Google launches Antigravity 2.0 with AI agents orchestration, but CLI shift sparks controversy

Google Transforms Antigravity Into Agent-First Development Platform

Google has launched Antigravity 2.0, repositioning its agentic coding app into a unified suite for agent-native software development. Announced at Google I/O, the update introduces a standalone desktop application that enables developers to orchestrate multiple AI agents and execute tasks simultaneously1

1

3

3

. The platform now allows users to design custom subagent workflows and schedule tasks that run automatically in the background, marking a shift from traditional coding assistance to true agent orchestration4

4

.

The revamped platform integrates seamlessly with Google AI Studio, Android, and Firebase, enabling developers to export projects from AI Studio directly to their local Antigravity app while preserving all context1

1

. Google has also added native voice command support to Antigravity, mirroring features already present in Gmail and Docs. The company positions this as a direct challenge to competitors like Claude Code and Codex, which have captured developer attention in recent months5

5

.

Gemini 3.5 Flash Model Powers Performance Leap

At the core of Antigravity 2.0 sits the Gemini 3.5 Flash model, which Google claims outperforms Gemini 3.1 Pro on nearly all coding benchmarks while running four times faster than any frontier model4

4

. The model was co-developed using Antigravity itself, demonstrating the platform's capabilities in real-world scenarios1

1

. This speed advantage becomes particularly valuable when multiplied across several agents working in parallel, allowing ideas to materialize in minutes rather than hours3

3

.

Google is also deploying Antigravity's coding capabilities in consumer products like Search, where users will see custom UIs generated in real-time as part of search answers. This integration enables users to build mini-apps while exploring topics within search results1

1

. The Gemini API now includes Managed Agents, which can spin up new agents that reason, use third-party tools, and execute code in isolated Linux environments with a single API call4

4

.

Antigravity CLI Replaces Open-Source Gemini CLI, Sparking Backlash

Google introduced the Antigravity CLI as a command-line interface tool for developers who prefer terminal-based workflows, but the launch comes with a controversial mandate. Starting June 18, 2026, the open-source Gemini CLI will stop serving requests for Google AI Pro and Ultra users, as well as those using it free of charge2

2

. Enterprise customers using Gemini Code Assist Standard or Enterprise licenses will retain access, but most developers must migrate to the closed-source Antigravity CLI2

2

.

The transition has triggered significant developer frustration. While the Gemini CLI GitHub page displays all the code that made the tool possible, the Antigravity CLI repository contains only a changelog, readme, and demonstration GIF2

2

. Google acknowledges there won't be feature parity between the two tools at launch, with some capabilities potentially taking time to arrive or not appearing at all2

2

. Developers have flooded GitHub with complaints about usage limits, with multiple users reporting they hit their weekly quota after just a couple of requests2

2

.

SDK and Pricing Changes Target Enterprise Adoption

Google launched an Antigravity software development kit that provides programmatic access to build custom agents based on the platform's technology. Developers can define custom agent behaviors and host them on third-party infrastructure, with Google releasing custom agent templates in AI Studio for enterprise users1

1

4

4

. Google Cloud customers can connect to Antigravity to build projects, with an export tool available in AI Studio for developers to transfer existing projects locally1

1

.

On the pricing front, Google introduced a new AI Ultra pricing tier at $100 per month, offering 5x higher usage limits in Antigravity compared to the Pro plan. The company also reduced its top-tier AI Ultra plan from $250 to $200, which provides 20x higher limits than the Pro plan1

1

. These changes mirror strategies from Anthropic and OpenAI, which have introduced tiered pricing systems to accommodate varying developer needs1

1

.

Android App and Cross-Platform Integration Expand Accessibility

Google AI Studio is gaining a dedicated Android application that allows developers to capture ideas on mobile devices and find working prototypes ready by the time they reach their desks3

3

4

4

. The app supports voice and text input, starts from high-quality example apps, and includes social sharing features. Pre-registration is now available in the Google Play Console3

3

.

The platform now enables developers to build native Android applications with a single prompt and publish apps directly to the test track within the Google Play Console4

4

. Google AI Studio has also integrated with Google Workspace, allowing AI agents to call relevant Workspace APIs and embed them directly into applications4

4

. This workflow transforms app development from traditional coding into project management, where developers lead teams of Gemini-powered AI agents3

3

.