Skip to main content

GoogleAI Coding Agent

Google Antigravity

Google's agent platform for software work. Agents plan, write code, run terminal commands, and test in a browser. They run from the Antigravity 2.0 desktop app, the CLI, the original IDE, or extensions for VS Code, Visual Studio, JetBrains, Zed, and Xcode. It runs Gemini models and, on eligible plans, Anthropic's Claude models.

The 30-Second Take

Pick Google Antigravity through Gemini Enterprise seats, because those Cloud terms keep your code out of training.

Its agents plan, edit code, run commands and test in a browser, from a desktop app, a CLI or editor extensions. One Google contract then covers Gemini and Claude models.

Best fit
Google Cloud organizations that can put engineers on Gemini Enterprise Standard or Plus seats.
Main constraint
Google says Antigravity lacks SOC 2 and ISO 27001 coverage, so hold back where audits need them.
First move
Start on Gemini Enterprise seats with terminal auto-execution set to review and the sandbox on.

Why

Watch the rules: Google changed Antigravity's quota and plan terms four times in eleven months, so budget against today's and recheck each quarter. Google also says it lacks SOC 2 and ISO 27001 coverage, so hold back where audits need them.

Practitioner sentiment, all practitioners−16 Mixed

98 practitioners, 140 sources reviewed, Updated Nov 1, 2026How this is measured

Read the GILD Review
Save a product your organization uses.Keep up with Google Antigravity without adding it to your stack.Visit Website
Operating stack
Agentic Software Development

Comparisons

How does Google Antigravity compare with the alternatives?

See all 4 comparisons

  1. Google Antigravity vs Cursor

    • AI Coding Agents

    Where engineers work decides it, since both agents plan, edit code, run commands, and check UI in a browser. Cursor centers on its own editor, with cloud agents and models from several providers. Antigravity centers on a desktop agent manager and CLI, reaches other editors through extensions, and comes with Gemini Enterprise seats.

  2. Google Antigravity vs Gemini CLI

    • AI Coding Agents

    Move each team to Antigravity CLI before its Code Assist contract ends. Gemini CLI now serves only existing Code Assist subscribers and paid API users, and Google no longer sells new Code Assist subscriptions. Antigravity CLI is the successor for terminal work, with the desktop app and editor extensions around it.

  3. Google Antigravity vs Claude Code

    • AI Coding Agents

    Claude Code and Antigravity run agents from desktop apps, terminals, and editor extensions, so compare the models you need and where you buy them. Claude Code uses Anthropic's agent software and contract. Antigravity runs Gemini and, on eligible plans, Claude under a Google contract.

Operator Insights

What are engineering leaders learning from deploying Google Antigravity?

Buying and deployment

What engineering leaders should verify before a broader rollout.

Sourced from Google documentation

How you buy it

US$30

Per seat per month on Gemini Enterprise Standard. Individuals use it free or on Google AI Pro.

Source 5, antigravity.google, Antigravity pricing: Antigravity pricingSource 6, one.google.com, Google AI plans: Google AI plansSource 2, Google, Gemini Enterprise quotas and pay-as-you-go usage: Gemini Enterprise quotas and pay-as-you-go usage

Details

Individuals use it free, on Google AI Pro at US$19.99 per month, or on Google AI Ultra at US$99.99 per month. Companies get it in Gemini Enterprise Standard or Plus seats, from US$30 per seat per month, or on a no-seat-fee Pay-as-you-go edition, both on an invoiced Cloud Billing account.

Cost drivers

US$10 a seat

Of monthly credit on Standard, US$15 on Plus, pooled weekly with no rollover.

Source 2, Google, Gemini Enterprise quotas and pay-as-you-go usage: Gemini Enterprise quotas and pay-as-you-go usageSource 7, Google, Antigravity docs: Enterprise: Antigravity docs: Enterprise

Details

Standard seats include US$10 and Plus US$15 of Antigravity credit per user per month, pooled and enforced weekly with no rollover. Overage runs at pay-as-you-go rates if enabled, and Claude Opus 5.5 and Sonnet 5.5 bill per token on top.

Where it runs

Desktop, CLI, IDEs

The 2.0 app and CLI on three systems, and extensions for VS Code, JetBrains, Zed, and Xcode.

Source 8, antigravity.google, Antigravity blog: Introducing Google Antigravity 2.0: Antigravity blog: Introducing Google Antigravity 2.0Source 3, Google, AI developer tools overview: AI developer tools overviewSource 7, Google, Antigravity docs: Enterprise: Antigravity docs: Enterprise

Details

The Antigravity 2.0 desktop app runs on macOS, Linux, and Windows, alongside the Antigravity CLI and Antigravity extensions for VS Code, Visual Studio, JetBrains, Zed, and Xcode. Google does not support the standalone Antigravity IDE for enterprise deployments.

What to verify

Auto-execution on

Gemini Enterprise defaults terminal commands to Always proceed, with the sandbox off.

