🇺🇸 USA · Google Antigravity

Status: 🟩 COMPLETE 🟦 LIVING Last updated: 2026-06-26 Plain-English tagline: Google’s agentic IDE — combines an editor surface with autonomous coding agents powered by Gemini. Released late 2025 as Google’s strategic answer to Cursor + Claude Code.


Front-matter facts

FieldValue
VendorGoogle LLC (Mountain View, USA)
Country / origin🇺🇸 USA
Recommended for Australian users?✅ Yes — fully accessible from AUS via Google account; integrates with Google Cloud for AUS data residency
Privacy summaryFree / Google AI Pro tier: Gemini Apps Activity training applies; opt-out at myactivity.google.com. Workspace / Enterprise: never trains. Vertex AI: never trains.
Free tierYes — Antigravity Free; rate-limited Gemini access
Paid tiersBundled with Google AI Pro (US249.99/mo); Workspace enterprise pricing for business deployments
First releasedAnnounced and limited preview November 2025; broader rollout 2026
Last reviewed2026-06-26
Official sitehttps://antigravity.google.com

What it is

Antigravity is Google’s agentic IDE — Google’s strategic entry into the AI-first coding tool space that Cursor (2023) and Claude Code (2024) established. It combines:

  • An editor surface (built on a fork or wrapper of VS Code, with Google-specific enhancements)
  • Gemini-powered agents that plan and execute multi-step coding tasks
  • Native integration with Google Cloud (Vertex AI, Cloud Run, GKE, Firebase, etc.)
  • Workspace integration (deploy to Google Cloud, sync to Drive, etc.)
  • Browser-based + desktop options

Antigravity was announced in November 2025 alongside Google’s broader push into agentic AI (Project Mariner browser agent, Project Astra visual assistant, etc.). It’s positioned as the “do real work with Gemini agents” companion to Gemini Code Assist (the older Gemini-in-IDE plugin).

Naming origin: “Antigravity” reportedly reflects Google’s framing that current AI coding tools are still working against gravity (manual file-by-file editing); Antigravity aims to lift you above that.


What you’d use it for

  • Multi-file coding tasks with Gemini agents — give a task, agents plan and implement
  • If you’re already deeply in Google Cloud — natural integration with Vertex AI, Cloud Run, Cloud SQL, Firebase, BigQuery
  • For Workspace organisations with Gemini AI Pro/Ultra subscriptions — Antigravity is included
  • Long-context coding — Gemini 3 Pro’s 1M-token context is well-suited for whole-codebase tasks
  • Deploying to Google Cloud as part of the coding flow
  • Mobile-first workflow — Antigravity has stronger mobile support than Cursor or Claude Code (rolling out)
  • If you’re a Google Cloud customer with AUS data residency requirements — Antigravity inherits Vertex AI’s australia-southeast1 Sydney / australia-southeast2 Melbourne data residency options

How to sign up + first 5 minutes from Australia

  1. Go to antigravity.google.com. Sign in with your Google account.
  2. If you have Google AI Pro or Google AI Ultra, Antigravity is included — no separate signup.
  3. If you’re in a Workspace org with Gemini enabled, your admin may have already enabled Antigravity.
  4. Download the desktop app (Mac / Windows / Linux) or use the web editor.
  5. Connect a Google Cloud project if you want deployment integration.
  6. Connect a GitHub / GitLab / Cloud Source Repositories repo.
  7. Try one of these:
    • “Add OAuth login using Firebase Auth to this Next.js project”
    • “Migrate this app from Cloud Functions to Cloud Run”
    • “Add a /reports endpoint that queries BigQuery and returns JSON”
  8. Antigravity plans, writes code, runs tests in a cloud sandbox, opens a PR.

