Live on avots.ai · all three modes

Gemini Omni Flash

Google's audio native video model, and the first mainstream one that EDITS a ready video by plain text. Tell it "make the pen golden and add falling snow" and the scene changes while you, your motion and your voice stay exactly the same. It also makes new clips with synchronized sound and dialogue, and animates photos into talking videos. All three modes are live on avots.ai right now.

Try it on avots ✦ See pricing
What it does

Three modes, one model

Omni Flash is built on Gemini's world knowledge, so motion, physics and sound stay coherent in every mode.

Edit a ready video

Upload a clip up to 30 seconds and describe the change: new background, new style, new objects. The person, lip sync and the original voice are preserved. This is the Higgsfield style workflow, on your own footage.

Video with real sound

Text to video with synchronized audio and spoken dialogue in the same pass. No separate voiceover step, no silent drafts.

Photos that talk

One photo in, a moving clip out, with ambience or speech generated from your prompt. A single step instead of the portrait, TTS and lip sync chain.

Scene stays coherent

Edits are iterative: ask for one change, then another, and the scene keeps its identity between turns.

Per second pricing

You pay for the seconds you process, about 180 tokens per second on avots. A 10 second edit is roughly 1,800 tokens. That price is for 720p. Since September 2026 avots runs generation 1.1 of the model (launched in August 2026): you can also pick 360p at about a third of the price or 1080p at one and a half times it.

Ready-made styles

In Studio the Video Styles tool ships 8 one-tap looks: retro browser, neon, magazine, comic, warm glow, green pop, doodles and claymation.

Open Video Styles in Studio
One balance

How to use Gemini Omni Flash on avots.ai

No separate account, no extra API key. One avots balance works in every surface.

Web app

Studio → Video Styles for one-tap restyling, or pick Gemini Omni Flash Edit in the chat model picker and attach a video.

Telegram

In @AvotsAIbot: Творчество → 🎨 Video styles, pick a look, send a clip.

MCP

From Claude or Cursor via the avots MCP server: the edit_video tool takes a video URL plus your instruction.

Compare

Gemini Omni Flash vs Veo 3.1, Sora 2 and Seedance 2.0

 Gemini Omni FlashVeo 3.1Sora 2 ProSeedance 2.0
Edits YOUR ready video by textyes, voice preservednonono
Native audio and dialogueyesyesyessilent
Photo to talking videoyes, one stepimage start onlyimage start onlyimage start only
Billingper secondper clipper clipper clip
On avots.ailive nowlive nowlive nowlive now

Different jobs, different models: generate a fresh scene with Veo or Sora, keep character continuity with Seedance, and edit what you already filmed with Omni Flash. On avots they all share one balance.

FAQ

Questions people ask

What is Gemini Omni Flash?

Google's audio native video model: text to video with dialogue, photo to video with speech, and text-instruction editing of ready clips. It runs on Gemini's world knowledge, so motion and physics look right.

Can it really edit my footage?

Yes. Up to 30 seconds per clip on avots. You describe the change, the scene updates, and you, your motion, your lip sync and your voice stay untouched.

How much does it cost?

About 180 tokens per second of processed video. 1,000 tokens cost 0.99 EUR one time, plans start at 4.90 EUR for 3,000 tokens a month. You only pay for what you process.

Where do I start?

Sign up, open Studio, pick Video Styles, upload a clip. Or ask Claude through the avots MCP server to edit a video for you.

Create an account ✦ All models
Start creating