Google I/O 2026 drops Gemini Omni, Gemini 3.5 Flash, and a $100 AI Ultra tier
Google just fired its biggest salvo of 2026. The I/O keynote ran nearly two hours and packed more launches than most companies manage in a full year. Two new models, a personal AI agent, smart glasses, a shopping cart that roams the internet, and a pricing cut that changes the economics of top-tier AI access.
Gemini Omni is the headliner. It takes any input (text, image, audio, video) and generates any output. Available today for paid subscribers. The pitch is simple: one model instead of five. Video editing from a prompt. Image generation that understands spatial relationships. A unified multimodal architecture that doesn’t hand off between separate vision and language systems.
Gemini 3.5 Flash also launched today. It now powers Google Search and the new AI Mode. The model is positioned as the first in a new family combining what Google calls “frontier intelligence with action”. That’s Google-speak for: the model can reason and then do things, not just talk. It replaces the old daily prompt limits with a compute-based model that refreshes every five hours.
The pricing move is the most aggressive signal. Google slashed AI Ultra from $250/month to $100/month. That undercuts OpenAI’s premium tier and makes frontier model access dramatically cheaper. A $150/month discount on a subscription isn’t a sale. It’s a land grab. Google processes 9.7 trillion tokens a month and the Gemini app has 900 million monthly active users, doubled year-over-year. They’re not competing on model quality alone. They’re competing on distribution and price.
Gemini Spark is the agentic play. A 24/7 personal AI agent running on Google Cloud, even when your laptop is closed. It works across Google Workspace, third-party apps, and the web. The live demo showed it planning a block party: pulling RSVPs from Gmail, tracking who’s bringing what, drafting follow-up emails, creating a live tracker in Sheets, generating a Slides deck. Every high-stakes action requires user approval before executing. It launches next week for Ultra subscribers.
Android XR smart glasses from Samsung are confirmed for Fall 2026. Two tiers, iPhone compatible. Google also announced Android Halo, a notification layer in the status bar that surfaces Gemini Spark updates and proactive suggestions.
Other launches: Ask YouTube turns the video platform into a conversational search engine. Universal Cart is an AI-powered shopping experience across the web. Gmail Live, Docs Live, Google Pics, and Daily Brief fill out the product suite. Antigravity 2.0 updates the agentic coding tool.
Demis Hassabis told the crowd that AGI is “just a few years away.” That claim is now standard boilerplate at every major AI event. What’s not boilerplate is the volume of simultaneous product launches. Google isn’t filing the edges anymore. They’re shipping.
The competitive positioning is clear. Against OpenAI: cheaper Ultra tier, Spark vs. Operator, Gemini Omni vs. GPT-5.5’s multimodality. Against Anthropic: distribution muscle that Claude can’t match. Against Apple: actual shipping hardware in XR this fall. The gap between Google’s AI rhetoric and shipped product finally collapsed at I/O 2026.
Sources: Google Blog, Cybernews, Tom’s Guide, BuildFastWithAI