Google has updated two generative models this round, one for video and one for pictures.
The one in the video is called Gemini Omni 1.2 Flash and is officially available today. The duration can be adjusted from up to 10 seconds to 15 seconds, and you can choose between 3 and 15 seconds in whole seconds.
The price is down 30% from the Omni 1.1, at about $0.07 per second for 720p. That works out to a little over a dollar for a 15-second 720p clip.
The key feature is called conversational editing. If you are not satisfied with the film after it is generated, you don’t have to start over, just continue chatting: replace the products in the picture, adjust the lighting, change the aspect ratio, the entire 15 seconds can be changed. People who make e-commerce short videos should be most aware of this. In the past, if you changed a detail, you would have to redraw the entire section. Getting the version you are satisfied with depends entirely on luck.
Drafts can be produced in 360p or 720p first, and then enlarged to 1080p or 4K when finalized. Trial and error in low-definition first, and then release the film in high-definition. The way to save money is paved for you. The output is watermarked with SynthID. Currently on Gemini API and Google Cloud.
The one in the picture is Nano Banana 2 Lite, which is designed for speed. The Vincent picture has a 4-second delay and costs $0.034 per picture with 1K resolution. It is already available on AI Studio, Gemini API and Enterprise Edition Agent platform, and will be included in the searchable AI Mode and Gemini App later.
Both are lightweight files such as Flash and Lite. Google did not shout slogans on image quality this time, but only on duration, speed, and price. At this stage of video generation, everyone is already competing to see who is more suitable for batch running.