Google I/O was a large one with a host of 2025 announcements, which was mostly focused on AI. We got updates and new features for Gemini 2.5 model (Pro and Flash), more powerful imagene 4 image generator, and VO3AI video generator. Google also unveiled the flow – a dedicated AI film manufacturing tool that connects imagene, VO and Gemini models to create cinematic views with simple text signals.
Gemini 2.5 The text brings extended language support in more than 24 languages ​​with text-to-speech and expressive voices. Google claims that arguments, multimodulies, coding and long reference capabilities for both flash and pro models improved.
https://www.youtube.com/watch?v=n3lpxbvzvevo
The larger new update on the Pro model is highly complex mathematics and coding tasks added deep thinking mode. This feature is still in its “experimental” phase and will soon be released to reliable examiners. Google says that it can consider several hypotheses before responding.
Gemini 2.5 Pro is now the leading AI mode on Webdev Arena and LMARENA benchmark. This provides up to 1 million token windows for increased coding and web app building tools and long reference understanding. Gemini 2.5 also improves security against indirect early injections.
Gemini 2.5 flash is available to all users in preview in the Gemini app, while its normal release is coming later in June. It will be followed shortly after the commercial release of Gemini 2.5 Pro.
Imagene 4 Now you can output images until 2K resolution. Google claims that this card, poster and comics have improved text accuracy. Imagene 4 Today is available from Gemini, Google Workpagon, Whisk and Vertex AI apps.
Imagene 4 sample (shared by Google)
VO3 Is Google’s latest AI video model and has a better text-to-video recognition. It can output videos with sound, character dialogue and background noise. Veo 3 is available today for the US and Wartax AI Enterprise users for Google AI Ultra Subscribers.
https://www.youtube.com/watch?v=yqg0ouaofh0
VO2 The camera is achieving movements, object add and removal function. Users can also add images for styling control and outpanning to expand the frame beyond the original borders.
Flow Google’s new AI is a film production tool that adds the capabilities of Veo, Imagen and Lyria models in more detail for cinematic views. Google claims that the flow can help storytellers to create extraordinary cinematic clips that excel in physics and realism. Users can control camera speed, angle and approach. You can also edit and extend the video already generated. Google flow is now available to ultra customers in Google AI Pro and US.
