Gemini Omni Flash, explained: what Google’s video model does and how to use it from Europe

Gemini Omni Flash is Google's fast video-generation model: it turns a text prompt — with an optional reference image — into short 720p clips, and it can keep editing what it just generated through follow-up instructions. Google released it in public preview in the Gemini API on June 30, 2026. In ilisai it is the model behind the video generator, billed per second in credits.

This guide covers what the preview does as of July 19, 2026: clip lengths and formats, how remixing works, what Google charges for the model through its API, what a second of video costs in ilisai, and the regional restriction that matters if you work from Europe.

What is Gemini Omni Flash?

Gemini Omni Flash (model id gemini-omni-flash-preview) is a video-generation model that Google released in public preview in the Gemini API on June 30, 2026. It generates clips of 3 to 10 seconds at 720p and 24 frames per second, in 16:9 or 9:16, from a text prompt or a prompt plus a reference image, and it can edit videos it has generated when you ask for changes in a follow-up request.

Two design choices define it. It is built for short clips — ten seconds is the ceiling, and pricing is per second, which keeps experimentation cheap. And it is stateful: the model runs on Google's Interactions API, where every generation is an "interaction" the model can pick up again. Ask for "the same scene at night" and it edits the clip it just made instead of starting over.

The -preview suffix in the model id is not decoration. As of July 19, 2026 this is a public preview: Google can change formats, limits and pricing before a stable release.

What the preview can do (July 2026)

Capabilities per Google's model documentation, as of July 2026:

Reference images. You can hand the model an image to animate or to use as a visual anchor for the scene. The input image does not need to match the output dimensions exactly — arbitrary sizes are accepted, which earlier video APIs did not always allow.

Remixing. Every clip the model generates stays editable. A follow-up request references the earlier interaction, so you can iterate — change the lighting, swap the camera angle, adjust the motion — without re-describing the whole scene. Per Google's documentation (July 2026), interactions are retained for up to 55 days by default on paid API usage (the window is configurable shorter); a clip whose interaction has expired can no longer be remixed.

What it does not do. There is no 1080p output and no clip longer than ten seconds in the preview. And one regional limit stands out: users in the EEA, Switzerland and the UK cannot upload their own footage for the model to edit — Google blocks uploaded-video editing in those regions. Editing videos the model itself generated works everywhere. More on this below.

Gemini Omni Flash in ilisai

As of July 19, 2026, Gemini Omni Flash (preview) is the default model in ilisai's video generator. Sign in, open the video page, describe the clip, and pick your settings: a duration preset of 4, 6, 8 or 10 seconds, 16:9 or 9:16, at 720p. You can attach a reference image, and every generated video has a Remix button for follow-up edits. No Google account and no API key — generation runs through ilisai's own integration with the Gemini API.

Pricing is deterministic. A second of 720p video costs 25 credits, and 1 credit is €0.01 (prices as configured on July 19, 2026):

The cost depends only on the duration you request, so you can work it out before you generate. A job that fails without delivering a video is not charged: the reserved credits are released.

For comparison, Google's own API pricing for the model works out to roughly $0.10 per second of 720p video — $17.50 per million video output tokens at 5,792 tokens per second, plus $1.50 per million input tokens (Google's pricing page, July 2026). Credits in ilisai buy the same generation without managing an API key or a billing account.

If you already generate images in ilisai, the flow will feel familiar: prompt, settings, result, iterate. Our Nano Banana Pro prompt guide shows the same iteration logic applied to images.

Using it from Europe: what works and what doesn't

Three facts, straight from Google's regional availability notes (July 2026):

• Generating videos works. Text-to-video and reference-image-to-video are available in the EEA, Switzerland and the UK, same as everywhere else — with one documented carve-out: Google does not support uploading or editing images containing minors (and blocks certain recognizable people) in these regions, so a reference image has to respect that.
• Remixing generated videos works. Editing a clip the model produced is available in Europe without restriction.
• Editing your own uploaded footage does not. Google blocks uploaded-video editing for users in the EEA, Switzerland and the UK. This is a provider-side restriction, not an ilisai choice — and as of July 2026, ilisai does not offer uploaded-video editing anywhere.

ilisai is built for European users first: prices in euros, credit billing with no surprise charges, and honest labeling when a provider restricts a capability regionally — you should see a limit like this in a blog post before you hit it in a product. One thing to know about data: video generation runs on Google's Gemini API, so your prompt and any reference image are processed by Google to produce the clip.

Frequently asked questions

What is Gemini Omni Flash?

A video-generation model from Google, in public preview in the Gemini API since June 30, 2026. It produces 3–10 second clips at 720p and 24 fps from a text prompt, optionally guided by a reference image, and supports remix-style editing of the videos it generates.

How much does a Gemini Omni Flash video cost in ilisai?

25 credits per second at 720p, with 1 credit equal to €0.01 (as of July 19, 2026). An 8-second clip costs 200 credits, or €2.00. The cost is fixed by the duration you request, and a failed job that delivers nothing is not charged.

Do I need a Google account or an API key?

Not in ilisai — sign in and open the video generator. If you prefer Google's own API, the model is available there at roughly $0.10 per second of 720p video (Google's pricing, July 2026).

Can it edit videos I upload?

Not from Europe. Google blocks uploaded-video editing in the EEA, Switzerland and the UK (July 2026), and ilisai does not offer uploaded-video editing anywhere yet. Remixing videos generated in the app works normally, including in Europe.

How long can I keep remixing a clip?

Per Google's documentation (July 2026), interactions are retained for up to 55 days by default on the paid tier, so a clip whose interaction has expired can no longer be remixed. Within that window you can iterate as often as you like — each remix is a new generation, billed by its requested duration like any other.

Is Sora still available in ilisai?

No. OpenAI announced that the API behind Sora video generation shuts down on September 24, 2026, and ilisai's video generator moved to Gemini Omni Flash. Videos you generated with Sora earlier remain in your library.

Will this article stay accurate?

Gemini Omni Flash is a preview, so the details can move. Every fact above carries the date it was checked — July 19, 2026, unless stated otherwise — and we update launch coverage when a model changes materially.

Sources

• Google — Gemini Omni documentation
• Google — Gemini Omni Flash model page
• Google — Interactions API
• Google — Gemini API pricing
• OpenAI — What to know about the Sora discontinuation