All the AI Products Announced at Google I/O 2025

Google I/O 2025 was a showcase of the company’s relentless push into artificial intelligence, with nearly every major product and service receiving significant AI-driven upgrades or entirely new capabilities. Below is a comprehensive rundown of the most important AI product announcements and features unveiled during the event.

Gemini: The Centerpiece of Google AI

Gemini 2.5 Pro: The latest flagship model, now leading industry benchmarks (notably the LMArena leaderboard), powers a wide range of new features across Google products. Gemini 2.5 Pro is designed for advanced reasoning, coding, and multimodal tasks, and is available in both consumer and developer offerings¹ ⁵.
Gemini 2.5 Flash: A more affordable, lightweight version of Gemini, optimized for everyday tasks and set to roll out in early June⁵.
Gemini 2.5 Pro Deep Think: Tailored for complex mathematical and coding challenges, this variant is currently available to “trusted testers”⁵.
Gemini Live: An agentic AI that combines camera input, voice commands, and web browsing, enabling users to interact naturally and hands-free. It can interpret images, remember contextual details, and perform tasks on the user’s behalf, extending last year’s Project Astra concepts³.
Gemini Agent Mode: Soon to launch in the Gemini app, this feature allows users to delegate research and task execution to an AI agent capable of acting across Google services⁵.
Gemini in Chrome: An AI assistant that answers questions based on the current webpage, rolling out soon in the U.S.⁵.

AI in Google Search

AI Mode for Search: Now available to all U.S. users, this mode enables handling of longer, more complex queries by breaking them into components and synthesizing results. It includes features like “query fan-out” and will soon offer personal context and deep search capabilities, including data visualization tools⁵.
AI Shopping Features: Leveraging the Shopping Graph (50+ billion products), users can describe what they want and see visual options, including virtual try-ons using uploaded photos. An AI shopping assistant tracks availability and price drops, with experimental features live in Google Labs⁵.

Workspace and Gmail AI Enhancements

Personalized Smart Replies in Gmail: Using personal context from across Google services, Gemini generates tailored email responses. This feature is expected to launch for Workspace users this summer⁵.
Inbox Cleanup: New AI-powered tools help users mass-delete emails based on custom criteria, streamlining inbox management².

Android XR and Wearables

Android XR OS: Google’s new operating system for AR/VR devices, designed for both immersive headsets and smart glasses. Announced in partnership with Samsung and Qualcomm, the Project Moohan XR headset and new smart glasses (in collaboration with Gentle Monster and Warby Parker) are set for release later this year. Developers will have early access to build XR apps in 2026⁵.
Smart Glasses with Android XR: Featuring cameras and lenses that connect to smartphones, these glasses signal Google’s renewed commitment to wearable AR⁵.

Generative AI: Images, Video, and Music

Imagen 3: The latest image generation model, offering improved detail and text rendering in visuals, addressing previous limitations of AI art generators⁵.
Veo 3: A new video generation model with better scene physics, sound effects, and dialogue synthesis, available now alongside a video editing tool called Flow⁵.
Lyria 2: Google’s new generative music model, capable of producing more realistic and nuanced compositions5.

Developer-Focused AI Tools

Jules (Coding Agent): Now in public beta, Jules is an asynchronous coding agent that assists developers with codebase management and file modifications⁵.
Native Audio Text-to-Speech: A new model that can replicate voices across languages, enhancing accessibility and localization for developers⁵.
Gemini Canvas: A collaborative AI co-creation platform, enabling teams to build and iterate on ideas together⁵.

AI Infrastructure and Pricing

TPU Ironwood: Google’s seventh-generation AI accelerator, delivering 10x the performance of the previous generation and 42.5 exaflops per pod. This hardware underpins all new AI models and will be available to Google Cloud customers later this year¹ ⁵.
Google AI Ultra Plan: A new subscription tier offering maximum rate limits and access to the full suite of AI products, priced at $249.99/month (with a base plan at $19.99/month)⁵.

Conclusion

Google I/O 2025 underscored the company’s commitment to making AI ubiquitous, practical, and powerful. From the next-gen Gemini models and agentic AI assistants to immersive XR experiences and generative content tools, Google is weaving artificial intelligence into the fabric of daily life, work, and play. The breadth and depth of these announcements signal not only technological leadership but also a vision for an AI-first future across every device and service Google touches¹ ³ ⁵.