Source 4, Google, AI developer tools settings: AI developer tools settingsSource 3, Google, AI developer tools overview: AI developer tools overviewSource 10, Google, Antigravity terms: Antigravity termsSource 9, Google, Antigravity docs: FAQ: Antigravity docs: FAQ

Details

In Gemini Enterprise, terminal auto-execution defaults to Always proceed and the sandbox to off. Google states that Antigravity in Gemini Enterprise does not support SOC 1, 2, or 3, ISO 27001, or FedRAMP. On personal sign-ins Google uses interactions to improve its products and machine learning technologies, and users can opt out of data collection in Settings.

Signals

What is the current evidence saying about Google Antigravity?

What engineering teams and individual builders report publicly about Google Antigravity, across the original IDE, Antigravity 2.0, the CLI, and the IDE extensions. Every finding links to its sources. Each source states whether it still applies after Google's May 19, 2026 product and plan changes.

Whose sources count

Practitioner sentiment, all practitioners

−16 Mixed net

of 102 sources that take a side, each author counted once.

How this figure is made

10 more state no overall view and cast no vote.

Leaders and teams have 13 accounts; GILD prints their sentiment from 20.

Current sentiment

Positive, mixed and negative sources

The score counts them as 32 positive, 18 mixed and 48 negative votes, one per author.

Public accounts lean negative. Most sources that discuss quotas raise a problem, and reliability draws concern from 24. Praise goes most often to multi-step execution, plan and walkthrough artifacts, and access to Claude models.

Use a task pilot to measure accepted work and review time.

Evidence field. All 140 sources, one mark each.

Arrange by

Each mark is one source: positive, mixed, negative, no opinion stated, or documentation (not counted).

By sentiment, All practitioners. Every source GILD read, grouped as the field draws it.
GroupSourcesPositiveMixedNegativeNo opinion stated and not counted in the scoreDocumentation, checked and not counted
Positive35350000
Mixed18018000
Negative49004900
Not counted21000210
Documentation17000017
Every source GILD read, by the quarter it was published, as the field's time axis counts it. Documentation is dated by the day it was checked.
QuarterSources
Q4 202516
Q1 202625
Q2 202639
Q3 202637
Q4 20266
Documentation, checked17

By sentiment

17 documentation pages check facts and are not counted.

Hover a mark for a quick look, click it or press Enter to open its source.Tap a mark to open its source, then step to its neighbors with Previous and Next.

What practitioners praise

Top praise of 5. Each mark is a source on it; filled marks praise it.

  1. Multi-Step Execution17 of 18 sources on it praise it

Most common concerns

Top concern of 7. Each mark is a source on it; filled marks raise it.

