We’re unveiling a new generation of tools designed to help developers transform ideas into production-ready applications, including major enhancements to Google Antigravity, an upgraded Gemini API, and native Android development capabilities within Google AI Studio.
At Google I/O 2026, we’re accelerating the transition from simple prompts to intelligent action through the introduction of Gemini 3.5 Flash. By combining frontier-level intelligence with exceptional performance, Gemini 3.5 Flash surpasses Gemini 3.1 Pro across nearly every benchmark while operating four times faster than competing frontier models. This provides developers with the speed and responsiveness required for real-world agentic workflows.
To help developers take advantage of these capabilities, we are introducing the new Google Antigravity 2.0 desktop application, Managed Agents within the Gemini API, and native Android vibe coding support in Google AI Studio. Below are the major updates for developers building the next generation of agent-powered applications.
The Google Antigravity Ecosystem
Google Antigravity serves as our agent-first development platform, enabling developers to transform concepts into fully deployable applications. Today, we are significantly expanding the Antigravity ecosystem to support the creation, management, and deployment of agents that can operate across multiple developer environments.
New Ways to Use Antigravity
Antigravity 2.0
Antigravity 2.0 introduces a dedicated standalone desktop application designed specifically for agent-centric development. The platform acts as a central workspace where developers can coordinate multiple agents and execute tasks simultaneously.
Key capabilities include:
- Dynamic subagents that enable parallel execution of complex workflows
- Scheduled tasks that automate background operations
- Deep integrations with Google AI Studio, Android development environments, and Firebase
Antigravity CLI
Developers who prefer terminal-based workflows can use the new Antigravity CLI, a lightweight and highly responsive interface for creating and managing agents directly from the command line.
The CLI eliminates the need for a graphical interface and enables rapid agent creation and deployment. Existing Gemini CLI users are encouraged to transition to Antigravity CLI to take advantage of the latest capabilities.
Antigravity SDK
The Antigravity SDK provides programmatic access to the same agent infrastructure that powers Google’s internal products.
Optimized specifically for Gemini models, the SDK enables developers to:
- Define custom agent behaviors
- Create specialized workflows
- Deploy agents on their preferred infrastructure
Antigravity in Gemini Enterprise Agent Platform
For enterprise customers, Antigravity now integrates directly with the Gemini Enterprise Agent Platform.
This allows Google Cloud users to connect Antigravity workflows directly to their cloud projects, simplifying deployment and enterprise-scale operations.
Google AI Ultra Subscription Starting at $100
We are also introducing a new Google AI Ultra subscription tier priced at $100 per month.
Designed for advanced users and professional developers, the plan includes a usage limit within Google Antigravity that is five times higher than the existing Google AI Pro plan.
For a limited period, both new and existing Google AI Ultra subscribers will receive an additional $100 in Antigravity credits. These credits automatically become available if subscribers exceed their standard quota allocation.
Users can claim the offer directly through the Antigravity application. The promotional offer remains available until May 25, 2026, subject to applicable terms and conditions.
Managed Agents in the Gemini API
We are introducing Managed Agents within the Gemini API, making it possible to launch fully capable AI agents with a single API call.
These agents can reason, execute code, utilize external tools, and operate within isolated Linux environments. Managed Agents are powered by the Antigravity agent harness, built on Gemini 3.5 Flash, and are accessible through both the Interactions API and Google AI Studio.
Antigravity Agent Harness
Developers gain access to the same underlying infrastructure and technology stack used by Google’s own AI agents.
The harness has been specifically optimized to work alongside Gemini models, with particularly strong performance when paired with Gemini 3.5 Flash.
Persistent and Isolated Environments
Each interaction automatically generates a dedicated environment that preserves files, state information, and context between sessions.
This persistence enables seamless multi-step interactions and long-running workflows without losing continuity.
Custom Agent Creation
Developers can extend Antigravity agents using custom instructions and reusable skills defined through markdown-based configurations.
Google AI Studio now includes new custom agent templates within the Playground environment, allowing developers to begin building specialized agents immediately.
Google AI Studio Mobile and Ecosystem Integrations
Google AI Studio continues to evolve as a platform where developers can rapidly transform ideas into working applications using the latest Gemini models.
Today’s updates significantly expand both where and how developers can use the platform.
Google AI Studio Mobile App
Available for pre-registration this week, the new Google AI Studio mobile application allows developers to capture ideas while on the move and begin generating working prototypes immediately.
The goal is simple: start with an idea on your phone and have a functioning prototype waiting when you return to your workstation.
Google Workspace Integration
AI agents can now directly access and interact with relevant Google Workspace APIs.
This enables developers to integrate Workspace functionality natively into their applications and workflows without requiring extensive custom integrations.
Export to Antigravity
A new export capability allows developers to transfer entire projects from Google AI Studio directly into Antigravity with a single click.
Project context, files, and workflow history are preserved throughout the process, allowing developers to continue working exactly where they left off.
Native Android Support
Google AI Studio now includes native Android development support.
Developers can generate high-quality Android applications directly from prompts, dramatically reducing the time required to build and prototype mobile experiences.
Additionally, support for the Google Play Console has been introduced, allowing developers to publish applications directly to testing tracks from within Google AI Studio.
The Build with Gemini XPRIZE Hackathon
We are also launching the Build with Gemini XPRIZE Hackathon, a global competition featuring a total prize pool of $2 million, making it the largest hackathon prize pool ever offered.
Participants will have access to Gemini models, developer tools, and platform capabilities to create applications that address some of the world’s most important challenges.
Potential solution areas include:
- Reducing food waste
- Advancing scientific discovery
- Improving sustainability initiatives
- Solving large-scale societal problems
Finalists will present their projects live at the Moonshot Gathering in Los Angeles this September, where they will compete for the grand prize.
The Future of Agentic Development
The industry is rapidly moving beyond single-turn interactions toward collaborative, always-on AI agents that can work continuously alongside developers.
This shift fundamentally changes how software is designed, developed, and deployed—and we are only at the beginning of that transformation.
Our goal is to make this transition as seamless as possible. With the introduction of the Antigravity platform, Managed Agents in the Gemini API, and expanded capabilities within Google AI Studio, we are removing much of the infrastructure complexity that traditionally slows development.
By reducing operational friction, developers can focus on what matters most: turning ideas into real-world applications faster than ever before.
