Google announced IDE extensions for its Antigravity 2.0 coding agent on Thursday, expanding the agent from a standalone desktop application into VS Code, Visual Studio, JetBrains IDEs (IntelliJ IDEA, PyCharm, WebStorm, GoLand, CLion, Rider), and Zed. The extensions allow developers to open agent conversations in a side panel, review inline diffs, inspect plans, and delegate multi-step engineering tasks without leaving their editor or moving into the Antigravity desktop app.
Antigravity 2.0 represents a significant shift in Google's strategy: moving from a standalone IDE to a ubiquitous agent harness distributed across developer workflows. The desktop app and CLI are now unified, sharing state and coordination logic. Google also released a Managed Agents API for cloud orchestration and an Antigravity SDK for self-hosted deployments. Enterprise customers get access through Gemini Enterprise Standard, Plus, or Emerging Market plans, with budget caps and usage controls.
Gemini 3.1 Pro, the backbone of Antigravity 2.0, scores 80.6% on SWE-bench Verified and 54.2% on SWE-bench Pro, positioning it behind Claude Opus 5 (89.1%) but ahead of GPT-4.1 alternatives in standard benchmarks. The agent includes specialized subagents for parallel task execution, cross-platform terminal sandboxing, credential masking, and hardened Git policies—designed for enterprise safety.
The expansion addresses a critical gap in the competitive landscape: while Cursor dominates inline editing and Claude Code dominates the terminal, Google Antigravity was confined to a desktop surface, making adoption friction high in enterprises that mandate IDE standardization. By meeting developers in their editor of choice (a lesson learned from past Google dev tool failures), Google aims to land inside engineering teams where Cursor, Claude Code, and Copilot already operate.
For architects: Antigravity's IDE-first strategy signals Google's intent to compete on distribution, not just model quality. The Managed Agents API and self-hosted SDK also position Antigravity as infrastructure for organizations building custom agent workflows. Key decision point: enterprises must now evaluate whether Gemini 3.1 Pro's SWE-bench score (80.6%) justifies switching from Claude Code (89.1%) in terminal workflows, balanced against Google's enterprise access controls and the existing Gemini Cloud ecosystem.