Google Antigravity 2.0 Turns Parallel AI Agents Into Your Entire Dev Team

The original Antigravity launched in November 2025 alongside Gemini 3 as a free, agent-first IDE. It was Google's answer to Cursor, a capable but narrowly scoped tool. Six months later, the picture looks very different. Google announced Antigravity 2.0 at I/O 2026, upgrading its agentic coding tool into a full developer platform with a revamped desktop app, a new CLI built in Go, and an SDK for custom agents.

The company announced Google Antigravity 2.0 as a standalone desktop application built entirely around agent orchestration, alongside an Antigravity CLI, an Antigravity SDK, Managed Agents in the Gemini API, and enterprise support through the Gemini Enterprise Agent Platform. Google is moving its developer tooling away from IDE-centric assistance and toward multi-agent workflow management as the primary abstraction.

As someone who covers this beat daily, I can say this one actually moves the needle. The shift from "AI-assisted coding" to "AI-orchestrated development" is not a marketing reframe. It's a fundamentally different architecture.

What Is Google Antigravity 2.0?

Unlike standard coding assistants that just autocomplete lines, Antigravity provides a "Mission Control" for managing autonomous agents that can plan, code, and even browse the web to help you build. Antigravity is designed as an agent-first platform. It presupposes that the AI is not just a tool for writing code but an autonomous actor capable of planning, executing, validating, and iterating on complex engineering tasks with minimal human intervention.

In the months since its original launch, Antigravity existed as a dedicated application that borrows heavily from Microsoft's VS Code but significantly integrates AI coding assistance. Version 2.0 is a clean break from that model. It is a new standalone desktop application, separate from the existing Antigravity IDE, designed fully around an agent-optimized experience.

How Multi-Agent Orchestration Works

The core architectural shift in 2.0 is parallelism. This architecture addresses a key limitation of previous IDEs that had more of a chatbot experience, which were linear and synchronous. In a traditional chat interface, the developer must wait for the AI to finish generating code before asking the next question. In Antigravity's Manager View, a developer can dispatch five different agents to work on five different bugs simultaneously, effectively multiplying their throughput.

Antigravity 2.0 introduces "dynamic subagents," which are for parallelized workflows, scheduled tasks for background automation, and new integrations with Google AI Studio, Android, and the Firebase mobile development platform.

Antigravity 2.0 adds more structure to daily development work. It can group conversations into Projects, work across multiple repositories, and schedule routine actions, while Antigravity CLI retains core Gemini CLI features such as Skills, Hooks, Subagents, and plugins.

The demo Google ran on stage during I/O made the strongest possible case for what this looks like in practice. During the keynote, Google claimed that Antigravity 2.0 was able to make the core framework of a working operating system in nearly 12 hours. The company also claimed that the AI platform launched 93 separate sub-agents during the task, processed billions of tokens, and completed the project for under $1,000 in computing costs. The company even demonstrated the system live on stage by attempting to run the classic game Doom on the AI-created OS.

Key Technical Highlights

• Gemini 3.5 Flash at the core. Google DeepMind Director of Software Engineering Varun Mohan noted that 3.5 Flash outperforms Gemini 3.1 Pro on almost all coding benchmarks while running four times faster than any other frontier model, giving developers the "high-speed engine needed for real-world agentic workflows."
• Self-referential development. In a neat bit of self-reference, Google says Gemini 3.5 Flash was itself co-developed using Antigravity.
• Voice commands. The desktop app gains native voice command support, letting developers talk to their agents rather than type.
• Managed Agents in the Gemini API. The Managed Agents feature in the Gemini API provides infrastructure-level isolation for agent execution. With a single API call, developers can spin up an agent that reasons, uses tools, and executes code in an isolated Linux environment.
• Persistent environments. Each interaction creates an environment that can be resumed in follow-up calls with all files and state intact, enabling multi-turn sessions without reinitializing context.

The CLI, SDK, and Vibe Coding on Android

Antigravity CLI handles keyboard-first work that does not need a full editor, including shell-heavy tasks and quick agent control. The Antigravity CLI is set to fully replace the previous Gemini CLI, meaning developers will need to migrate any of their existing agentic workflows to the new tool. According to the Google Developers Blog, Gemini CLI and Gemini Code Assist IDE extensions will stop serving requests on June 18, 2026.

Google also revealed the new Antigravity SDK, which provides programmatic access to the same agents found in Antigravity, enabling users to define custom agent behaviors and host them on third-party infrastructure.

On the vibe coding front, AI Studio's updated "vibe coding" experience integrates the Google Antigravity coding agent to convert prompts into production-ready apps, with support for multiplayer features, secure API key storage, and Firebase backend integration. Users can now describe an app idea in plain language and preview it inside an integrated Android emulator. Apps can also be tested directly on connected Android smartphones.

Google is also using Antigravity's coding capabilities in consumer products like Search, where users will get a custom UI generated in real time as part of an answer. Users will be able to build mini-apps while exploring a topic within search.

Pricing

Google is introducing a new AI Ultra tier at $100 per month, which offers five times the usage limits of the AI Pro plan. The existing top-tier AI Ultra subscription drops from $250 to $200 per month, delivering 20 times the Pro limits.

The $100 price point puts Google in direct competition with OpenAI's recently launched ChatGPT Pro tier and Anthropic's Claude Max, both of which also charge $100 per month for enhanced access to coding tools.

Higher ceilings matter when several agents are planning, running shell work, and evaluating code at once because those workflows can burn through requests far faster than a single assistant pane. This is not a vanity tier. It's a functional requirement for teams running multi-agent pipelines at any serious scale.

What This Means for the Industry

Competition helps explain why Google expanded the pitch beyond one coding assistant pane. Augment Code, JetBrains AI Assistant, and Windsurf already sell broader workflow products. Antigravity 2.0 is Google's attempt to consolidate that space under one brand with tighter Gemini integration than any third party can offer.

It's a workflow that makes app development feel less like coding and more like project management, leading a team of Gemini-powered AI agents. That framing is accurate, and it also signals where the developer tools market is heading. The question is no longer whether AI can write code. It's whether it can manage a full project lifecycle without a human in the loop at every step.

Final Thoughts

The OS-in-12-hours demo is the kind of claim that demands skepticism, but also demands attention. Initially, the OS failed because keyboard drivers were missing, but then the company instructed Antigravity 2.0 to generate the required drivers in real time, after which the game reportedly became functional. That detail is actually more interesting than the headline number. It shows agents recovering from failure autonomously, which is the hard part of agentic systems that most demos conveniently skip.

What I'll be watching closely is the Managed Agents infrastructure. The agent harness itself is the same technology and infrastructure that powers Google's own agents, co-optimized with Gemini 3.5 Flash. Persistent isolated environments mean each interaction creates an environment that can be resumed in follow-up calls with all files and state intact. That's a serious engineering commitment, and if it holds up under real workloads, it's a meaningful advantage over competitors who bolt agents onto existing IDEs. The migration deadline for Gemini CLI users is June 18, 2026, so the pressure to evaluate this stack is already real.

What do you think? Are you migrating your workflows to Antigravity 2.0, or are you sticking with Cursor, Windsurf, or another tool? Drop your thoughts in the comments.

---