Google Antigravity

Google Antigravity

The first agent-native development platform

Google Antigravity is a revolutionary IDE designed for the Gemini 3 era. It introduces an "Agent-First" interface with a "Mission Control" view to orchestrate multiple autonomous agents. Antigravity allows developers to delegate end-to-end engineering tasks—planning, coding, browsing, and verifying—to agents powered by Google's most capable models.

Kategorie

Agentic IDE

Bewertung

4.9

Benutzer

2M+

Plattform-Unterstützung

macOSWindowsLinuxWeb

Preispläne

Kostenloser Plan

$0

Public Preview: $0/month (Individual Plan)

Empfohlen

Pro-Plan

$0 /month

Individual: Access to Gemini 3 Pro, Claude Sonnet & Opus 4.5, gpt-oss-120b. Unlimited tab completions & command requests.

Empfohlen

Pro-Plan

Google One /sub

Developer Plan: For Google AI Pro/Ultra subscribers. More generous rate limits refreshed every 5 hours.

Enterprise-Plan

定制

Organization Plan: Coming soon. Self-serve and enterprise plans for teams of all sizes.

Kernfunktionen

1 Core Features

Agent Manager

A "Mission Control" dashboard to spawn and monitor multiple agents working in parallel

Integrated Browser

Agents can actively browse localhost and the web to debug UI and verify fixes

Gemini 3 Pro

Powered by Google's most capable reasoning model, optimized for long-horizon tasks

2 Model Optionality

Multi-Model Access

Native access to Claude Sonnet/Opus 4.5 and open-source gpt-oss-120b within the agent

Unlimited Autocomplete

Free unlimited tab completions powered by distilled Gemini models

3 Innovation

Vibe Coding

Build apps from vague natural language prompts with "human-in-the-loop" guidance

Self-Healing

Agents detect terminal errors and automatically attempt fixes without user intervention

Vor- und Nachteile

Vorteile

  • Unbeatable Free Tier: Access to Gemini 3 Pro and Claude Opus 4.5 for free during preview
  • Deep integration with Google One subscription for power users
  • True "Agentic" workflow: manages multiple agents for coding, testing, and browsing
  • Integrated browser allowing agents to "see" and fix UI bugs is a game changer
  • Parallel execution: one agent fixes a bug while another writes tests

Nachteile

  • Organization/Team plans are not yet available (Coming Soon)
  • Rate limits on the free plan are "generous" but opaque
  • Heavier resource usage than standard VS Code due to local agent processes

Ideal Für

Individual developers wanting free access to SOTA models (Gemini 3, Opus 4.5)

Google One subscribers looking for a "daily driver" IDE

Full-stack developers who want to offload repetitive CRUD tasks

Product Managers wanting to prototype ("Vibe Coding")

Tech-Stack

Architektur

Hybrid (VS Code Fork + Cloud Agents)

KI-Modelle

Gemini 3 ProClaude Sonnet 4.5Claude Opus 4.5gpt-oss-120b

Plattformen

macOSWindowsLinuxWeb

Community-Ressourcen