Google's annual I/O developer conference has long been a showcase for the company's latest innovations, and the 2026 edition was no exception. The keynote, led by CEO Sundar Pichai, was heavily focused on artificial intelligence, with new models, tools, and integrations spanning virtually every major product. From the next generation of its Gemini AI family to expanded smart glasses partnerships and a universal shopping cart, Google laid out a comprehensive vision for an AI-powered ecosystem. Below is a detailed breakdown of the 13 biggest announcements from the event.

Gemini 3.5

The flagship announcement was the launch of Gemini 3.5 Flash, a new AI model that becomes the default for the Gemini app and AI Mode in Search. Google promised significant improvements in speed, agentic task handling, and coding capabilities. The model also features enhanced guardrails to reduce harmful outputs while avoiding false positives on safe queries. A larger variant, Gemini 3.5 Pro, is slated to follow next month. This update continues Google's rapid iteration cycle, competing directly with OpenAI's GPT series and Anthropic's Claude models. The focus on agentic capabilities—allowing the AI to perform multi-step tasks autonomously—signals a shift toward more proactive assistants.

Gemini Omni

Alongside the 3.5 updates, Google introduced an entirely new model family called Gemini Omni. The first model, Omni Flash, launches today in the Gemini app, Google Flow, and YouTube Shorts. Unlike Google's existing Veo video generation model, which only accepts text prompts, Omni Flash can generate video clips from a mix of text, photos, video, and audio inputs. Google described this as a step toward a universal content creation tool that can "create anything from any input." The move positions Google to compete with multimodal models from Meta and others, aiming to simplify creative workflows for consumers and professionals alike.

Gemini Spark: An Always-On AI Agent

Taking the concept of persistent AI assistance further, Google announced Gemini Spark, an always-on agent that runs continuously on virtual machines in Google Cloud. Spark can handle tasks like drafting emails, creating study guides, monitoring for hidden fees, and more. It integrates with Google Workspace (Docs, Gmail, Sheets, Slides) and third-party apps like Canva and Instacart. Plans also include macOS integration to access local files through the Gemini app. This marks Google's answer to OpenAI's rumored "agent" projects and Microsoft's Copilot extensions, bringing true background automation to everyday users.

Vibe-Coding Full Android Apps in AI Studio

Google AI Studio now allows users to build complete native Android applications using natural language prompts. The feature includes an embedded Android emulator for real-time previews, and apps can be tested directly on a connected phone. Developers can export projects to Android Studio or GitHub, or as ZIP files. Google also promised future support for publishing apps exclusively to friends and family, as well as Firebase integration. This democratization of app development could lower barriers for non-programmers, though concerns about quality control and security remain.

Project Aura Smart Glasses Update

Google's smart glasses initiative, Project Aura, received a significant update. Developed in collaboration with Xreal, the glasses now feature a redesigned external compute puck with a fingerprint sensor and lanyard for portability. New software capabilities include widgets for display glasses, Gemini integration with Google Calendar and Keep, and improved overall performance. While still a prototype, Project Aura represents Google's continued investment in augmented reality, competing with Meta's Ray-Ban Stories and Apple's rumored AR headset.

Android XR Glasses from Warby Parker and Gentle Monster

Expanding its XR portfolio, Google announced two new partnerships for audio-only smart glasses launching this fall. Warby Parker and Gentle Monster will each release Android XR glasses that support live translation, navigation assistance via Gemini, and notification summaries. These are akin to the Ray-Ban Meta smart glasses but built within the Android XR ecosystem, giving users a fashion-forward option with Google's AI baked in. The partnerships highlight Google's strategy of leveraging established eyewear brands to reach mainstream consumers.

Universal Cart for Cross-Merchant Shopping

Google unveiled a "Universal Cart" that aggregates products from YouTube, Search, Gemini, and Gmail into a single checkout experience. The cart works with major retailers like Nike, Target, Walmart, Ulta Beauty, Sephora, Wayfair, and Shopify. It can also flag potential issues—such as incompatible parts for a gaming PC—and apply loyalty perks from Google Wallet. The feature rolls out in Search and Gemini this summer, with YouTube and Gmail integration later. This directly challenges Amazon's one-click shopping and PayPal's checkout solutions by leveraging Google's ubiquitous presence across web services.

Gmail Live: Voice-Driven Inbox Search

Gmail is gaining a voice-activated search tool called Gmail Live. Users can click an icon in the search bar to ask questions verbally, and the AI extracts relevant information from emails rather than just returning a list of threads. For example, finding a hotel confirmation code becomes a simple spoken query. Similar voice features are coming to Google Docs and Keep, with data pulled from Drive and Gmail. This represents a natural progression of the Gemini Live experience, making information retrieval more conversational.

Pics: AI-Powered Image Editing in Workspace

A new Workspace app called Pics enables iterative AI-generated image edits. Users click on a specific part of an image and leave a comment describing the desired change, rather than writing a full prompt. Powered by Nano Banana 2 and Gemini, Pics aims to make photo editing more intuitive. Google plans to incorporate these capabilities into other Workspace apps, potentially challenging Adobe's Firefly tools by embedding AI editing directly into productivity software.

Search Evolves with Agents, Generative UI, and Mini Apps

Google redesigned the search box to accommodate longer queries and added AI-generated suggestions. Search now accepts text, images, files, videos, and even Chrome tabs as input. More significantly, Google introduced "information agents" that provide summarized updates on topics by pulling from blogs, news, and social media. These agents launch this summer for AI Pro and Ultra subscribers. Additionally, "generative UI" can produce interactive tables, graphs, and simulations, while "mini apps" create custom dashboards for repeated tasks like event planning. This transformation positions Search as a full-fledged assistant rather than a simple link directory.

AI Ultra Subscription Price Cut

Google slashed the price of its premium AI Ultra subscription from $249.99 per month to a starting tier of $100 per month, with a $200 tier that adds Project Genie access. The move mirrors OpenAI's pricing structure and aims to attract more power users. The cheaper tier still includes advanced models and higher usage limits, while the top tier provides exclusive early access to experimental tools. This aggressive pricing could accelerate adoption of Google's AI ecosystem among developers and enterprises.

AI Detection Tools Expanded to Chrome and Search

Addressing concerns about AI-generated content, Google expanded its SynthID watermarking and C2PA Content Credentials to Chrome and Search. Users can now upload or select images in Search (including via Lens and Circle to Search) to view provenance details. A future Chrome feature will allow users to circle suspicious images on websites to check their origins. This initiative aims to increase transparency and trust, though its effectiveness depends on widespread adoption by content creators and platforms.

Lifelike AI Agents for Google Beam

Formerly Project Starline, Google Beam now incorporates lifelike AI agents that can hold conversations during calls. A demo featured "Sophie," an agent that could answer questions, read documents held up to the camera, and provide recommendations. While not rendered in 3D, the agent uses the underlying Beam hardware (the $25,000 HP Dimension system) to create a more personal interaction. Google also demonstrated group calls via Beam with integration into Google Meet and Zoom, signaling a future where AI companions are a native part of video conferencing.

Source: The Verge News