generate native audio, but they are built for different jobs. Omni Flash is a general-purpose, chat-editable model that accepts any input type; Google's API docs keep Veo 3.1 available for specific workflows such as its own extension, frame-specific generation and image-based direction features.

NeedStrong fit
Fast drafts and iterative, chat-based editsGemini Omni Flash
Google's dedicated cinematic video model with native soundVeo 3.1
Multi-shot sequences and strong motionKling or Seedance 2.5
Your own 1080p clips up to 15 seconds from text or imageKen 1, Kenerate's own video model

The honest answer is that no single model wins every shot. Running the same prompt through two or three models and keeping the best take is often quicker than fighting one model's weaknesses.

Who Gemini Omni Flash is for

  • Social creators who need short vertical clips with sound and quick revisions.
  • Marketers who want product shots, quick restyles and variations without a reshoot. Google's own developer demos include an e-commerce product studio.
  • Storyboard and concept artists who want rough motion fast, especially with 360p drafts.
  • Developers building apps on the Interactions API, where edits chain from one interaction to the next.

If you need long uninterrupted takes, exact typography or precise voice control, plan to combine Omni Flash with other tools.

A creator at a dark desk reviewing a vertical short-form video of a sneaker splashing through a puddle on a phone mounted on a small tripod, soft rim lighting, photographic
A creator at a dark desk reviewing a vertical short-form video of a sneaker splashing through a puddle on a phone mounted on a small tripod, soft rim lighting, photographic

Where to try Gemini Omni Flash

  • Gemini app and Google Flow: for Google AI Plus, Pro and Ultra subscribers.
  • YouTube Shorts and YouTube Create: included for creators at no cost.
  • Google Vids: Gemini Omni 1.1 Flash for Google and Workspace accounts.
  • Google AI Studio and the Gemini API: for developers.
  • Kenerate: Gemini Omni Flash is one of the models in the Kenerate Video Studio, alongside Veo 3.1, Kling v3 Pro, Seedance 2.5, Sora 2 Pro, Wan 3.0 and Ken 1. You can then upscale the result, add an ElevenLabs voiceover in the Voice Studio, or score it with Ken Music. New accounts get 25 free credits, and credits never expire.

For step-by-step access, see how to use Gemini Omni Flash in the app, API and Kenerate, and for better results read the Gemini Omni Flash prompt guide.

Bottom line

Gemini Omni Flash is Google's fast, any-input video model: short clips with sound, generated from text, images, audio or video, and refined through conversation. The 1.1 release fills in the obvious gaps with scene extension, keyframe control and 4K output in the API. It is a strong default for quick social and product video, with clear limits on clip length, text accuracy and audio control that are worth planning around.