Gemini 3.5 Flash and Omni Redefine Google’s Model Lineup
Google I/O opened with a clear message: Gemini is the backbone of its AI strategy. Gemini 3.5 Flash, long-rumoured after a brief backend leak as “Gemini 3 Fast,” finally got its stage debut as a streamlined, high-speed model aimed at everyday tasks and lighter workloads. Alongside it, Gemini Omni AI took the spotlight as Google’s next-generation multimodal flagship, designed to handle text, images, and video inside a unified architecture. Omni is already surfacing in Google Flow and the Gemini app, where it powers generative graphical answers and more consistent video outputs. While Gemini 3.5 Pro stayed in the background, Google hinted that benchmark slides and a “coming soon” label would follow the company’s familiar rollout pattern. Together, Flash and Omni show Google’s intent to offer both fast-response and highly capable tiers within a cohesive Gemini stack for developers and end users.
A Redesigned Gemini UI and AI Studio on the Go
The Gemini app itself is getting a “complete redesign” that signals how deeply Google plans to weave AI into everyday workflows. The new interface, described as using a Neural Expressive design language and a Liquid Glass aesthetic, brings refreshed typography, colors, and a more dynamic layout. Gemini Live now lets users switch seamlessly between typing and talking, pause the microphone for longer explanations, and receive responses broken into visual modules instead of long walls of text. At the same time, Google AI Studio is expanding beyond the browser with a dedicated mobile companion app, giving developers the option to write and test code directly from their phones. On desktop, the Gemini app is introducing Spark, an agent mode built to work with local folders, connectors, and skills, nudging developers toward an “agent-first” approach to building and debugging AI-driven applications.
Spark, Modern Web Guidance, and New Agent Tools for Developers
For developers, Google I/O delivered a slate of tools that make agents first-class citizens in web and app development. Spark, the new agent mode in the Gemini desktop app, can interact with local files and external connectors, paving the way for semi-autonomous workflows that monitor projects, run tasks, and chain skills together. On the web side, Google introduced Modern Web Guidance in preview, a collection of resources designed specifically for coding agents to build and refine websites. Complementing that are new Chrome DevTools aimed at helping both humans and agents diagnose issues more smoothly. WebMCP, another new offering, lets developers turn their existing web pages into agent toolkits, giving AI agents more autonomy over UI-driven tasks. Together, these updates position Gemini not just as a chatbot but as a development platform, with Spark and these web tools acting as the on-ramps for agentic applications.
Android XR, Intelligent Eyewear, and Gemini Everywhere
Google used the stage to push Android XR software and intelligent eyewear from concept toward reality. Co-developed with partners like Warby Parker and Gentle Monster, the new Android XR glasses are the first consumer-ready “intelligent eyewear” set to arrive later this year, running on Android XR and tightly integrated with Gemini. During the keynote, Google showed AI glasses recognizing what the wearer was seeing, playing entrance music, and handling a coffee order via DoorDash while the phone stayed in the pocket, only asking for final confirmation before purchase. Google also confirmed its first audio-only smart glasses, which deliver private spoken Gemini assistance without a visible display. These devices sit alongside broader Gemini integrations across phones, laptops, and other wearables, underscoring a vision where Gemini is embedded everywhere—from the search bar to XR hardware, and from casual users to power developers building spatial computing experiences.
Gemini for Science and the Road Ahead for Developers
Beyond consumer and developer tools, Google closed I/O by positioning Gemini as a driver of scientific discovery. The new Gemini for Science initiative combines Google’s AI stack to help researchers track fresh papers, summarize research updates, and stay on top of rapidly expanding scientific literature. Google highlighted use cases like building digital twins of the Earth for better climate and weather simulations, and accelerating areas such as drug discovery. For developers, this scientific focus hints at new APIs and datasets geared toward high-impact domains, while existing tools like Google Flow and Pics show how creative agents and multimodal models will trickle into specialized workflows. With on-demand I/O sessions and codelabs rolling out after the keynote, the message is clear: learning to build with Gemini 3.5 Flash, Gemini Omni AI, and Android XR is no longer optional for developers who want to stay aligned with Google’s AI-first ecosystem.