I/O 2026 Highlights: Gemini AI Updates from Google’s Largest Developer Event

In Focus

• Google unveiled Gemini 3.5, a family of new AI models set to launch next month
• The first model under this family will be Gemini 3.5 Flash
• The tech giant introduced two AI agents, Gemini Spark and Daily Brief

The Google I/O 2026 developer conference closes on May 20 in California. Google is using the annual event to showcase AI products and preview new hardware and software to developers before making them public. The tech company has shared big product updates, with AI at the center stage. From improving Gemini and enabling AI agents, to launching new AI chips, here are the main highlights of the Google I/O 2026 event.

What Gemini Updates Did Google Announce at I/O 2026?

Based on CEO Sundar Pichai’s keynote address, it’s clear that Google will continue to build products around its Gemini AI models. Here are some of the notable Gemini AI updates that dominated Google I/O announcements in 2026:

1. Gemini 3.5 Pro Launches in June 2026

Google unveiled Gemini 3.5, a family of new AI models that will launch next month. The company is already testing the AI model internally.

2. Gemini 3.5 Flash Rolls Out Globally at No Cost

The first model on the Gemini 3.5 family will be Gemini 3.5 Flash, which is four times faster than frontier models and is less costly. Available globally starting May 19, 2026 at no cost, Gemini 3.5 Flash is now the default AI model in the Gemini App and AI Mode in Google Search.

3. Gemini Omni Flash to Supercharge Video Generation

The tech giant introduced Gemini Omni Flash, a new AI model from a family of multimodal models designed to “create anything from anything”. The intuitive AI model uses text, image, video, and audio inputs to generate videos. Omni Flash is accessible to Ultra and AI Plus Pro subscribers in Google Flow, the Gemini app, and YouTube Shorts.

4. Boosting Productivity with AI Agents

Google launched two AI agents during its annual event. Both agents are designed to enhance personal productivity among users and are accessible to subscribers only.

Gemini Spark, an AI Agent that Runs on Google Cloud

This 24/7 AI agent operates on Google Cloud and continues to run even when a laptop is closed or a phone gets locked. According to the tech giant, Gemini Spark will connect to Gmail, Workspace, and Docs upon launch.

Gemini Spark rolls out in beta to AI Ultra subscribers in the U.S. in the coming week. The AI agent can:

• Review credit card statements and identify hidden subscriptions
• Take meeting notes in Google Docs and send them via email
• Track inboxes and identify deadlines

A Gemini Agent that Creates Morning Digests

AI Pro, Plus, and Ultra subscribers get access to a new AI agent called “Daily Brief”. The agentic tool searches Gmail accounts, Tasks, and Calendar and generates a personalized digest highlighting user’s priorities.

What Hardware Updates Featured in Google’s Announcements?

Google introduced new AI chips, TPU 8t and 8i, during the annual event. The company said the 8t processor is designed to support large-scale AI training, supporting higher volumes of raw compute compared to its predecessor. The 8i processor is designed for inference. The two chips offer up to double better performance per watt, allowing the tech giant to distribute AI training across over a million TPUs globally.