  1. Quotas & Plan Changes28 of 37 sources on it raise it as a concern
All 140 sources behind these findings
  1. Antigravity agent stuck in endless file-reading loop (Gemini models)SuneraX, Antigravity user (self-described: App development), Forum post, published Oct 12, 2026
  2. Antigravity CLI: Workspace and Google AI Pro accounts are practically unusableHans, runs a team's AI agents on its own server (self-described: Daily operations with several AI agents on a company server), Forum post, published Oct 12, 2026
  3. Comment on: I made an open-source Boom 3D clone. Wanna check it out?joaoviniciusgs, Hacker News commenter (self-described: Solo open-source Rust and Tauri app), Hacker News comment, published Oct 12, 2026
  4. PlansGoogle (Official), Product documentation, checked Oct 12, 2026
  5. ModelsGoogle (Official), Product documentation, checked Oct 12, 2026
  6. Antigravity in Gemini EnterpriseGoogle (Official), Product documentation, checked Oct 12, 2026
  7. Agent permissionsGoogle (Official), Product documentation, checked Oct 12, 2026
  8. Terminal sandboxGoogle (Official), Product documentation, checked Oct 12, 2026
  9. Managing AI credits and quotas (CLI)Google (Official), Product documentation, checked Oct 12, 2026
  10. IDE extensionsGoogle (Official), Product documentation, checked Oct 12, 2026
  11. Migrating from Gemini CLIGoogle (Official), Product documentation, checked Oct 12, 2026
  12. FAQGoogle (Official), Product documentation, checked Oct 12, 2026
  13. ChangelogGoogle (Official), Product documentation, checked Oct 12, 2026
  14. Google Antigravity Additional Terms of ServiceGoogle (Official), Product documentation, checked Oct 12, 2026
  15. Changes to Antigravity plansGoogle (Official), Product documentation, checked Oct 12, 2026
  16. Bringing Antigravity to Gemini EnterpriseGoogle (Official), Product documentation, checked Oct 12, 2026
  17. AI developer tools overviewGoogle (Official), Product documentation, checked Oct 12, 2026
  18. Configure AI developer tools settingsGoogle (Official), Product documentation, checked Oct 12, 2026
  19. Quotas and pay-as-you-go usageGoogle (Official), Product documentation, checked Oct 12, 2026
  20. Gemini Code Assist Standard and Enterprise subscriptionsGoogle (Official), Product documentation, checked Oct 12, 2026
  21. Comment on: OpenSCAD the Programmers Solid 3D CAD Modellergeorgemcbay, Long-time OpenSCAD user (self-described: Hobby CAD), Hacker News comment, published Oct 10, 2026
  22. Comment on: Gemini 4 Argontinyhouse, Hacker News commenter, former Pro subscriber (self-described: Compared more than five harnesses), Hacker News comment, published Oct 1, 2026
  23. Comment on: Gemini 4 Argonmapontosevenths, Hacker News commenter, Google AI Ultra subscriber (self-described: Terminal workflows), Hacker News comment, published Oct 1, 2026
  24. Comment on: Gemini 4 Argontaylorfinley, Hacker News commenter (self-described: Personal local-model setup), Hacker News comment, published Sep 30, 2026
  25. Comment on: Gemini 4 Argonsmartbit, Hacker News commenter (self-described: Documentation and coding in the terminal), Hacker News comment, published Sep 30, 2026
  26. Comment on: Gemini 4 Argonhuydotnet, Hacker News commenter (self-described: Personal; Google AI Pro through their phone carrier), Hacker News comment, published Sep 30, 2026
  27. Comment on: Gemini 4 Argonmartythemaniak, Hacker News commenter (self-described: Robotics hobby project), Hacker News comment, published Sep 30, 2026
  28. Comment on: Gemini 4 Argonpiyh, Hacker News commenter (self-described: Parallel work in worktrees), Hacker News comment, published Sep 30, 2026
  29. Stress Test: 100% Antigravity + Gemini Flash, 102.8B tokens and 474K LOCJanson79jc, solo developer (self-described: Solo, 474K line codebase over 465 sessions), Telemetry report (GitHub repository), published Sep 27, 2026
  30. Comment on: Plan mode is deadparasti, Hacker News commenter (self-described: Hobby projects; Claude Code at work), Hacker News comment, published Sep 26, 2026
  31. Severe slowdowns and frequent 503 errors on Gemini Flash 3.8 during specific hoursFreya Smith, uses Antigravity for work (self-described: Daily work between 16:00 and 22:00 UTC), Forum post, published Sep 25, 2026
  32. Problem with antigravity CLI using shell commands for everythingflying_manta, Antigravity CLI user (self-described: Daily CLI use), Forum post, published Sep 23, 2026
  33. We All Have a "Serious Work" AI and a "Just Vibing" AI. When Did That Happen?Dhruv Jani, developer (self-described: Daily IDE work), Personal blog post, dev.to, published Sep 22, 2026
  34. Comment on: GPT-5.6 Luna vs. GPT-6 Astra: Is a US$1.20 Model Good Enough for Code Review?dlahoda, Engineer running a review orchestration on their team's repository (self-described: Team repository with several review bots), Hacker News comment, published Sep 14, 2026
  35. Comment on: Ask HN: What default model do you use and why?vallerie, Hacker News commenter (self-described: Plan and test-harness driven work), Hacker News comment, published Sep 12, 2026
  36. Comment on: Benchmark: CadQuery vs. OpenSCAD for agentic CAD workevilduck, Hacker News commenter (self-described: Home and hobbyist CAD for 3D printing), Hacker News comment, published Sep 12, 2026
  37. Comment on: GPT-6 Astrarafaelmn, Hacker News commenter (self-described: Long-planned side project), Hacker News comment, published Sep 5, 2026
