Gemini Omni is worth using in 2026, but mainly as a fast, editable video model rather than a pure cinematic generator. Its standout feature is conversational editing: you change a finished clip by describing the edit, and you don't start over. As of November 1, 2026, the version to pick is Gemini Omni 1.1 Flash. It added 360p drafts, 1080p and 4K output, video references and first-and-last-frame control, and at 720p it costs about $0.10 per second on Google's API.
Many Gemini Omni reviews were written in May 2026, right after Google I/O, and they still describe the launch version: 720p only, no end frame and tight daily limits. That no longer holds. This review covers both Omni models (the original Gemini Omni Flash and Gemini Omni 1.1 Flash), what changed in 1.1, the real price per clip, where Omni falls short and how it compares with Veo 3.1.
What is Gemini Omni?
Gemini Omni is Google DeepMind's family of "any input" video models, announced at Google I/O on May 19, 2026. One request can mix text, images, video and audio, and the output is a short video with native sound (speech, sound effects and ambience). Google built it around multi-turn editing. You generate a clip, then ask for changes such as "make it night", "swap the mug for a glass" or "move the camera lower". The model keeps the rest of the scene consistent from one turn to the next.
The first model was Gemini Omni Flash, where "Flash" is Google's name for its fast, lighter tier. On August 27, 2026, Google released Gemini Omni 1.1 Flash (gemini-omni-1.1-flash in the API) as a production-ready update. Every output carries an invisible SynthID watermark.
For a full feature walkthrough, read our explainer on what Gemini Omni Flash is. This post focuses on the verdict: what it does well, what it doesn't, and what it costs.
Gemini Omni review at a glance
| Gemini Omni Flash (1.0) | Gemini Omni 1.1 Flash | |
|---|---|---|
| Released | May 19, 2026 (Google I/O) | August 27, 2026 |
| Inputs | Text, image, video, audio | Text, image, video, audio |
| Native audio | Yes | Yes |
| Clip length | 3–10 s | 3–10 s per generation (Google API: scene extension in 10 s steps to 40 s) |
| Resolutions | 720p | 360p draft, 720p, 1080p, 4K |
| Aspect ratios | 16:9, 9:16 | 16:9, 9:16 |
| Reference images (on Kenerate) | Up to 4 | Up to 4 |
| Reference video clips (on Kenerate) | No | Up to 3 clips, each ≤ 3 s, 9 s total |
| First/last frame | No end frame | Yes |
| Video edit mode | Yes | Yes |
| Google API price (720p) | No longer the stable model | ≈ $0.10 per second |
| Kenerate credits (720p) | 39 credits/second | 30 credits/second |
Specs come from Google's Gemini Omni 1.1 Flash announcement and the Gemini API pricing page. Kenerate limits and credits come from our model configuration, checked on November 1, 2026.
What is Gemini Omni good at?
Is the conversational editing really useful?
Yes. It is the main reason to choose Omni over other video models. With most generators, any change means a new generation and a new roll of the dice. With Omni, the clip you already like is your starting point. SeaArt's hands-on review reached the same conclusion: Omni is strongest "as an interactive video editing model". It handled spatially aware edits well, such as effects that follow a moving hand.
In practice this suits:
- Product and ad iterations: keep the shot and change the background, colour or prop.
- Explainers and education: Google points to Omni's world knowledge and physics, and Google quotes its customer GMI Cloud citing accuracy for educational content.
- Social cut-downs: 9:16 output and short clips fit Reels, Shorts and TikTok.
Does it understand mixed inputs?
That is the point of the "Omni" name. You can send a product photo, a short reference clip and a written direction in one request. On Gemini Omni 1.1 Flash, the video reference (up to three seconds in Google's API) helps keep a character or setting consistent across shots. That has been a weak spot for most prompt-only video models.
How fast is it?
Flash models are tuned for speed. In 1.1, Google says 360p drafts generate up to 60% faster than 720p at about a third of the cost. The intended workflow is to draft several 360p variations, pick the best and re-render that one at 1080p or 4K. Google's own sample app, "Draft Room", does exactly this with 3–4 drafts side by side.
Where does Gemini Omni fall short?
An honest review needs the limits as well:
- Short clips. One generation is 3–10 seconds. Google's API can chain scene extensions up to 40 seconds, but each step is another paid render.
- Only two aspect ratios. It offers 16:9 and 9:16, with no 1:1, 4:5 or 21:9. Square feed posts need a crop.
- Prompt-only cinematic shots vary. Reviewers found that complex transformations, dense multi-object edits and pure text-to-video "hero shots" vary from run to run. SeaArt rated its fidelity ceiling below Veo 3.1 and its prompt-only shots as less cinematic than Seedance 2.0.
- The launch model is weaker. Gemini Omni Flash (1.0) is 720p only and has no end-frame control. If a review complains about "720p max", it is reviewing 1.0.
- Consumer limits. In the Gemini app and Google Flow, Omni needs a Google AI Plus, Pro or Ultra subscription. Early reviewers said they reached the daily limits quickly, even on paid plans.
What changed in Gemini Omni 1.1 Flash?
