What is Google Antigravity? Everything You Need to Know After Google I/O 2026

8 min read
  • google antigravity
  • what is google antigravity
  • antigravity 2.0
  • google io 2026
  • ai coding tools 2026
  • gemini cli replacement

Something big came out of Google I/O 2026 two days ago, and most people outside the developer community are still catching up on what it actually is.

Google announced Antigravity 2.0 — a major expansion of a platform that launched quietly in November 2025 and has since become one of the most ambitious things Google has built in the AI coding space. It is not just a coding assistant. Google is explicitly positioning it as a platform for building and managing teams of autonomous AI agents.

Here is what it is, what is new, and why it is worth paying attention to.

What Is Google Antigravity?

Google Antigravity is Google's agent-first development platform. The official description from Google is straightforward: "a platform for developers to take an idea and turn it into a production-ready app."

The original version launched on November 18, 2025, alongside Gemini 3. At that point it was primarily an agentic coding environment — an AI-powered IDE that could handle complex, multi-step development tasks autonomously rather than just suggesting code completions.

Version 2.0, announced at Google I/O on May 19, 2026, is a different kind of announcement. Google is not just updating a coding tool. It is expanding Antigravity into a full platform ecosystem — with a standalone desktop app, a command-line interface, a developer SDK, cloud-managed agents, and enterprise infrastructure. The coding environment is now just one surface on top of a much larger agent harness.

As Koray Kavukcuoglu, CTO of Google DeepMind, put it during the I/O press briefing: "We are expanding Antigravity beyond a coding environment and turning it into a platform to develop and manage teams of autonomous AI agents."

What Is New in Antigravity 2.0

The 2.0 announcement introduced five distinct pieces that together form what Google is calling the Antigravity ecosystem.

Antigravity 2.0 Desktop App

A standalone desktop application built entirely around agent orchestration — separate from the existing Antigravity IDE. Think of it as a command centre rather than a code editor. You can run multiple agents simultaneously, orchestrate parallel tasks, design custom subagent workflows, and schedule tasks that run automatically in the background without you prompting them each time.

That last feature — scheduled background tasks — is practically significant. It converts the agent from a single-turn tool you prompt manually into something closer to a persistent automation pipeline. You define the task once, and the agent runs it on schedule. Voice command support is also built in, consistent with Google's recent additions to Gmail and Docs.

Antigravity CLI

A command-line version of the platform for developers who prefer terminal-based workflows. Built in Go for speed, it supports asynchronous workflows, can control multiple agents simultaneously, and shares the same underlying agent harness as the desktop app — meaning improvements to core agents apply across both surfaces automatically.

This is also the official replacement for Gemini CLI, which is being retired. Gemini CLI launched less than a year ago and garnered over 100,000 GitHub stars and millions of users. Google's reasoning for retiring it: the tool was built for single-model terminal access, and developers increasingly need multiple agents that can communicate with each other and divide workloads. Antigravity CLI is that upgrade. The migration window closes June 18, 2026, after which Gemini CLI will stop processing requests.

Antigravity SDK

A programmatic interface that gives developers access to the same agent harness powering Google's internal products. The SDK is co-optimised for Gemini 3.5 Flash — Google's new model, released at I/O alongside Antigravity 2.0, which outperforms Gemini 3.1 Pro on almost all benchmarks while running four times faster than competing frontier models. Gemini 3.5 Flash was co-developed alongside Antigravity 2.0 as the platform's primary workhorse.

Managed Agents in the Gemini API

A single API call can now spin up a full agent that reasons, uses tools, and executes code inside a persistent, isolated Linux environment. State and files persist across follow-up calls — enabling multi-turn sessions without re-initialisation. This is the cloud infrastructure layer that lets enterprise teams deploy Antigravity-powered agents at scale without managing the underlying infrastructure themselves.