  38. Comment on: Gemini 3.8 Flash and 3.8 Flash Cyberallthetime, Hacker News commenter (self-described: Company codebase: web, React Native, CMS, ETL and websocket services), Hacker News comment, published Sep 3, 2026
  39. Antigravity's terms of service and third-party useGergely Orosz, author of The Pragmatic Engineer (self-described: Writes about engineering teams and their tooling; deciding whether to use Antigravity himself), X post, published Sep 3, 2026
  40. Comment on: Google Antigravity introduces Boost deep reasoning (/boost)dpweb, Hacker News commenter, student plan (self-described: Personal devops and home network), Hacker News comment, published Sep 1, 2026
  41. Comment on: Google Antigravity introduces Boost deep reasoning (/boost)sanarothe, Non-software engineer (self-described: Personal projects on a personal plan), Hacker News comment, published Sep 1, 2026
  42. Comment on: Google Antigravity introduces Boost deep reasoning (/boost)dan_voronov, Blogger reviewing AI programming tools (self-described: Long-term testing of Antigravity 1 and 2), Hacker News comment, published Sep 1, 2026
  43. Comment on: Google Antigravity introduces Boost deep reasoning (/boost)leoedin, Hacker News commenter, former subscriber (self-described: Personal use), Hacker News comment, published Sep 1, 2026
  44. Comment on: Our decision on Cursor following its acquisition by SpaceXhn_throw2025, Hacker News commenter, former Cursor subscriber (self-described: Uses Antigravity 2.0 beside a JetBrains IDE), Hacker News comment, published Aug 29, 2026
  45. Comment on: Why does Opus 5 feel worse to work with?slaser79, Hacker News commenter (self-described: Holds Anthropic, OpenAI and Google subscriptions), Hacker News comment, published Aug 14, 2026
  46. Nothing to See Here: How an AI Agent Buried Its Own CommandsJesus Padilla Soqui, Straiker STAR team (self-described: Agentic AI security company assessing the Antigravity platform), Security research write-up, Straiker blog, published Aug 14, 2026
  47. Preventing Quota Crashes via Antigravity CLI Agent HooksTanaike, Google Developer Expert (self-described: Autonomous Antigravity CLI sessions), Personal blog post, dev.to (Google Developer Experts), published Aug 12, 2026
  48. Comment on: Discovery Loopqlte, Hacker News commenter, Google AI Pro subscriber (self-described: Personal use after the Gemini CLI shutdown), Hacker News comment, published Aug 5, 2026
  49. Comment on: So what is your programming setup? Hype is making fundamentals confusingmalkamius, Hacker News commenter (self-described: Hobby game add-ons, run in a VM), Hacker News comment, published Jul 27, 2026
  50. Comment on: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyberdudeinhawaii, Hacker News commenter, former Google AI Ultra subscriber (self-described: Daily use of several providers), Hacker News comment, published Jul 22, 2026
  51. Comment on: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cybertobias2014, Hacker News commenter (self-described: Complex problems, run semi-autonomously), Hacker News comment, published Jul 22, 2026
  52. Comment on: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyberstonewhite, Hacker News commenter buying AI tools for their company (self-described: Company that standardized on Google AI Ultra), Hacker News comment, published Jul 21, 2026
  53. Comment on: Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cybermythz, Hacker News commenter (self-described: Solo, full-time development), Hacker News comment, published Jul 21, 2026
  54. Comment on: Kimi K3, Qwen 3.8, and Anthropic's (Potential) Unravellingmwryl, Hacker News commenter (self-described: Team building products on low-cost models; replaced Antigravity for development), Hacker News comment, published Jul 20, 2026
  55. Trusted automation with Google Antigravity: Scaling secure finance integrations from 40 Days to 5Rohan Soni, Google, writing Inspired Entertainment's case study with its CTO (self-described: Integration team building Java Spring Boot wallet adapters), Vendor case study, Google Developer forums, published Jul 20, 2026
  56. The Week of Sandbox Escapes Day 1: Escaping Antigravity's Allow-Default SeatbeltDan Lisichkin, security researcher at Pillar Security (Pillar Research team testing agentic IDE sandboxes on macOS), Security research write-up, Pillar Security blog, published Jul 20, 2026
  57. Starting and iterating on a Kaggle competition in Google AntigravityAndrey Lukyanenko, ML engineer and Google Developer Expert (self-described: Kaggle Playground competition), Personal blog post, andlukyane.com, published Jul 9, 2026
  58. Comment on: The classifiers Anthropic puts in front of Fable are too zealousSwellJoe, Developer building a security auditing harness (self-described: Security auditing harness and benchmarks), Hacker News comment, published Jul 8, 2026
  59. Ghostfabric: How An AI Coding Agent Was Tricked Into Exfiltrating Data Over VXLANChristopher Tuz, AI security researcher on the STAR Labs team at Straiker (Agentic AI security company red teaming coding agents), Security research write-up, Straiker blog, published Jul 8, 2026
  60. Comment on: I'm opening VSCode less and less every dayyogendra6698, Builds AI agents professionally (self-described: Daily development), Hacker News comment, published Jul 3, 2026
