Google I/O 2025 was large with a series of ads, mostly focused on artificial intelligence. We have received updates and new features for Gemini 2.5 models (Pro and Flash), together with the most powerful images of Imagen 4 images and the VIO 3 AI video generator. Google has also revealed Flow: a cinematographic tool to the dedicated that combines Image models, Veo and Gemini to create cinematographic scenes from simple text instructions.
Gemini 2.5 It brings a support for the language extended to over 24 languages with text voices with discussion and expressive. Google states that reasoning, multimodality, coding and long context features improve both for flash and pro models.
https://www.youtube.com/watch?v=n3lpxbvzveo
The new new update on the Pro model is the method of reasoning to the depth added for highly complex mathematics and coding activities. The function is still in its “experimental” phase and will soon be released to the trusted testers. Google says that it can consider multiple hypotheses before responding.
Gemini 2.5 Pro is now the main mode on the webdev Arena and Lmarena Benchmarks. It offers improved web app coding and construction tools and up to a window of 1 million token for a long understanding of the context. Gemini 2.5 also improves security against indirect prompt injections.
Gemini 2.5 Flash is available in preview of all users in the Gemini app, while its general version will arrive later in June. The commercial release of Gemini 2.5 Pro will follow shortly after.
Imagen 4 It can now produce images with a resolution of up to 2K. Google says that it has improved the accuracy of the text in cards, posters and generated comics. Imagen 4 is available starting today through the Gemini apps, Google Workspace, Whisk and Vertex Ai.
Imagen 4 samples (shared by Google)
Veo 3 It is the latest video model AI of Google and improved functionality recognition of the Text-Video prompt. It can produce videos with audio, character dialogues and background noises. Veo 3 is available starting today for Google Ultra subscribers to the United States and Vertex to Enterprise.
https://www.youtube.com/watch?v=yqg0oouofh0
Veo 2 It is obtaining camera movements, addition of objects and removal features. Users can also add images for style control and exit to extend the frames beyond the original edges.
Flow It is the new film tool AI of Google that combines the skills of the Veo, Imagen and Lyria models for cinematographic scenes with more details. Google says that the flow can help narrators create exceptional cinematographic clips that excel in physics and realism. Users can control the movement, corners and perspectives of the camera. It is also possible to edit and extend the previously generated videos. Google Flow is now available for Google subscribers to the pros and ultra in the United States.
Start a new Thread