Google showed off a lot of cool stuff during its Google I/O '25 Keynote. We’ve rounded up the main news.
- Gemini Live: Now sees through your camera and can talk to you about objects you tap. Available today on mobile, FREE;
- Google Meet is bringing a voice translator, for now only for the English and Spanish language pair, with more promised in the «coming weeks»;
- The Gemini app will feature Agent Mode: a basic AI agent that can perform simple tasks such as searching and booking hotels;
- Jules, the coding agent, is now in public beta. It’s literally OpenAI’s Codex, but powered by Gemini;
- The Gemini 2.5 Flash model has been updated: it is now not much inferior to the Pro model, but works much faster. It is already available in AI Studio for free;
- Gemini 2.5 Pro will feature Deep Think, which gives the model more time to think and improves its response. It’s literally like OpenAI’s high models;
- AI Mode will appear in Search: Gemini 2.5 Pro will collect info and give you a short answer (do you have AI with search? Keep searching with AI). Starting today, it works in the US;
- AI search will adapt to your preferences. For example, if you book seats at restaurants on the terrace, it will suggest such places for reservation;
- You can upload your photo and try on the clothes you are going to buy;
- New image generator Imagen 4. Works better with text, I wonder if it will support Ukrainian;
- Veo 3 is out TODAY. The East generates video WITH SOUND;
- Augmented reality glasses with auto-translation function. During the demo, everything fell apart.
