![antigravity 2point0]()
Google has officially introduced Antigravity 2.0, a major redesign of its AI-powered development platform that shifts the product from a traditional AI-assisted IDE into a fully “agent-first” software engineering environment. The announcement was made during Google I/O 2026 through the official Antigravity blog.
Download Antigravity 2.0
According to Google, Antigravity 2.0 is designed around autonomous AI agents capable of planning, coding, testing, debugging, and orchestrating long-running development tasks with minimal human supervision. Instead of simply adding AI features into an editor, Google says the platform is now built around agents as the primary workflow model.
The update introduces several major additions to the Antigravity ecosystem, including:
A redesigned desktop application
Antigravity CLI for terminal-first workflows
An SDK for building custom AI agents
Multi-agent orchestration capabilities
Shared agent infrastructure across tools
Google says the new Antigravity CLI replaces the earlier Gemini CLI tooling as part of a broader unification strategy around a single agentic development platform. In a separate developer announcement, Google confirmed that Gemini CLI will transition into Antigravity CLI beginning this year.
The company describes Antigravity as a platform where developers can delegate engineering tasks to autonomous agents that work across the editor, terminal, and browser simultaneously. Google’s official codelab explains that agents can plan architectures, execute workflows, browse documentation, run browser tests, and validate implementations independently.
One of the most notable aspects of Antigravity 2.0 is its multi-agent workflow system. Developers can run several AI agents in parallel for coding, reviewing, debugging, and research tasks across large repositories and long-running projects.
Google also confirmed deeper integration with Gemini 3.5 Flash, the company’s newest AI model optimized for agentic reasoning and coding workflows. Antigravity 2.0 acts as one of the core execution environments for Google’s broader “Agentic Gemini Era” strategy introduced at I/O 2026.
During I/O, Google demonstrated how Antigravity integrates with Gemini Spark, an always-running AI agent platform capable of continuously handling workflows in the background. Together, the systems represent Google’s push toward persistent autonomous AI agents operating across developer and productivity environments.
The launch also reflects the rapidly intensifying competition in AI coding platforms. Google is positioning Antigravity 2.0 against tools such as OpenAI Codex, Claude Code, Cursor, Windsurf, and Replit. Multiple reports noted that Google has aggressively redesigned Antigravity specifically around agentic software engineering workflows.
Google says Antigravity is built to support asynchronous development workflows where agents continue executing tasks even after developers leave their machines. The company described this as moving beyond “AI autocomplete” toward autonomous software engineering systems.
The platform also supports Model Context Protocol (MCP), allowing agents to connect securely with external tools, APIs, enterprise systems, and third-party services.
Industry observers view Antigravity 2.0 as one of Google’s strongest moves yet into the AI-native software development market, especially as enterprise and developer demand shifts from chat-based coding assistants toward full workflow automation platforms.
Developers can explore the platform and official resources here: