TL;DR: Google’s agent-first IDE, Antigravity, launched with massive promises (multi-model support, deep CLI/browser integration, and reasoning modes) along with a 5-hour quota reset. However, it is currently locking AI Pro users out for up to 7 days, with Google instead offering a gentle nudge to upgrade to their $200/month Ultra plan.

Google advertised AI coding with 5-hour reset limits in its agent-first IDE, Antigravity, but those who signed up for the AI Pro plan have been met with a reality that starkly contradicts those promises.

The Grand Promise vs. The 120-Hour Reality Check

When Google first launched Antigravity, they promised a revolutionary, seamless AI coding experience. It wasn’t just billed as a smart auto-complete tool; it was designed to be a full-fledged agentic workspace. Out of the gate, developers were promised the following:

  • A Multi-Model Ecosystem: The freedom to choose from a variety of top-tier models, including:
    • Gemini 3.1 (High & Low)
    • Gemini Flash
    • Claude Sonnet and Opus
    • GPT OSS models.
  • Deep Tool & Workspace Integration: Agents that hook seamlessly into most VSCode features, giving them the ability to review code, edit files, manage version control via Git, and execute commands directly in the CLI.
  • Advanced Agentic Capabilities: The power to launch a browser, autonomously interact with web pages, and even generate images directly within the workflow.
  • Flexible Execution: A toggle between a deep “Planning” (reasoning) mode for complex architectural problem-solving and a “Fast” mode for rapid iterations.

Many users could run 2-4 agents at the same time concurrently and rarely run out of credits. Combined with an advertised 5-hour reset limit for quotas, it was positioned as the ultimate daily driver for developers.

Around this time, in the early days of Antigravity, Claude Code and GitHub Copilot users (competitors to Antigravity) were always complaining about quota limits. It was so terrible that some were reverting to the old era of copy-pasting code. Even more terrible, it was said that some had reverted to typing code manually in the editor line-by-line like the ancient developers. These conditions seemed so primitive to Antigravity users.

Judgemental Volturi looking down whilst dressed in royal robes symbolizing antigravity users

In those days, Antigravity users looked down with disdain on these outdated developers who didn’t have an agentic IDE at their disposal … until the tables turned. Right now those who actually signed up for the AI Pro plan are facing a very different reality.

Instead of a quick 5-hour reset, users are getting hit with lockouts lasting 120 hours (~5 days), and in some cases, a full 7 days.

AI Pro Plan

.

Cover image for post titled Many Google AI Pro Users Are Locked Out of Antigravity IDE

Now, Google’s suggestion to resolve the issue is to move up to the $200 AI Ultra plan, a move considered by many to be a slap in the face. Unsurprisingly, users are hesitant to upgrade. Instead, many are weighing whether to downgrade back to the free tier or cancel their monthly and yearly subscriptions entirely.

One disgruntled user on the Google Dev forums put it bluntly:

“I no longer use those Antigravity agents, only my Claude subscription. I do not want to feel like I am having a stroke every day because of unfair terms and the complete lack of communication.” (google.dev user)

Included below is a screenshot of the Antigravity model console for an AI Pro user, showing the exorbitant wait time for the usage limits to reset:

Screenshot of the Antigravity console showing usage limits

The Irony of the IDE Bug

To compound users’ frustrations, the IDE has an internal issue-reporting feature designed to capture logs and send feedback. Naturally, users attempted to use this to report the lockouts, only to find that the issue submission tool itself is broken. The irony is palpable.

Screenshot of the antigravity issue submission error, saying an error occured while submitting your feedback please try again

Will Upgrading Save You?

If you’re thinking about biting the bullet and upgrading to the highest tier, think again. According to current AI Ultra subscribers, the $200 price tag might not solve your woes either.

“I’m an Ultra subscriber, do not upgrade bro. Right now it’s the same situation with quota. No support, no announcements, just silent quota cut off.” (@YND)

Alternatives to Antigravity

We can’t let you go without providing a few working alternatives. If you are looking to jump ship, consider these tools:

  • Claude Code (Anthropic)
  • OpenAI Codex
  • Open-Source Tools: Cline, Aider, and Continue.dev

References

How long has this been going on? About three months. And yes, we brought receipts:

Video: Locked Out Featuring Claude Code and Copilot Users

For a perfect summary of the community’s current vibe, complete with Antigravity users being mocked by Claude Code and GitHub Copilot devs, this video from Richard Djarbeng’s channel captures it perfectly:

Media mentions:

Also posted on dev.to for Richard Djarbeng

P.S. The writer is fully aware of the irony that this post was formatted using Gemini 3.1 Pro from Google.

Related Concepts & Posts

The following cards provide further definitions of concepts discussed above, and possible links to related posts.

Antigravity

Agent-First Google

An agent-first, AI-native IDE from Google that features autonomous browser subagents and advanced coding agents for an agentic development experience.

Visit Antigravity

Claude Code

CLI Agent

Anthropic’s command-line interface for AI-native coding, offering a powerful alternative to integrated IDE agents.

Visit Claude Code

Cline

Open Source

An open-source AI assistant that can use your CLI and editor, providing a flexible and transparent coding experience.

Visit Cline

Aider

Popular

A popular command-line chat tool that lets you write code with AI, known for its efficiency and strong editing capabilities.

Visit Aider