Antigravity Enterprise Agent Platform

Enterprise support through Google Cloud, with custom agent templates in AI Studio for organisations to get started. Google Cloud customers can connect to Antigravity to build and deploy projects. There is also an Antigravity export tool in AI Studio that lets developers export existing projects and continue work locally.

CodeMender — The Security Agent That Patches Its Own Bugs

Alongside the Antigravity expansion, Google announced CodeMender — an AI code security agent developed by Google DeepMind.

Most security tools find vulnerabilities. CodeMender goes further: it autonomously identifies vulnerabilities, recommends precise fixes, tests those fixes in a secure environment, and applies patches across dependent systems — with developer approval at each step.

The full process as described by Thomas Kurian, CEO of Google Cloud: "CodeMender is an AI code security agent. Leveraging Agent Platform capabilities and advanced Gemini models, CodeMender autonomously identifies vulnerabilities within your code. It then recommends precise fixes, securely tests them, and can apply patches and necessary changes across dependent systems, with your approval."

The key phrase is "with your approval" — Google is being deliberate about keeping humans in the loop at each stage. A select group of security experts is currently testing a CodeMender API, with broader availability to follow.

WebMCP — The Open Web Standard You Should Know About

One of the less-covered but significant announcements from I/O was WebMCP — a proposed open web standard that lets developers expose structured tools (JavaScript functions, HTML forms, and similar elements) so browser-based AI agents can execute complex tasks with greater reliability and precision.

The experimental WebMCP origin trial starts in Chrome 149, with support for Gemini in Chrome coming soon. The idea is to give AI agents working inside browsers a standardised, structured way to interact with web content — rather than scraping and interpreting unstructured HTML.

It is early-stage, but the direction is meaningful: Google is building web infrastructure for a world where AI agents are routine users of the web, not just humans.

How Antigravity Compares to Cursor and Kiro

The AI coding and agent tool landscape is consolidating fast. Here is where Antigravity sits:

Cursor remains the leading choice for individual developers who want AI to accelerate their existing coding workflow. Fast, familiar, IDE-centric. Best for rapid iteration on known codebases.

Kiro (Amazon's spec-driven IDE) plans before it codes — generating requirements documents and design files before writing anything. Best for complex features where design clarity matters more than speed.

Google Antigravity is playing a different game entirely. It is not primarily positioning itself as a tool for individual developers writing code — it is building infrastructure for teams of autonomous agents working in parallel. The desktop app, CLI, SDK, and managed cloud agents are all aimed at developers who want to orchestrate multi-agent workflows, not just accelerate single-developer coding sessions.

The honest comparison: Cursor is a co-pilot for your code. Kiro is a planner that happens to write code. Antigravity is a platform for deploying fleets of agents that build and maintain software with minimal human intervention.

Pricing

Google announced an AI Ultra plan at $100 per month alongside the Antigravity 2.0 launch, offering five times more usage than the existing AI Pro plan. Full pricing details for the enterprise tier have not been fully disclosed. The Antigravity CLI is available to existing Google AI Pro and Ultra subscribers, with the Gemini CLI migration deadline set for June 18, 2026.

Why This Matters

Google I/O 2026 had a clear throughline in its developer announcements: the company has fully committed to a world where AI agents, not humans, do the orchestration. The framing from the I/O team was explicit — Google has "transitioned from AI that simply assists you, to agents that can independently navigate complex tasks across your entire workflow."

Antigravity 2.0 is the most concrete expression of that bet. It is Google's answer to the question of what developer tooling looks like when the primary user of those tools is an AI agent rather than a human typing in an editor.

Whether that vision arrives on Google's timeline or takes longer than expected, the direction is clear. The coding tools you use in two years will look less like IDEs with AI features bolted on, and more like platforms for managing autonomous systems that happen to produce software.

Antigravity is Google's first serious attempt to build that platform.

The Neuron covers AI clearly — no hype, no jargon. Subscribe to the newsletter for weekly breakdowns of what actually matters in AI.