  61. Comment on: Google limits Meta's use of its Gemini AI modelsmoshegramovsky, Hacker News commenter (self-described: Personal C++ development), Hacker News comment, published Jun 28, 2026
  62. Comment on: Computer use in Gemini 3.5 Flashanigbrowl, Hacker News commenter (self-described: New paying Gemini API customer), Hacker News comment, published Jun 25, 2026
  63. Comment on: AI's Affordability Crisistrollbridge, Hacker News commenter running a small development shop (self-described: Small shop that competes with offshore development firms), Hacker News comment, published Jun 23, 2026
  64. Comment on: The text in Claude Code's "Extended Thinking" outputlinsomniac, Hacker News commenter (self-described: Spec-first workflow across agents), Hacker News comment, published Jun 22, 2026
  65. Comment on: John Jumper to join Anthropicbjackman, Hacker News commenter (self-described: Personal use of Gemini and Claude), Hacker News comment, published Jun 20, 2026
  66. Comment on: Running local models is good nowpolotics, Hacker News commenter (self-described: Solo Android app, about 10,000 lines of Kotlin), Hacker News comment, published Jun 16, 2026
  67. Comment on: DiffusionGemma: 4x Faster Text Generationevilturnip, Hacker News commenter (self-described: Personal projects), Hacker News comment, published Jun 10, 2026
  68. Comment on: Microsoft starts canceling Claude Code licensesseabrookmx, Hacker News commenter (self-described: Team at a large Google Cloud customer), Hacker News comment, published May 23, 2026
  69. Comment on: Google's Antigravity bait and switchmark_l_watson, Retired developer and author (self-described: One-month Google AI Ultra test), Hacker News comment, published May 22, 2026
  70. Comment on: Google's Antigravity bait and switchheyyeah, Hacker News commenter (self-described: Installing Antigravity IDE after 2.0), Hacker News comment, published May 22, 2026
  71. Comment on: Antigravity 2.0 Tops the OpenSCAD Architectural 3D LLM Benchmarku8, Developer of CLI developer tooling (self-described: Personal site build), Hacker News comment, published May 22, 2026
  72. Comment on: Google's Antigravity Bait and Switchdaft_pink, Hacker News commenter (self-described: Workspace user), Hacker News comment, published May 21, 2026
  73. Comment on: Google's Antigravity Bait and Switchspankalee, Hacker News commenter (self-described: Existing Antigravity user), Hacker News comment, published May 21, 2026
  74. My AI Agent Hit Google's Quota Wall in 8 Minutes. 36 Hours Later, Google Tripled the Limits.Joske Vermeulen, runs an autonomous agent startup race (self-described: Autonomous agent on cron, US$20 Pro plan), Personal blog post, dev.to, published May 21, 2026
  75. Antigravity CLI First Impressions: Fast, Rough, and Not ReadyJason Agostoni, developer and author of the Ship Bench benchmark (self-described: Ship Bench run on Windows), Personal blog post, dev.to, published May 21, 2026
  76. OpenSCAD LLM Benchmark: Building the PantheonModelRift engineering team (self-described: Company benchmarking coding agents for its OpenSCAD product), Company benchmark post, ModelRift blog, published May 21, 2026
  77. Google's Antigravity Bait and SwitchSid, independent developer on Google AI Ultra (self-described: Daily production work), Personal blog post, Sid's Blog, published May 21, 2026
  78. Recent Antigravity Changes Have Seriously Hurt the ExperienceMadhur Mishra, Antigravity user (self-described: Daily development), Forum post, published May 20, 2026
  79. Comment on: Gemini CLI will stop working from June 18, 2026MisterPea, Hacker News commenter (self-described: Personal use across Google tools), Hacker News comment, published May 20, 2026
  80. Comment on: Gemini CLI will stop working from June 18, 2026No1, Hacker News commenter (self-described: First trial), Hacker News comment, published May 20, 2026
  81. Comment on: Gemini CLI will stop working from June 18, 2026tegeek, Hacker News commenter (self-described: Four months of daily use with Kiro and Claude Code), Hacker News comment, published May 20, 2026
  82. Comment on: Qwen3.7-Max: The Agent Frontiercft, Hacker News commenter, Google AI Pro subscriber (self-described: Personal; Flash for auxiliary tasks), Hacker News comment, published May 20, 2026
  83. Comment on: Gemini CLI will stop working from June 18, 2026almog, Hacker News commenter, Google AI Pro subscriber (self-described: Personal firmware reverse engineering), Hacker News comment, published May 20, 2026
  84. I'm done with AntigravityVicente Cortes, Google One Pro subscriber for over a year (self-described: Personal projects), Forum post, published May 19, 2026
  85. Comment on: Cursor Introduces Composer 2.5smnscu, Hacker News commenter (self-described: Personal development), Hacker News comment, published May 19, 2026
  86. Comment on: Gemini 3.5 Flashlern_too_spel, Hacker News commenter (self-described: Personal account, small project), Hacker News comment, published May 19, 2026
  87. Comment on: A History of IDEs at Googlemcoliver, Hacker News commenter, Workspace administrator (self-described: Moving their organization off the Workspace AI Ultra add-on), Hacker News comment, published May 14, 2026
  88. Comment on: AlphaEvolve: Gemini-powered coding agent scaling impact across fieldsjatora, Hacker News commenter (self-described: Personal use of several agents), Hacker News comment, published May 8, 2026
