// AI NEWS · AI VIDEO

Google Launches Gemini Omni Flash, a Video Model You Edit by Chatting

The fast tier of the new Gemini Omni family hit public preview on June 30, 2026 — generate a clip, then refine it turn by turn in conversation instead of re-prompting.

2026-07-02 · by Moe Ameen

What happened

Google launched Gemini Omni Flash in public preview on June 30, 2026, releasing it through the Gemini API (model ID gemini-omni-flash-preview) alongside Nano Banana 2 Lite, its fast image model. Omni Flash is the fast, cost-efficient tier of the new Gemini Omni family — a video generation and editing model that accepts text, images, and video as references and outputs a clip.

The headline feature is conversational, stateful editing. Rather than writing one prompt and taking whatever comes back, you generate a clip and then keep talking to it — asking for a different time of day, a camera move, or a wardrobe change — and each turn edits the existing shot while preserving the elements you did not mention. Google frames it as combining Gemini's multimodal reasoning with video, so scene continuity and physics hold together across edits rather than resetting every generation.

At launch, clips are capped at 10 seconds, with longer durations described as coming soon, and output is available in 16:9 or 9:16. Pricing is usage-based at $0.10 per second of video output — the same rate Google charges for Veo 3.1 Fast — so a full clip costs roughly a dollar. Every video carries Google's invisible SynthID watermark for AI provenance. The model is reachable through the Gemini API, Google AI Studio, the Gemini app, and Google Flow. Being a preview, some capabilities are still off — audio references and multi-video referencing are unsupported, character consistency can drift when scenes change, and editing uploaded video is restricted in the EEA, Switzerland, and the UK. Treat specific limits as a launch-window snapshot.

Why it matters for creators

  • Conversational editing changes the workflow: refining a shot by describing the change is faster than re-rolling a prompt, which lowers the effort of getting a usable clip.
  • The price is aggressive — $0.10 per second matches Veo 3.1 Fast, so a 10-second clip is about a dollar, cheap enough to iterate several variations.
  • 9:16 output means clips come out social-native for Reels, TikTok, and Shorts, not just landscape you have to reframe.
  • The 10-second cap keeps it a shot generator for now, not a full video tool — plan on generating openers and B-roll rather than complete pieces.
  • SynthID on every clip bakes in AI provenance, which matters as platforms and audiences push for disclosure of AI-made content.
  • It ships with no publishing layer — no captions, no scheduling, no cross-platform posting — so the model gets you a clip, not a finished post.

How to act on this with Kompozy

Two things are worth acting on the day a model like this lands, and Kompozy covers both. The first is the news itself: a clean, searchable launch with a clear hook (you edit video by chatting) is exactly the kind of moment creators can ride while interest is peaking. Feed the facts into Kompozy as a source and the engine fans one explainer into a package — a Blog Article that can rank for "Gemini Omni Flash," a Carousel that walks through what conversational editing actually is, a Listicle Video of the key specs over a portrait clip, and native Text Posts sized per platform. Autopilot schedules and publishes the set across all nine connected platforms plus your blog from one queue, and generation runs server-side on Trigger.dev workers, so you can approve a batch and let it ship while you move on.

The second is the clip. Generate and refine a 10-second shot in Omni Flash, export it, and drop it into Kompozy to do everything the model doesn't: burn in branded captions, reframe it per platform, stack hook text through HyperFrames, and fan that single clip into supporting posts — a quote card, a thread, a caption set — all in your voice via your Persona Brief. Omni Flash gives you the shot fast; Kompozy turns it into a week of on-brand, scheduled content while the launch is still hot.

Quick takeaways

  • Google launched Gemini Omni Flash in public preview on June 30, 2026, via the Gemini API, alongside Nano Banana 2 Lite.
  • Its defining feature is conversational, stateful editing — refine a generated clip by chatting, and each turn preserves what you did not change.
  • Clips are capped at 10 seconds (longer coming soon), in 16:9 or 9:16, with a SynthID watermark on every output.
  • Pricing is $0.10 per second of output — the same as Veo 3.1 Fast — so a full clip is about a dollar.
  • It has no publishing layer; use Kompozy to caption, reframe, fan out, and schedule the clips across nine platforms.

Frequently asked questions

What is Gemini Omni Flash?

It is a video generation and editing model from Google, launched in public preview on June 30, 2026 as the fast tier of the Gemini Omni family. Its standout feature is conversational editing — you generate a clip and then refine it turn by turn by chatting, with each edit preserving the parts you did not mention.

How long can Gemini Omni Flash clips be, and what do they cost?

Clips are capped at 10 seconds in the launch preview, with longer durations described as coming soon, in 16:9 or 9:16. Pricing is $0.10 per second of video output — the same as Veo 3.1 Fast — so a full 10-second clip costs about a dollar.

Where can I use Gemini Omni Flash?

It is available through the Gemini API, Google AI Studio, the Gemini app, and Google Flow. Note that in the preview, editing uploaded video is restricted in the EEA, Switzerland, and the UK, and audio references and multi-video referencing are not yet supported.

How can creators publish Gemini Omni Flash clips across platforms?

Omni Flash generates the clip but does not publish it. Bring the export into Kompozy to add branded captions, reframe it per platform, and schedule and publish it across TikTok, Reels, YouTube Shorts, X, LinkedIn, and more — then fan the same clip into a quote card, text posts, and a thread in your voice.

Related news

← All AI news · Get started →