Gemini Omni is Google DeepMind’s multimodal creation family that starts with video. The first shipping member is Gemini Omni Flash: you can mix text, images, video, and (initially) voice references, then generate or edit video through multi-turn chat. Consumer surfaces named in Google’s announcement are the Gemini app, Google Flow, and YouTube Shorts; developers call the stable API id gemini-omni-1.1-flash.
What Omni Flash is for
Google’s product blog frames Omni as Gemini’s reasoning plus creation: change lighting, action, characters, camera, or style with natural-language turns while the scene stays coherent. The DeepMind model page repeats the same pillars—conversational editing, physics- and knowledge-grounded generation, and multimodal references—and lists production controls such as scene extension, first/last-frame constraints, short video references, and drafting at 360p as well as higher resolutions.
The Gemini video overview FAQ names the consumer model Gemini Omni 1.1 Flash, says it can turn combinations of text, photos, or video into video (photo references up to five), supports scene extension and chat editing, and states that Omni replaces Veo inside the Gemini app. Generated videos in the Gemini app carry SynthID; Google AI subscription is required, with features varying by tier and geography (18+).
Where you can use it
Per the same blog rollout note:
- Gemini app and Google Flow: rolling out to Google AI Plus, Pro, and Ultra subscribers globally.
- YouTube Shorts and YouTube Create App: rolling out at no cost starting the week of the announcement.
- Developers / enterprise: the blog said API and enterprise rollout would follow in the coming weeks; the Gemini API model page and pricing page now list the paid GA SKU below.
Do not assume free-tier Gemini API access: pricing marks the Free tier as unavailable for this model.
API model id, modalities, and limits
The Gemini Omni Flash model card lists:
| Property | Official value |
|---|---|
| Model code | gemini-omni-1.1-flash (Stable) |
| Input | Text, Image, Video (up to 10s for editing and extension) |
| Output | Video |
| Context window | 1,048,576 tokens |
| Output video | 3s-10s at 360p / 720p / 1080p / 4K, 24 FPS |
| Preview id marked shut down | gemini-omni-flash-preview |
| Latest update on card | August 2026 |
Docs describe conversational refine/edit via the Interactions API, plus video extension, resolution upscaling, and advanced interpolation. Use the stable id for new work.
Preview shutdown date
Google’s deprecations table under “Gemini Omni Flash models” states:
| Model | Release | Shutdown | Replacement |
|---|---|---|---|
gemini-omni-1.1-flash |
August 27, 2026 | No shutdown date announced | — |
gemini-omni-flash-preview |
June 30, 2026 | October 22, 2026 | gemini-omni-1.1-flash |
Migrate any remaining preview traffic to gemini-omni-1.1-flash before that shutdown date. The same page recommends Omni Flash as the replacement for several Veo endpoints.
Gemini API pricing (Omni Flash)
From the Gemini Developer API pricing section “Gemini Omni Flash” (gemini-omni-1.1-flash), paid tier, USD per 1M tokens:
| Free tier | Paid tier | |
|---|---|---|
| Input (text / image / video / audio) | Not available | $1.50 |
| Output text (incl. thinking) | Not available | $9.00 |
| Output video | Not available | $17.50 |
Google notes video billing uses 5,792 tokens per second of 720p output, about $0.10 per second under Standard pricing. Paid-tier content is marked as not used to improve products; Free tier is not offered for this SKU. Pricing page last updated 2026-10-22 UTC when checked.
→ Model rankings: Compare live usage and Arena rows on the AI Rank models board.
Benchmarks and ranking gaps
DeepMind publishes human preference wins for editing / T2V / I2V / reference-to-video on internal and public benches, but does not put a single public scorecard with numeric Elo-style rates in the pages checked here. AI Rank’s Arena and OpenRouter snapshots at write time had no gemini-omni-1.1-flash row—so there is no third-party rank to quote yet. Treat vendor demos as demos, not as AI Rank measurements.
Sources
- Introducing Gemini Omni (Google blog)
- DeepMind: Gemini Omni
- Gemini API model: Omni Flash
- Gemini API pricing
- Deprecations (Omni Flash section)
- Gemini overview: video generation
Compare before you choose: pricing and rankings move—check the current model rankings on AI Rank before you commit.
Or 📱 get the AI Rank app to follow them on your phone.
Prepared with automated editorial assistance and checked against Google blog, DeepMind, Gemini API model/pricing/deprecations pages, and the Gemini video overview on October 24, 2026. We did not generate sample videos for this article.