  89. Comment on: Using coding assistance tools to revive projects you never were going to finishsaidinesh5, Hacker News commenter (self-described: Weekend home-automation projects), Hacker News comment, published Apr 27, 2026
  90. Comment on: Google is building a Claude Code challenger, Sergey Brin is involvedjdw64, Hacker News commenter (self-described: Subscribes to several AI tools), Hacker News comment, published Apr 25, 2026
  91. Comment on: Claude Code to be removed from Anthropic's Pro plan?mark_l_watson, Retired developer and author (self-described: Personal open-source projects), Hacker News comment, published Apr 22, 2026
  92. Comment on: Our eighth generation TPUs: two chips for the agentic eraWarmWash, Hacker News commenter (self-described: Personal tax return test and small greenfield projects), Hacker News comment, published Apr 22, 2026
  93. Prompt Injection leads to RCE and Sandbox Escape in AntigravityDan Lisichkin, security researcher at Pillar Security (Pillar Research team testing agentic IDEs for sandbox bypasses), Security research write-up, Pillar Security blog, published Apr 20, 2026
  94. Antigravity servers down for over 7 hours!Ill_indication, Freelance developer building for clients (self-described: Client work), Forum post, published Apr 15, 2026
  95. Using OpenCode as a fallback agent for AntigravityAlexander Tyutin, automation practitioner (self-described: Daily delivery work), Personal blog post, dev.to (Google Developer Group), published Apr 15, 2026
  96. Comment on: Pro Max 5x quota exhausted in 1.5 hours despite moderate usageAchshar, Hacker News commenter, Google AI Ultra subscriber (self-described: Personal development), Hacker News comment, published Apr 12, 2026
  97. Comment on: Pro Max 5x quota exhausted in 1.5 hours despite moderate usagerzkyif, Hacker News commenter, Google AI Pro subscriber (self-described: Personal ~15k line codebase), Hacker News comment, published Apr 12, 2026
  98. Comment on: Pro Max 5x quota exhausted in 1.5 hours despite moderate usageasdfasgasdgasdg, Programmer of twenty years (self-described: Personal embedded motor-control project), Hacker News comment, published Apr 12, 2026
  99. Antigravity: My Approach to Deliver the Most Assured Value for the Least MoneyAlexander Tyutin, automation practitioner (self-described: Delivery workflow with quality gates), Personal blog post, dev.to (Google Developer Group), published Apr 1, 2026
  100. Google Antigravity: Secret Quota Cuts, Break Trust & Make It Unusable for Productionaaron_wu, Aaron Wu, long-time paying user (self-described: Production work, upgraded from Pro to Ultra), Forum post, published Mar 26, 2026
  101. [Ultra] Dramatic quota reduction after update: this needs an official explanationMariano Abad, Google AI Ultra subscriber (self-described: Several agents running complex tasks in parallel), Forum post, published Mar 23, 2026
  102. Testing Antigravity: Building a Data-Intensive POC at 300km/hPatricio Navarro, data engineer and Google Developer Expert (self-described: GCP data pipeline prototype), Personal blog post, dev.to (Google Developer Experts), published Mar 22, 2026
  103. Comment on: Google Antigravity IDE rug pulls its quotasriversflow, Hacker News commenter, Google AI Pro subscriber (self-described: Personal building on a Pro subscription), Hacker News comment, published Mar 11, 2026
  104. Comment on: Agentic Engineering Patternsjpadkins, Hacker News commenter (self-described: Hobby game development), Hacker News comment, published Mar 5, 2026
  105. Comment on: Agentic Engineering Patternsmaccard, Hacker News commenter (self-described: Personal evaluation), Hacker News comment, published Mar 4, 2026
  106. Comment on: Beta Player: unofficial Bandcamp desktop and mobile player with remote controleremef, Hacker News commenter (self-described: Solo open-source app, a month of full-time work), Hacker News comment, published Mar 3, 2026
  107. Comment on: I don't know how you get here from "predict the next word"measurablefunc, Hacker News commenter (self-described: Personal Rust compiler and VM project), Hacker News comment, published Feb 26, 2026
  108. Adding Capacitor to Glif with Antigravity: The Good, The Bad, and The Reality CheckDomenico Tenace, open-source developer (self-described: Turning a Nuxt app into a mobile app), Personal blog post, dev.to (Playful Programming), published Feb 26, 2026
  109. Unacceptable Antigravity Quotas for Gemini 3.1 Pro: Workflow Completely Blocked!Abisheak_Shill, Abisheak Shill, Google AI Pro subscriber (self-described: Next.js redesign for a client project), Forum post, published Feb 21, 2026
  110. Comment on: Ask HN: Are you using an agent orchestrator to write code?nprateem, Hacker News commenter (self-described: Solo on a 500k+ line codebase), Hacker News comment, published Feb 13, 2026
  111. Comment on: Zed editor switching graphics lib from blade to wgpumercutio2, Hacker News commenter (self-described: Personal feature work), Hacker News comment, published Feb 13, 2026
  112. Comment on: Claude Code is suddenly everywhere inside MicrosoftTheAceOfHearts, Hacker News commenter (self-described: Personal agentic coding), Hacker News comment, published Feb 2, 2026
  113. Google AI Pro Subscription: Antigravity Quota Not Working as Advertised (10-Day Lockout Instead of 5-Hour Reset)SMTH_Gaming, SMTH Gaming, Google AI Pro subscriber (self-described: Code analysis of a local repository), Forum post, published Jan 25, 2026
