Google I/O 2026 is in full swing and it’s generally about AI, with Gemini Omni being a new model that Google is unveiling during the event. Where Gemini 3.5 Flash is the model that Google positions as an everyday helper, Gemini Omni is a more powerful model that is likely capable of handling more complex tasks.
Think of Gemini Omni as a “no limits” AI model so to speak. Of course, it’s not really without limits. Surely there are some within it somewhere. Having said that, Google specifically says that Gemini Omni will let you “create anything from any input.” People will no doubt put that to the test once it becomes available. Which as one would have it, is rolling out today as Gemini Omni Flash. This will be available to all users globally from the Gemini app. You will need to be subscribed to Gemini Plus, Pro, or Ultra to access it though.
The Gemini Omni AI model will start with video
Google’s goal is to push Gemini Omni out to several different capabilities, but for now the company says it’s starting with video. Being able to create anything from any input is quite a claim, and one that will have to be tested to really grasp how powerful it actually is.
According to Google, you can take Omni and combine images, video, audio, and text as an input and have it generate high-quality video. It uses Gemini’s real-world knowledge to generate the video, built on learning from Gemini Nano Banana that was launched last year. Google also confirms that it’s “easy to edit” the videos once they’re completed. All through the same conversation you had with Gemini Omni from the beginning.
It’s also possible to take a video you recorded and then have Omni edit it. You can ask it to change what’s happening. If things aren’t quite where you want them to be in the beginning, you can further refine the output across multiple turns in the conversation.
It’ll also be possible to use your own voice to create video using what Google is calling Avatars. This makes a digital version of you in an Avatar form, and then generates videos that look and sound like you. Naturally, this could have some privacy implications that will probably cause some concern. So Google made sure to implement SynthID watermarks in the videos that are generated. This way people can tell fairly easily that videos were AI generated.
The post Gemini Omni is the no limits AI model that can create anything appeared first on Android Headlines.