Google's August 27 post lists five upgrades over the earlier Omni model:
- Scene extension. The model now reads up to 10 seconds of prior context, where earlier models used only the last second. Clips extend in 10-second steps to a cumulative 40 seconds.
- First and last frame. You set the start and end keyframes, and the model fills the motion in between. This is useful for transitions and before/after reveals.
- 360p drafts. These are up to 60% faster than 720p at about a third of the cost.
- 1080p and 4K output. Both are upscaled tiers for final renders.
- Video references. Up to three seconds of video can be used as input for character and scene consistency.
The upgrade closes most of the gaps reviewers listed in May. The original gemini-omni-flash-preview API model was scheduled for retirement on September 30, 2026, so new projects should start on 1.1.
How much does Gemini Omni cost?
Google API pricing
Google's pricing page lists Gemini Omni 1.1 Flash video output at $17.50 per million tokens, billed at 5,792 tokens per second of 720p video. That works out to about $0.10 per second at 720p, and there is no free tier on the API. Using Google's token rates, eesel.ai calculated the other resolutions:
| Resolution | Approx. cost per second (Google API) | 8-second clip |
|---|---|---|
| 360p draft | ~$0.034 | ~$0.27 |
| 720p | ~$0.10 | ~$0.81 |
| 1080p | ~$0.15 | ~$1.22 |
| 4K | ~$0.30 | ~$2.43 |
Price on Kenerate AI
On Kenerate AI, both Omni models run in the same video studio as Seedance, Kling, Veo and Wan, and you pay credits per second:
| Model on Kenerate | 360p | 720p | 1080p | 4K | Edit mode |
|---|---|---|---|---|---|
| Gemini Omni 1.1 Flash | 9 cr/s | 30 cr/s | 45 cr/s | 90 cr/s | 6 cr per input second + 12 / 36 / 54 / 108 cr per output second |
| Gemini Omni Flash | — | 39 cr/s | — | — | 6 cr per input second + 39 cr per output second |
An 8-second 720p clip on Omni 1.1 Flash costs 240 credits. On the original Omni Flash, the same clip costs 312 credits. For comparison, Veo 3.1 at 720p with sound is 120 credits per second on Kenerate, so an 8-second clip costs 960 credits. Video models need a paid plan. The Starter plan is $15 for 1,300 credits, and current plans are on the pricing page.
Is Gemini Omni free?
Only in some places. Google rolled Omni Flash out at no cost in YouTube Shorts (Remix) and the YouTube Create app, with limits it hasn't published. A limited launch promo of about 10 free videos in the Gemini app ended in August 2026. The Gemini app and Google Flow now need a Google AI Plus, Pro or Ultra plan, and the API has no free tier. Watch out for unofficial "Free Gemini Omni" listings in software directories. They are not Google products.
Gemini Omni vs Veo 3.1: which should you use?
Both are Google models, but they do different jobs. The short version: Omni edits and Veo renders.
| Gemini Omni 1.1 Flash | Veo 3.1 | |
|---|---|---|
| Best for | Fast drafts, iterative edits, any-input prompts | Polished 8-second hero shots with tight dialogue |
| Clip length (on Kenerate) | 3–10 s, any whole second | 4, 6 or 8 s |
| Editing | Chat-style video edit mode | Re-generate or extend |
| 720p price on Kenerate | 30 cr/s | 120 cr/s with sound |
Choose Omni when you expect to revise the shot several times or when you need to mix photo, clip and text inputs. Choose Veo 3.1 for a single, final, dialogue-heavy shot. Our Gemini Omni Flash vs Veo 3.1 comparison covers this in more depth. If neither fits, Seedance 2.5 on Kenerate makes clips up to 30 seconds with multi-shot control.
How to try Gemini Omni on Kenerate AI
- Open the video studio. Go to the Gemini Omni video generator or pick "Gemini Omni" in the model list of the Kenerate video studio.
- Choose the version. Pick Gemini Omni 1.1 Flash for 360p–4K and video references, or Gemini Omni Flash for a fixed 720p run.
- Pick a mode. Text-to-video, image-to-video (upload a start frame), reference-to-video (up to 4 images, plus up to 3 clips on 1.1) or video edit.
- Set length and ratio. Choose 3–10 seconds and 16:9 or 9:16. The credit cost updates before you generate.
- Draft cheap, then finish. On 1.1, run one or two 360p drafts (9 credits per second), then render the best prompt at 720p, 1080p or 4K.
- Edit instead of restarting. Use video edit mode to change the background, lighting or a prop in the clip you already like.
Is Gemini Omni worth it? Final verdict
As of November 2026, Gemini Omni 1.1 Flash is one of the most practical Google video models to work with. Edits don't throw the shot away, 360p drafts keep testing cheap, and on Kenerate it costs a quarter of Veo 3.1 per second at 720p. It is not the best choice for a 20-second cinematic sequence or for square formats, and prompt-only hero shots can vary from run to run. If your work involves revisions (ads, product shots, explainers, social clips), Omni is worth testing. If you need one perfect shot, keep Veo 3.1 or Seedance 2.5 in the mix.
