# Google I/O 2025: Key Enhancements for the Gemini App
During the Google I/O 2025 event, the tech leader revealed a host of thrilling updates for its Gemini app. Key announcements included the debut of **Gemini 2.5 Flash** and **Gemini Live**, now featuring camera and screen sharing options for iOS users. These improvements are poised to transform the user engagement, rendering the app more adaptive and user-friendly.
## Model Enhancements
### Gemini 2.5 Flash
The **Gemini 2.5 Flash** update, first showcased in April 2025, introduces considerable performance upgrades across various metrics, including reasoning, multimodality, coding, and handling long context tasks. Remarkably, it attains this improved performance while utilizing 20-30% fewer tokens, enhancing efficiency. This version is now accessible to all Gemini app users, with a preview version (05-20) also available in Google AI Studio and Vertex AI. Developers and corporate clients can anticipate general availability for production in early June, with **Gemini 2.5 Pro** arriving shortly thereafter.
### Gemini 2.5 Deep Think
Another critical revelation was the launch of **Gemini 2.5 Deep Think**, known for its superior reasoning abilities. This model shines in Mathematics (as demonstrated by its performance in the USAMO 2025), Code (LiveCodeBench v6), and Multimodality (MMMU) benchmarks. The upgraded reasoning mode employs novel research methodologies that enable the model to assess multiple hypotheses before providing a response. Google is focusing on safety assessments and expert input prior to a broad rollout, granting initial access to selected testers via the Gemini API.
### Imagen 4 and Veo 3
Alongside the Gemini advancements, Google unveiled **Imagen 4**, which offers lifelike detail and enhanced outputs for text and typography, in addition to speed improvements. This tool is presently available within the Gemini app. Moreover, **Veo 3**, featuring built-in audio generation capabilities, is also open to users in the Gemini app (US) for those enrolled in Google AI Ultra. This function allows for the generation of sound effects, background sounds, and character dialogues.
## New Gemini Functionalities
### Gemini Live Camera and Screen Sharing
A standout feature introduced at I/O 2025 is the **Gemini Live** camera and screen sharing option, now available for iPhone and iPad users following its successful introduction on Android. The fullscreen Gemini Live interface will present new buttons to facilitate these functions, making it available to all users at no charge. This rollout aims to bolster cooperation and communication among users.
### Future Integrations
In looking ahead, Gemini Live will soon embrace various Google applications and extensions, including Google Maps, Calendar, Tasks, and Keep. Users will be empowered to generate Calendar events and search Maps directly from their interactions, with additional first-party services expected to be incorporated in the future.
### Deep Research
Starting today, the **Deep Research** functionality enables users to merge public data with their private PDFs and images, fostering a thorough understanding that cross-references unique insights with overarching trends. Future integrations with Gmail and Drive are also anticipated.
### Gemini Canvas and Interactive Quizzes
The **Gemini Canvas** feature has been upgraded with a fresh “Create” menu, allowing users to produce web pages, infographics, quizzes, and audio summaries from their written content. Furthermore, interactive quizzes have been launched, enabling users to take part in customized learning experiences. For example, users can request a practice quiz on a specific subject, obtain immediate feedback, and subsequently engage with personalized quizzes based on their results.
### Agent Mode
For Google AI Ultra subscribers, the forthcoming **Agent Mode** will utilize Project Mariner to simplify user objectives. This feature will permit users to express their goals, with Gemini orchestrating the necessary steps to accomplish them. The interface will feature a chat function alongside a browser window, integrating web browsing, research, and intelligent Google app functionalities.
## Conclusion
The updates revealed at Google I/O 2025 represent a significant advancement for the Gemini app, amplifying its features and user experience. With enhancements in model performance, new collaboration tools, and innovative research and learning solutions, Google continues to expand the horizons of AI technology. As these features are implemented, users can look forward to a more cohesive and efficient experience across their devices.
