Google Launches Antigravity 2.0: Agent-First Development Platform

At Google I/O 2026, Google announced a significant shift in its AI-assisted development strategy with the launch of Antigravity 2.0. Moving away from traditional IDE-centric tools, the company has introduced a standalone, agent-first platform designed to facilitate multi-agent workflow management. This new ecosystem provides developers with a unified harness for building and deploying production-ready applications through a combination of a desktop application, a command-line interface (CLI), a software development kit (SDK), and enterprise-grade deployment options.

A New Agent-First Development Environment

Antigravity 2.0 functions as a standalone desktop application that serves as a central hub for agent interaction. It allows developers to orchestrate multiple agents simultaneously, enabling parallelized workflows through dynamic subagents. A standout feature is the platform’s support for scheduled tasks, which allows developers to define background automation pipelines that trigger agents automatically, moving beyond simple single-turn prompts. The application also integrates natively with Google AI Studio, Android, and Firebase, and includes voice command support to streamline the development experience.

Expanding the Developer Ecosystem

To support diverse development preferences, Google has released a suite of tools that share a common agent harness. The Antigravity CLI offers a terminal-based, high-velocity alternative for creating agents, replacing the previous Gemini CLI while maintaining support for existing features like Agent Skills, Hooks, and Subagents. For teams looking to embed agentic capabilities into their own internal products, the Antigravity SDK provides programmatic access to the same harness used by Google’s own internal agents.
For enterprise environments, the Gemini Enterprise Agent Platform allows organizations to connect Antigravity directly to Google Cloud projects. Additionally, the Managed Agents feature within the Gemini API enables developers to spin up agents in isolated Linux environments with a single API call. These agents support persistent, multi-turn sessions, allowing developers to maintain file states and context across follow-up calls without reinitialization.

Performance and Cross-Platform Integration

The entire Antigravity ecosystem is powered by Gemini 3.5 Flash, which Google has designated as the default model. According to the company, Gemini 3.5 Flash offers superior performance across most benchmarks compared to Gemini 3.1 Pro while operating four times faster than other frontier models. This speed is particularly critical for parallelized agent workflows where latency can compound across concurrent calls.
Google is also broadening the reach of its development tools through a new AI Studio mobile app, which allows for on-the-go prototyping and one-click project exports to the desktop Antigravity environment. Furthermore, new integrations with Google Workspace allow agents to interact programmatically with services like Docs, Sheets, and Calendar. Developers can also build Android apps through natural language prompts and publish directly to the Google Play Console from within AI Studio. For users requiring higher usage limits, Google has introduced an AI Ultra plan priced at $100 per month.