Google enters the agentic era
Google CEO Sundar Pichai opened the company's I/O 2026 conference by declaring the dawn of the agentic era. Google processes 3.2 quadrillion tokens per month, a sevenfold increase from last year. More than 8.5 million developers build with Google models monthly, and Google's model APIs handle about 19 billion tokens per minute.
The centerpiece of the event was Gemini 3.5 Flash, a model that outperforms its predecessor Gemini 3.1 Pro on key benchmarks while running four times faster than other frontier models. It costs less than half the price of comparable models. Google said if top enterprises shift 80 percent of workloads to Flash, they could save more than $1 billion annually.
Antigravity 2.0 powers autonomous agents
Google launched Antigravity 2.0, a standalone desktop application for managing cohorts of autonomous AI agents. It runs a version of Flash optimized to be 12 times faster than other frontier models. The platform is available starting today.
Gemini Spark is a personal AI agent that runs 24/7 on Google Cloud VMs. It executes long-horizon background tasks, connects via the Model Context Protocol, and integrates with the Gemini app. A US beta begins next week for Google AI Ultra subscribers. The agent will expand to email, chat, and Chrome later this summer.
New features across Google products
Google also announced new AI features across its product lineup. Ask YouTube reimagines video discovery by surfacing relevant video segments based on nuanced interests. Voice-Powered Docs allows users to create and edit documents entirely through voice commands.
TPU 8th generation chips were unveiled, including the TPU 8t for large-scale pretraining with three times the raw computing power of its predecessor, and the TPU 8i for high-speed inference. Both deliver up to two times better performance-per-watt. Google's capital expenditure has grown from $31 billion in 2022 to roughly $190 billion this year, mostly focused on custom silicon.