2 min read AI-generated

Google I/O 2026: Gemini 3.5 Flash, an Agent Called Spark, and the $100 Subscription Tier

Copy article as Markdown

Google went all-in on AI at I/O 2026. Gemini 3.5 Flash claims frontier-level intelligence at 4x speed, Gemini Spark works around the clock as a personal agent, and the new AI Ultra tier costs $100 per month.

Featured image for "Google I/O 2026: Gemini 3.5 Flash, an Agent Called Spark, and the $100 Subscription Tier"

Google I/O 2026 made one thing clear: everything is AI now. Google unveiled three new Gemini models this week, launched a personal agent, and completely restructured its subscription tiers. Here’s what matters.

Gemini 3.5 Flash: Fast and Smart

The new default model is called Gemini 3.5 Flash. Google claims it delivers frontier intelligence at four times the speed of comparable models. On coding and agentic benchmarks, it reportedly outperforms Gemini 3.1 Pro.

Gemini 3.5 Flash is now the default model in Google Search’s AI Mode and available through the API in Google AI Studio. That’s an aggressive move — Google is pushing its best fast model straight into Search, reaching billions of users.

Gemini Spark: Your Agent That Never Sleeps

The real surprise was Gemini Spark. It’s a personal AI agent running on dedicated Google Cloud VMs — meaning it keeps working even when your device is off.

Spark can check credit card statements for hidden subscriptions, monitor your kids’ school emails, and compile notes into Google Docs. It sounds like what Apple always promised with Siri but never delivered.

Spark launches in beta for AI Ultra subscribers. Which brings us to point three.

New Pricing: AI Ultra at $100 Per Month

Google restructured its subscription model. The new AI Ultra tier costs $100 per month and offers five times higher usage limits than the $20 Pro plan, priority access to Google’s coding tool Antigravity, and 20 TB of cloud storage.

That’s a direct shot at Anthropic Max and OpenAI Pro. The premium AI subscription price war is on.

What Else?

Google also introduced Gemini Omni — a multimodal model that can output video. They showed Android XR smart glasses built with Samsung, Gentle Monster, and Warby Parker. And a fully AI-powered search box that accepts images, videos, and entire Chrome tabs as input.

The I/O was impressively broad. Whether Google can reliably deliver on all of it is another question entirely.


Sources: CNBC, Google Blog, Tom’s Guide