Google DeepMind Launches Antigravity 2.0 and Managed Agents to Automate Production Workflows

Google DeepMindGoogle DeepMind

· Updated

Google DeepMind expanded its Antigravity ecosystem with a new desktop orchestrator, a developer SDK, and API-managed agents that run in isolated Linux sandboxes. By pairing these tools with the high-speed Gemini 3.5 Flash model, Google is shifting AI development from single-turn chat to autonomous, multi-step agentic engineering.

Google DeepMind launched the Antigravity 2.0 ecosystem, a suite of tools designed to move AI from reactive chat to autonomous agentic engineering. The release includes a standalone desktop application for orchestrating parallel agents, a new CLI, and an SDK for programmatic control. These are powered by Gemini 3.5 Flash, which recently entered the Arena.ai leaderboards.
Model
Gemini 3.5 Flash
Speed
4x faster than other frontier models
Pricing
$100 per month AI Ultra plan
Environment
Isolated Linux sandboxes
Hackathon prize
$2 million XPRIZE pool

This expansion addresses the infrastructure friction that has kept agents in the prototype phase. By introducing Managed Agents in the Gemini API, developers can spin up agents in persistent, isolated Linux environments. This mirrors the industry shift toward programmable agent infrastructure seen in the GitHub Copilot SDK and OpenAI's Agents SDK.

You can now use Google AI Studio to build Android apps via natural language and export projects to Antigravity for local development. A new 100 dollar per month AI Ultra mid-tier plan provides 5x higher usage limits. Managed agents are available via the Interactions API, and a mobile app is available for pre-registration.

View the full update on blog.google
Google DeepMind
Google DeepMind
@GoogleDeepMind
X

We’re expanding the @Antigravity ecosystem to help developers shift their energy away from tedious debugging and focus on what actually matters: architecture and design. 🧵 https://t.co/Dr4EkIDVcF

24retweets168likes
View on X

Still wondering? A few quick answers below.

Google Antigravity 2.0 is an agent-first development platform designed to turn ideas into production-ready applications. It features a standalone desktop app that serves as a mission control for orchestrating multiple AI agents in parallel. The ecosystem includes a command-line interface and an SDK to help developers automate debugging and focus on high-level software architecture.

Managed Agents allow developers to spin up autonomous AI agents with a single API call. These agents can reason, use tools, and execute code within isolated Linux environments that persist across multi-turn sessions. This infrastructure is powered by the Antigravity agent harness and is co-optimized to run on the high-speed Gemini 3.5 Flash model.

The new Google AI Ultra plan starts at 100 dollars per month. It is designed for professional workflows, offering five times higher usage limits in Google Antigravity compared to the standard AI Pro plan. Subscribers can also claim 100 dollars in bonus credits for a limited time to cover usage if they exceed their plan quota.

This is a global competition with a 2 million dollar prize pool, the largest ever for a hackathon. Developers are invited to use Google's latest models and agentic tools to build real-world applications that solve global challenges, such as reducing food waste. Finalists will pitch their projects live at the Moonshot Gathering in Los Angeles this September.

Yes, Google AI Studio now includes native Android support that allows you to build high-quality mobile applications using natural language prompts. The platform also integrates with the Google Play Console, enabling developers to publish their AI-generated apps directly to a test track. Projects can be exported to the Antigravity desktop app for further local development.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Google

See all AI news & updates from Google →

Share this update