Oops! Something went wrong while submitting the form.

Gemini Omni Review

Your honest AI Tool Reviews: user-friendly, non-technical and unbiased.

Why you can trust me: Testing Methodology & Editorial policy

Lili Marocsik

July 1, 2026

Artlist Video

Featured Tool

Video Generators

AI Video Creation Platform that lets you create videos with integrated generators, editing features, sound effects, and templates all in one place.

Visit

Gemini Omni

Score

Best Tool for :

Tool performance:

Setup/Onboarding:

User Experience:

Prompt coherence:

AI capabilities:

Audio quality:

Lip sync & motion:

Value for money:

Image quality:

Design capabilities:

Pros:

Video-to-video input available

Serene scene is truly beautiful

Great with text background in prompt 1

Cons:

Generally generations are a bit bumpy

The start frame video is very immature (P4)

Struggles with motion of the breakdancing nun

Monthly price starting from: $10

For Who?

Special FeatureVideo-to-Video

Try Gemini Omni Flash

Key features:

AI Video Generation

Multimodal Input/Output

Conversational Video Editing

Digital Avatar Creation

SynthID Watermarking

Tested Model:

Gemini Omni Flash

Last test:

July 1, 2026

Gemini Omni Flash - Exciting new feature: Video-to-Video

How I Made The Ravers Dance Behind Me

Gemini Omni Flash Review: Google Enters the Video Generation Game
So Google is entering the video generation battle. The Gemini Omni Flash review everyone's been waiting for is here, and I tested it so you don't have to get too excited too fast.
The feature that stands out straight away is video-to-video generation. You can upload an existing video as your input, and Gemini Omni Flash uses it as the base for the new generation. That's a first in this space. No other major model has made video-to-video this accessible yet.

Testing Video-to-Video with a Real Clip

I tested it using a Silly Lili video from my online course. The prompt: generate a 90s audience raving in the background, wearing typical 90s outfits, while I'm talking. The first generation was so cheesy I genuinely refused to upload it here. Think early image generation levels of cliche. The outfits looked like a Halloween costume store threw up on a stock photo. Too obviously AI generated to be useful for anything.

That said, I think a lot of that comes down to prompting. I refined it and the results did improve. Not amazing, but noticeably better. New technology takes time to get right and the prompting learning curve is real. I've added that second attempt as video 1 so you can judge for yourself.

Try Gemini Omni Flash

Other Prompts I Tested Gemini Omni Flash

Prompt 1: Robot and woman handshake, multishot scene with text overlay
I was testing multishot capability and text rendering in video. The scene came together reasonably well but the text rendering was off, which is a known weak spot across most video generators right now.

Prompt 2: Breakdancing nun in a public square
I was testing motion accuracy, audio, and SFX. The nun struggles with the footwork and the crowd moves too much in unison. It looks rehearsed rather than spontaneous, which kills the realism.

Prompt 3: Woman in a lacy dress with origami birds
I was testing aesthetic appeal and cinematic motion. This one looked genuinely beautiful. The detail on the dress came through well. The catch: the birds partially fly out of frame instead of staying in the choreographed spiral pattern I prompted.

Prompt 4: Space suit to Hawaiian backdrop transformation
I was testing start and end frame consistency. This is where Gemini Omni Flash really falls short. Kling handles this kind of scene swap better than anyone right now. Gemini is not close on this use case yet.

Try Gemini Omni Flash

Results Gemini Omni Flash (with simple non video-to-video prompts)

The breakdancing nun had some issues. She struggles with the footwork, and the crowd moves too much in sync, which looks unnatural. What did work was a scene with a woman in a detail-rich dress. That one looked genuinely beautiful. The detail on the fabric came through well. The catch: the birds in the scene were partially flying out of frame, which broke the illusion.

The start frame feature is where Gemini Omni Flash really struggles. I gave it a photo of me in a Mars landscape and asked it to swap in a Hawaiian backdrop. Kling handles this kind of image-to-video transformation better than anyone right now. Gemini is nowhere close on that specific use case.

My overall impression of Gemini Omni Flash

The debut of this new video-to-video model is bumpy. There's a lot of hype floating around on social media right now. Of course a new feature like video-to-video is super nice, it opens up a new world for creatives. Right now it's still very new and all new model innovatons have been imperfect, as it is now with Gemini Omni Flash. But I can see how this is going to be huge and if I'm being completely honest, I could have prompted much better too. So overall I think this model is dope solid and I will keep an eye on it and experiment more with it.

Try Gemini Omni Flash

How does Gemini Omni compare to other AI tools?

Read the Video Generators comparison to find out!

Read Comparison

Lili remembers the excitement of discovering the internet at 14 — a true window to the world. The AI boom now feels just as thrilling. Since 2023, she's tested many AI tools, seeing the good, the bad and the ugly (especially at the beginning). Before AI, she worked as a video marketer, crafting YouTube Ads for HelloFresh and Revolut. She believes AI should empower people, leading her to build this site for SMEs. When not exploring AI, she enjoys her 30 plants and modern art.