  114. Google Antigravity vs GitHub Copilot: Why I'm Team Antigravity NowDomenico Tenace, open-source developer (self-described: Refactoring their Easy Kit Utils library), Personal blog post, dev.to (Playful Programming), published Jan 23, 2026
  115. Comment on: Ask HN: Do you have any evidence that agentic coding works?defatigable, Hacker News commenter (self-described: Personal Chrome extension; uses Augment at work), Hacker News comment, published Jan 21, 2026
  116. Comment on: Ask HN: Do you have any evidence that agentic coding works?Kerrick, Hobbyist Rubyist (self-described: Solo open-source library, built in public), Hacker News comment, published Jan 21, 2026
  117. Antigravity broken: getting only "Agent execution terminated due to error."Etienne, Google AI Ultra subscriber (self-described: Daily development on macOS), Forum post, published Jan 15, 2026
  118. Vibe Mapping: Google's Antigravity is pretty damn amazingZachary Flower, developer (self-described: Hobby tool for a MUD codebase), Personal blog post, flower.codes, published Jan 15, 2026
  119. Comment on: What happened to WebAssemblyneomantra, Hacker News commenter (self-described: Open-source MCP server project), Hacker News comment, published Jan 9, 2026
  120. AudioNoise READMELinus Torvalds, creator of Linux (self-described: Personal audio effects project), Project README, GitHub, published Jan 9, 2026
  121. Comment on: How Google got its groove back and edged ahead of OpenAImythz, Hacker News commenter (self-described: Solo, full-time development), Hacker News comment, published Jan 8, 2026
  122. Comment on: How Google got its groove back and edged ahead of OpenAIjchw, Hacker News commenter (self-described: Personal Rust projects), Hacker News comment, published Jan 8, 2026
  123. Comment on: Ask HN: Any example of successful vibe-coded product?alwinaugustin, Hacker News commenter (self-described: Solo, shipping small SaaS apps), Hacker News comment, published Jan 4, 2026
  124. Comment on: Ask HN: Any example of successful vibe-coded product?contractorwolf, Software engineer (self-described: Solo, consumer app MVP), Hacker News comment, published Jan 3, 2026
  125. How Google’s Antigravity IDE Changed My Perspective on Small-Team Software DevelopmentSteve Sink, VP of Technology at Just Landed Jets (self-described: Private jet charter broker with a small team and one lead developer who deploys internal models), Company blog post, Just Landed Jets, published Dec 23, 2025
  126. Antigravity feels heavy and Claude Skills are lightPiotr Migdał, Quesma (self-described: Side-by-side test against Claude Code), Company blog post, Quesma blog, published Dec 16, 2025
  127. Comment on: AI agents are starting to eat SaaSscotty79, Hacker News commenter (self-described: Solo, one-evening tool build), Hacker News comment, published Dec 15, 2025
  128. AI IDE prompt Injection: Reading browser local storageManish, independent security researcher (self-described: Personal security testing), Security research blog post, introvertmac.wordpress.com, published Dec 15, 2025
  129. This blog is now on Astro thanks to Antigravity!Philipp Hartenfeller, Product Lead at United Codes (self-described: Personal blog migration), Personal blog post, hartenfeller.dev, published Dec 8, 2025
  130. Google Antigravity Security: The Zero-Click Exfiltration Google Won’t FixAaryan Bhujang, AI security researcher at Repello AI (AI security company red teaming coding agents), Security research write-up, Repello AI blog, published Nov 28, 2025
  131. Antigravity's Gemini ProblemAmit Gawande, developer (self-described: Standard agentic-IDE test against Claude Code), Personal blog post, Amit Gawande's Devlog, published Nov 26, 2025
  132. Google Antigravity Exfiltrates DataPromptArmor, AI security research team (self-described: Security team testing Antigravity against prompt injection), Security research write-up, PromptArmor, published Nov 25, 2025
  133. Forced Descent: Google Antigravity Persistent Code Execution VulnerabilityAaron Portnoy, Chief Product Officer at Mindgard (AI security company testing Antigravity's workspace trust model), Security research write-up, Mindgard blog, published Nov 25, 2025
  134. Comment on: Nano Banana Proelevatortrim, Hacker News commenter (self-described: Company weighing a move from Claude to Gemini for agentic coding and employee use), Hacker News comment, published Nov 21, 2025
  135. Using Antigravity for Statistical Physics in JavaScriptChristopher Krapu, data scientist (self-described: Personal visualization test), Personal blog post, christopherkrapu.com, published Nov 21, 2025
  136. Comment on: Google Antigravitysega_sai, Hacker News commenter, Gemini Pro subscriber (self-described: Personal trial), Hacker News comment, published Nov 19, 2025
  137. Comment on: Building more with GPT-5.1-Codex-Maxtimtimmy, Hacker News commenter (self-described: Evaluating Antigravity for a team codebase), Hacker News comment, published Nov 19, 2025
  138. Comment on: Google AntigravityYeroc, Hacker News commenter (self-described: Personal trial; only Claude Code is approved at work), Hacker News comment, published Nov 18, 2025
  139. Comment on: Google AntigravityTopfi, Hacker News commenter (self-described: Personal trial on launch day), Hacker News comment, published Nov 18, 2025
  140. Google AntigravitySimon Willison, independent developer and blogger (self-described: Personal open-source plugin work), Personal blog post, simonwillison.net, published Nov 18, 2025

