Google I/O 2026: Gemini 3.5, Search Overhaul, & Android XR

Everything Announced at Google I/O 2026: Gemini 3.5, Search Overhaul, & Android XR

🧠 Foundational Models: Gemini 3.5 Series & Gemini Omni

Google highlighted the staggering scale of its infrastructure, revealing that it now processes quadrillions of tokens monthly across its global products. To fuel this demand, Google DeepMind unveiled its newest flagship models:

1. Gemini 3.5 Flash (The New Default)

This is now the global default engine driving the consumer Gemini app and AI Search. Deployed for hyper-low latency and low operating costs, Gemini 3.5 Flash acts as an elite agentic and coding model. It pushes out responses up to 4x faster than rival frontier models and outperforms previous iterations on complex, multi-step tool-use benchmarks.

2. Gemini Omni (The “Create Anything” World Model)

The absolute showstopper of the keynote was Gemini Omni, a multimodal flagship model designed as a native world simulator. By merging technologies behind Google’s experimental Veo and Genie projects, Omni inherently understands physical forces like gravity, kinetic energy, and fluid dynamics.

Conversational Video Editing: Users can feed the model text, audio, image references, and video clips simultaneously to generate cinematic-quality scenes. Creators can iteratively edit clips using multi-turn natural language dialogue (e.g., “change the camera angle to a wide shot” or “make the sky look rainy”), while the AI maintains absolute character, background, and physics consistency.
SynthID Transparency: To ensure safety, every video track generated or altered using Gemini Omni is embedded with an imperceptible SynthID digital watermark for quick verification in Chrome and Google Search.
Rollout: The lightweight version, Gemini Omni Flash, is rolling out immediately to premium subscribers and inside YouTube Shorts Remix.

🔍 Google Search: Turning Queries Into Living Dashboards

Google Search is undergoing its most radical architectural evolution in 25 years, transforming from a static text index into a dynamic, interactive dashboard ecosystem.

The Intelligent Search Box: The desktop and mobile search bar now physically expands as you type. It leverages machine learning to anticipate human intent, helping users structure complex, multi-part questions rather than relying on fragmented keywords.
Multi-Modal Entrances: Users can now throw text, images, video files, and entire active Chrome tabs directly into the search bar as a single combined input query.
Generative UI Overlays: Powered by Gemini 3.5 Flash, the updated AI Mode doesn’t just summarize text; it uses “Generative UI” to automatically build personalized, real-time mini-dashboards, custom data visualizations, and use-case-specific interactive applets right over your results page based on what you are researching.

🤖 The New Gemini App Experience & “Gemini Spark”

The official Gemini application has crossed massive global adoption milestones, and it is receiving a ground-up physical and infrastructural facelift.

📱 The “Neural Expressive” Redesign

The mobile app debuts a fluid design language featuring organic animations, haptic feedback, and a streamlined pill-shaped prompt box. Furthermore, Gemini Live has been updated to run inline—meaning you can chat with the AI continuously in a small overlay window without losing your place across other active smartphone apps.

⚡ Gemini Spark: Your 24/7 Cloud Agent

Google officially introduced Gemini Spark, an always-on, autonomous digital assistant designed to independently navigate everyday tasks. Running persistently on Google Cloud virtual machines, Spark tracks your digital life via your data:

It can independently scan credit card statements to isolate and flag hidden subscription fee increases.
It automatically parses school message boards to pull together a clean “Daily Brief” of upcoming deadlines for your children.
Universal Cart Integration: By utilizing the new open Model Context Protocol (MCP), Spark can jump out of the Google ecosystem to interact with third-party merchant partners like Instacart or OpenTable to complete real-world purchases and bookings—though it explicitly surfaces an overlay prompt to request human confirmation before executing high-risk transactions or sending final emails.

👓 Hardware: Android XR “Intelligent Eyewear”

In a major hardware reveal, Google showcased its official return to smart glasses via the Android XR platform, built in deep lockstep partnership with Samsung and Qualcomm.

                       ┌──────────────────────────────┐
                       │   Android XR Smart Glasses   │
                       └──────────────┬───────────────┘
                                      ▼
               ┌──────────────────────┴──────────────────────┐
               ▼                                             ▼
       [ Audio Glasses ]                             [ Display Glasses ]
   • Arriving Fall 2026                          • Launching later in roadmap
   • Real-time translation overlay               • In-line HUD navigation data
   • Native Gemini conversation                  • Hands-free context streaming

Google is avoiding bulky, futuristic headsets by prioritizing everyday fashion. It has partnered with luxury eyewear brands Gentle Monster and Warby Parker to design frames that look completely indistinguishable from standard glasses.

Audio Glasses First: Dropping later this fall, the first wave consists of audio-centric glasses that place Gemini directly in your ear. They can handle real-time voice translation (cloning the speaker’s natural tone), snap on-the-go photos, and provide ambient audio tracking.
Cross-Platform Peace: In a surprise move for platform lock-in, Android XR glasses will feature full, native out-of-the-box compatibility with both Android smartphones and the Apple iPhone.

💼 Workspace: “Docs Live” Voice-Powered Productivity

Google Workspace introduced Docs Live, a feature that completely flips the document creation paradigm. Instead of typing out rigid text blocks or clean prompts, users can speak to Gemini in casual, unstructured streams of consciousness. The AI automatically parses the verbal brain-dump, structuring the thoughts into a perfectly formatted, executive-ready Google Doc in real time.

Voice-based editing and document management are scheduled to expand broadly into Gmail and Keep later this summer.

📺 YouTube: Meet “Ask YouTube”

Hunting through long-form videos for a single piece of information is a thing of the past. Ask YouTube allows viewers to issue complex conversational follow-ups regarding the entire YouTube library. The system creates structured text responses and drops interactive time-stamps that jump your player directly to the exact video frame that answers your specific question.

🛠️ Developer Keynote: Antigravity 2.0 & Coding Subagents

For developers, Google expanded Antigravity into a full, agentic software engineering pipeline.

Antigravity 2.0: This major framework upgrade introduces Predictive Auto-Scaling (which pre-provisions server capacity before massive traffic spikes hit) and Zero-Config Observability to easily debug multi-agent pipelines. Engineers can now spin up specialized AI subagents to independently maintain massive codebases, protected by local terminal sandboxing.
The Android Migration Agent: Google previewed an incredible new feature inside Android Studio. The agent autonomously analyzes foreign app files—whether built on iOS, React Native, or basic web frameworks—and automatically migrates and rewrites the entire project state into a native, clean Kotlin Android app in a matter of hours.

📊 The New “Compute-Used” Google AI Subscription Tiers

Because running background agents and generating physics-grounded videos demands astronomical computing power, Google is restructuring its premium tiers around a usage-based compute model. To accommodate power users, Google adjusted its tiers:

Plan Level	Monthly Pricing	Core Features & Usage Allocation
Google AI Pro	$20 / month	Standard premium access; baseline conversational features and 3.5 Flash core.
Google AI Ultra (New Mid)	$100 / month	5x higher usage limits; priority access to Antigravity 2.0 tools; 20TB cloud storage.
Google AI Ultra (Max)	$200 / month (Down from $250)	20x higher usage ceilings; exclusive access to Project Genie (the interactive 3D world builder).