Google Ai Pro (Antigravity) Quota Refresh Failure - Instant 140h Lockout After Reset

Issue Summary: My model quota was scheduled to refresh today, March 13, 2026, at 12:28 PM. I monitored the countdown in the settings menu. However, immediately after the countdown reached zero and the “Refresh” was triggered, all high-tier models (Gemini 3.1 Pro and Claude Sonnet 4.6) were instantly locked again for another 6 days (140+ hours). I have not sent a single prompt since the supposed reset.

My Environment:

  • Account Type: Google AI Pro (active until Oct 8, 2026)

  • Tool: Antigravity Extension for VS Code / IDE

  • Region: Malaysia

Observed Anomalies:

  1. Ghost Exhaustion: The “Baseline Quota” appears to be consumed instantly by background processes (potentially background indexing) without user interaction.

  2. Weekly Cap Trap: Even as a Pro subscriber, the system seems to be applying a strict “Weekly Cap” that overrides the promised 5-hour refresh cycle.

  3. Sync Error: The UI shows a successful refresh timestamp, but the API backend remains in a “locked” state.

1 Like

Same only I lost access to the feature until appealing but the email for appealing never showed and the email the AI gave didn’t bounce back but also didn’t reply. AWESOME!

Same, also used overages but not really usable. One query took 140 credits and only have 1k. See my thoughts Usage credits are almost zeroed and can not use to get effective result indepth if you want

Hi all,

To provide you with greater control and a seamless path to scale, we are evolving our Google AI plans by introducing built-in AI credits that can now be utilized directly within Antigravity. Under this updated structure, our Google AI Pro tier is tailored for practical builders, offering generous limits for Gemini Flash alongside a baseline quota to experience our most advanced premium models. For developers operating at the highest scale who require consistent, high-volume access to our most complex models, Google AI Ultra serves as the ideal comprehensive solution.

Furthermore, we’ve designed these plans with maximum flexibility in mind; if you are on the Pro plan and require additional capacity or deeper premium access for intensive development sprints, you can seamlessly top up your AI credits to customize your limits and continue shipping your best work without interruption

Thanks for the comment, but this is very vague. And let me say again: this was not part of the deal when I signed up for a year subscription.

There are three major issues:

  • Pro was sold to me as “Pro”, as in Professional. Somewhere else on this forum a Google employee suddenly framed it as something for hobbyist. Now you frame it as “Practical builders’. Which one is it? How will Google frame it next week?
  • I can’t use my pro account. I keep being locked out for days. And the lockout doesn’t discriminate between professionals, hobbyists and practical builders
  • How can I trust Google that they will not bait and switch me again, reframing the Ultra plan as something for the enthousiast and the new “Omega” plan offers more generous limits?
1 Like

Some transparency on paid account usage would be the bare minimum.
I waited all week for my quota to refill according to the timer, only to see it jump back to another 7‑day wait. At this point, I fully expect the same thing to happen again next week.
On top of that, my Claude usage only reached 80%, and after a short task with Sonnet 4.6, it suddenly dropped by 20%.
Google, we need real answers, and not the usual “we’re making things better” response, because right now, you’re not.

1 Like

Good day. The problem is not with the limits themselves, but with the lack of transparency; people do not understand how to best use the limits, what they are, or how they are being consumed.

My cases:

  1. It shows that I have an available limit for Opus usage and the AI Credit consumption option is enabled - and these two parameters are being consumed simultaneously: both Opus limits and AI Credits. This is a bug.

  2. On the limits panel, it shows me that I have an available limit for the model—but in fact, I receive an error in the chat stating that the limit is gone and to wait 7 days. This is a bug.

Furthermore, it is completely unclear what the 7-day lockout affects, as users reasonably note that they haven’t sent a single message to the chat, yet a 7-day limit has already been set. This is a non-transparent provision of services that leads to frustration and refusal to use the tool…