Google Antigravity 2.0 is launched, featuring a CLI, SDK, and AI agents.

      **TL;DR:** At I/O 2026, Google revealed Antigravity 2.0, enhancing its agentic coding tool into a comprehensive developer platform that features an updated desktop app, a new command-line interface (CLI) built in Go, and an SDK for creating custom agents. The update also introduces a $100/month AI Ultra plan and phases out the Gemini CLI for consumer users.

      Google has taken a bold step in the fast-growing agentic coding sector. On Monday at I/O 2026, the company launched Antigravity 2.0, evolving its original Cursor competitor into a vast developer platform equipped with a refreshed desktop application, a new command-line tool, and an SDK for developing custom agents.

      The initial Antigravity debuted in November 2025 alongside Gemini 3 as a free, agent-centric IDE. The 2.0 version represents a significant transformation. At its core is the newly designed desktop application, which emphasizes multi-agent orchestration. Developers now have the ability to deploy multiple agents to tackle problems concurrently, create custom workflows for subagents, and set up tasks that operate automatically in the background.

      Integration capabilities are extensive. Antigravity 2.0 seamlessly connects with Google AI Studio, Firebase, and Android, enabling developers to export projects from AI Studio directly to their local Antigravity instance while maintaining complete context. For users who favor terminal work, Google is introducing a completely new Antigravity CLI built in Go, which promises faster performance and improved responsiveness compared to its predecessor.

      The former Gemini CLI is being retired. Google announced that access to Gemini CLI and Gemini Code Assist IDE extensions for consumer users will end on June 18, 2026, affecting AI Pro, AI Ultra, and free-tier users. However, enterprise users with Gemini Code Assist Standard or Enterprise licenses will retain access. The message is clear: Antigravity is the way forward.

      Completing the platform is the Antigravity SDK, which provides developers with the tools to create custom agents on Google’s coding framework. Customers using Google Cloud can connect directly to Antigravity, and enterprise users have access to custom agent templates in AI Studio. The company has already utilized Antigravity internally to develop consumer features, including custom UI generation for Google Search.

      Driving this is the new Gemini 3.5 Flash model, which Google claims surpasses the coding and agentic benchmarks of Gemini 3.1 Pro while operating four times faster than competing frontier models. In an interesting nod, Google mentions that Gemini 3.5 Flash was co-developed using Antigravity. The desktop app also includes native voice command capabilities, allowing developers to communicate with their agents verbally instead of through typing.

      Regarding pricing, Google has launched an AI Ultra tier at $100 per month, which allows for five times the usage limits compared to AI Pro. The existing top-tier AI Ultra subscription has been reduced from $250 to $200 per month, offering 20 times the Pro limits. This $100 price puts Google in direct competition with OpenAI's new ChatGPT Pro tier and Anthropic's Claude Max, both priced at $100 per month for enhanced coding tool access. The premium tier at $200 aligns with Anthropic's leading Claude Max plan.

      This strategic timing is intentional. The market for agentic coding tools is thriving, and all major AI labs are vying for developer loyalty. GitHub recently halted new sign-ups for Copilot as agentic workloads transformed the economics of AI-assisted coding, and GitLab has begun reorganizing for what it calls the "agentic era." With Antigravity 2.0, Google is wagering that the future of coding will belong to those who can efficiently orchestrate the most agents across various platforms.