Nov 2025 to Oct 2026.

123 sources from practitioners, by month and sentiment.

The full readExplore all Google Antigravity Signals7 praise themes, 7 concerns, 3 points of agreement and 3 disagreements across all three groups, with all 140 sources open to inspect.

Where It Fits

Where does Google Antigravity belong in the stack?

Open Where It Fits: the Stack Field and the full map

Google Antigravity is Core across Build, supports Plan and Validate, and touches Ship and Operate.

Business area
Product & Engineering
Function
Software Development
Operating stack
Agentic Software Development
Primary role
Agent-First Development Platform
Google Antigravity beside the rest of its stackThe Stack Field, every product on every layer
FootprintGoogle Antigravity across the Agentic Software Development stack. Bar height is its role on each layer.

  1. Plan

    1. Work planning: Supports

      Work planning, supports. Agents write an implementation plan and task list before they build, and /plan in the CLI produces a plan artifact to approve; planning stays in the team's tracker.

  2. Build

    1. Codebase context: Supports

      Codebase context, supports. Agents read and search the workspace, a /codesearch command searches code, and rules, GEMINI.md, AGENTS.md, and skills add project context; a dedicated code-intelligence platform still owns cross-repository indexing.

    2. Coding: Core

      Coding, core. This is Antigravity's primary stack role: agents write and change code across a workspace, with Gemini 3.8 Flash by default and Claude models on eligible plans.

    3. Agent runtime: Core

      Agent runtime, core. Antigravity 2.0 is a standalone app for launching and monitoring several agents, subagents, and scheduled tasks. The CLI, the IDE, and the editor extensions give the same agent a terminal and an editor.

    4. Tool execution: Core

      Tool execution, core. Agents run terminal commands in an OS-level sandbox, call MCP servers and plugins, and drive Chrome, each governed by Deny, Ask, and Allow permission rules.

  3. Validate

    1. Debugging: Supports

      Debugging, supports. Agents trace failures from terminal output, logs, and the browser, but Antigravity is not a production debugging or error-tracking tool.

    2. Testing: Supports

      Testing, supports. Agents run tests and use the browser to click through a web app, saving screenshots and recordings as artifacts; the team's test suites and CI stay the gate.

    3. Code review: Supports

      Code review, supports. Walkthrough artifacts summarize each change and reviewers can leave comments on any artifact for the agent to act on; pull request review stays with the team's review tool.

    4. Application security: Touches

      Application security, touches. Gemini Enterprise administrators can restrict browser URLs, MCP servers, and file access, but Antigravity does not scan code for vulnerabilities as an application-security platform.

  4. Ship

    1. Source control: Touches

      Source control, touches. Antigravity 2.0 includes Git version control and worktrees for parallel agents, but repositories, branches, and merges live in the team's source-control system.

    2. CI/CD: Touches

      CI/CD, touches. Headless CLI runs with agy -p support scripts and CI jobs, but Antigravity is not a delivery platform.

  5. Operate

    1. Observability: Touches

      Observability, touches. Agents can investigate an incident from logs and MCP-connected tools, and Gemini Enterprise can log prompts and responses, but Antigravity does not provide observability.

Market mapThe layers Google Antigravity covers, and who else owns each one.

  1. Coding & implementation

    CoreBuild

  2. Agent interface & developer runtime

    CoreBuild

  3. Tool execution & integrations

    CoreBuild

  4. Work planning & task decomposition

    SupportsPlan

  5. Codebase context & intelligence

    SupportsBuild

7 more layers on the full map. Core, Supports and Touches describe coverage. They say nothing about quality.

Models

Which models run in Google Antigravity?

The current releases GILD tracks that public sources record running in Google Antigravity. Each model's page charts its public benchmark scores.

  1. Gemini 3.1 Pro PreviewA preview Gemini Pro release for advanced reasoning and multimodal work.Gemini ProBenchmarks
  2. Gemini 3.8 FlashA Gemini Flash release for long-horizon software engineering, agents, and enterprise workflows.Gemini FlashBenchmarks
  3. Gemini 4 ArgonGoogle's Gemini 4 Argon release for advanced reasoning and multimodal work. Google announced it on September 30, 2026 as a limited rollout to cyber defenders in its Fairwind Program; it has been generally available on the Gemini API and Agent Platform since October 13, 2026.Gemini ProBenchmarks
  4. gpt-oss-120bOpenAI's largest open-weight model, 117B parameters with 5.1B active, with a 131,072-token context window. Weights under the Apache 2.0 license; OpenAI lists no API price for it, so it runs on your own hardware or a host's.gpt-ossBenchmarks

More on Google Antigravity

  1. GILD ReviewGILD's evaluation, with the buying facts
  2. Research3 pieces
  3. Guides4 guides
  4. Operator Insights1 operator account
  5. Where It FitsOn the Agentic Software Development stack
  6. Related Products8 products