What it costs — what you actually get

Free (limited)

  • Limited Antigravity tasks per month
  • Free Gemini Flash usage in editor
  • Suitable for trying it out

Google AI Pro — US36 incl GST)

  • Antigravity full access at Pro limits
  • Gemini 3 Pro in agents
  • Cloud sandbox tasks at moderate quotas
  • Workspace integration if you also have Workspace

Google AI Ultra — US$249.99/month

  • Heavy Antigravity usage
  • Gemini 3 Deep Think (extended reasoning) in agents
  • Higher cloud sandbox quotas
  • For developers using Antigravity many hours daily

Workspace + Gemini for Business / Enterprise

  • Antigravity included at enterprise tiers
  • No training on your code
  • AUS data residency via Vertex AI (Sydney / Melbourne)
  • Admin controls, audit logs
  • SAML SSO

Free for university students

  • Antigravity inherits Google AI Pro’s student program — verified .edu.au students get 12 months free

Vertex AI billing for advanced workloads

  • For very heavy agent workloads, can route to Vertex AI pay-per-token at Google Cloud rates

How it compares to alternatives

CapabilityGoogle AntigravityCursor ProClaude CodeOpenAI Codex (cloud)
SurfaceIDE + cloud agentStandalone IDETerminal CLICloud in chatgpt.com
Underlying modelGemini familyPick (Claude/GPT/Gemini/etc.)Claude familyGPT-5 / codex-1
Long-context1M tokens (Gemini 3 Pro)Varies by chosen model1M (Claude Sonnet)256K
Cloud deployment integrationBest (Google Cloud native)ManualManualLimited
Workspace / Office integrationBest for Workspace usersNoneNoneLimited
MCP supportNewerMatureNative (best)Newer
Free tier (for students)Yes (12 months AUS uni)Yes (cursor.com/students)LimitedNo
Multi-model pickerGemini-focusedBest (any model)Claude-focusedOpenAI-focused
AUS data residencyVia Vertex AI Sydney / MelbourneNoneVia AWS Bedrock SydneyNone native
Price (consumer)US249 (Ultra)US$20Bundled with Claude Pro/MaxUS$20-200 (ChatGPT)

When to pick Antigravity:

  • You live in Google Cloud and Workspace
  • You want Gemini 3 Pro’s long context for whole-codebase work
  • You’re an AUS student or org with Google Workspace
  • You want AUS data residency via Google Cloud Sydney/Melbourne
  • You prefer Google’s enterprise security / billing structure

When to pick Cursor / Claude Code / OpenAI Codex:

  • Pick Cursor for tab autocomplete + Composer + multi-model freedom
  • Pick Claude Code for terminal-first / agentic work with the richest Skills + MCP ecosystem
  • Pick OpenAI Codex if you live in ChatGPT / OpenAI infrastructure

Privacy / data handling

Free + Pro + Ultra: Gemini Apps Activity controls apply — same as general Gemini consumer products. Opt out at myactivity.google.com/product/gemini.

Workspace Business / Enterprise: Antigravity does NOT train on your code. Tenant-isolated. Contractually committed.

Vertex AI routing: for heavy agent workloads routed via Vertex AI, no-training applies and AUS data residency options available (australia-southeast1, australia-southeast2).

Source-code reading by Gemini: when Antigravity reads your codebase, it sends content to Gemini (in the Workspace tenant or via Vertex AI). For sensitive code, prefer Enterprise / Workspace tier.

Cloud sandbox tasks: ephemeral sandboxes destroyed after task completion. Logs may be retained per Google Cloud retention policy.


Recent changes

  • 2026: Generally available rollout expanded; Gemini 3 Pro + Gemini 3 Deep Think integration
  • 2025-12: Initial general availability after limited preview
  • 2025-11: Antigravity announced at Google Cloud Next / Gemini event

Gotchas

  • “Antigravity” is the new name — earlier Google AI coding products (Project IDX, Duet AI for Developers, Gemini Code Assist) overlap or have been re-branded. Verify which product name is current at antigravity.google.com.
  • Antigravity ≠ Gemini Code Assist — Code Assist is the older IDE-plugin product for VS Code / JetBrains; Antigravity is the standalone agentic IDE. Both still exist (as of mid-2026).
  • Workspace integration depth varies by Workspace SKU — admin must enable Antigravity at the tenant level for Workspace users.
  • AUS data residency via Vertex AI is opt-in for Enterprise — Workspace consumer tiers route through US data centres.
  • Gemini’s safety filtering can be more conservative than competitors on certain code-related topics — rephrase or switch tools if you get an unexpected refusal.
  • Cloud sandbox costs are bundled at consumer tier but heavy workloads may eventually push into Vertex AI billing — monitor usage if you’re a heavy user.
  • PR integration with GitHub is solid; with GitLab / Bitbucket the integration is weaker.

See also


Sources