Gemini is Getting an Agent Mode, Personalization, and Chrome Integration

At the annual developer conference, I/O 2025, Google announced exciting new features for Gemini including an Agent mode and some personalization features. First, the Agent mode in Gemini can live browse the web to perform actions for you. For instance, if you are looking for an apartment, the Agent mode can find the listings that match your criteria, and schedule tours for you.

The Agent mode in Gemini uses MCP to access web listings on supported websites. Initially, it will be available to those who have subscribed to the new Google AI Ultra plan which costs $249.99 per month. Currently, this plan is only available in the US and Google is offering 50% off for the first three months to new users.

Apart from that, Gemini Live is now available to all users on Android and iOS. It’s becoming more personal and proactive. While interacting with Gemini Live, you can now ask it to create an event in your Google Calendar, get the latest details from Google Maps, and more. Google is starting to integrate more Google apps such as Tasks, Keep, Calendar, and Maps into the Gemini Live experience.

Moreover, Google’s latest image generation model, Imagen 4 and video generation model, Veo 3, are now integrated into the Gemini app. Along with video, the new Veo 3 model can natively generate audio including sound effects, background noises, and dialogue between characters. Note that Veo 3 is currently available in the US to Google AI Ultra subscribers.

Google says the new Gemini 2.5 Flash model is becoming the default model on the Gemini app. It’s smaller and cheaper, yet manages to come within the striking distance of the flagship Gemini 2.5 Pro model. Next, you can now upload files, images, and PDFs while using the Deep Research AI agent in Gemini to generate deeper insights.