My credits were completely exhausted even though I didn’t use GPT-o1, Claude Opus, or Gemini 1.5 Pro. This contradicts the promise of credit renewal every 5 hours for Pro users, making it impossible to continue my projects. I chose AntiGravity over options like Cursor and Windsurf, but this limitation is a major setback. At the very least, the Flash model should remain available throughout the month. If this isn’t resolved, I’ll have to reconsider my tools and potentially move back to Cursor.
I had the same problem; it only lasted about 2-3 uses before the limit was reached. The waiting time was about 5 days.
1 Like
Same problem here, with a PRO account. This is very frustrating and makes me consider using Antigravity.
The credit system seems terrible to me; credits are practically consumed immediately even with optimized prompts.
2 Likes
I waited for over 7 days expecting it to be reset, but the wait was extended to another 4 days and 20 hours.
This is a PRO plan, but I cannot use it at all. Shouldn’t I be given a refund?
3 Likes
As everyone else here, I join the “I’m restricted to Gemini 3 Flash since a week” club.
Today was supposed to be the day I got my Gemini quota back, but it just reset to 0 again, withouth any usage.
1 Like
Hello @chunduriv.
This does not resolve the underlying technical issue or change the facts on the ground. As we have already pointed out, framing the documented 140+ hour system lockouts as an “evolution” of the plans in order to force users to buy additional credits at API prices is completely unacceptable.
We want to officially reiterate that silently altering the terms of an active subscription (removing the “5-hour refresh” promise from your documentation)
The developer community will continue to systematically document every quota bug, every phantom date shift, and every one of these evasive forum responses. We require your engineering team to deploy a structural fix for this bug
3 Likes
I have a wild theory as to why Google would suddendly stress everybody on their ai usage limits.
I think Google would like everybody to break the habit of using only the most expensive premium models e.g. Gemini 3.1 Pro (High), Claude Opus 4.6 (Thinking), etc. and try using Gemini 3 Flash for a change.
Gemini Flash works surprisingly well and it’s enough to handle a lot of my needs. However, I have a habit of using only the premium models. Why use lesser models at all if premium models are available and the rate limits are not a problem?
I have decided to cancel my pro membership and I would implore everyone else to do so as well. No matter their motivation: The lack of response on Google’s side to multiple threads of countless people experiencing a breaking issue, is shocking. The offered “solution” is just another money grab and the handling of this entire situation disrespectful towards the paying customer.
I am disappointed, since I really enjoyed Antigravity in combination with Gemini, but this has me switching to the competition.
2 Likes
Yes same problem so the best advice is to cancel the Pro Subscription and change to Claude at least they shows transparently how the quota is working in their models.
2 Likes
Google’s pro subscription is really hilarious. Before even using it, my Claude subscription showed a refresh every 4 hours, and after one cycle, it changed to a 7-day interval. I really can’t understand Google’s thinking process. OpenAI and Anthropic are both increasing the usage of their platforms, but Google is trying to limit the users. This is simply beyond comprehension!!! I will cancel my Google subscription. I suggest that friends of Vibe Coding switch to OpenAI or Anthropic as soon as possible.
I am experiencing the exact same issue. I’ve been emailing support for over a month now, and I haven’t received a single helpful solution.
It is honestly disappointing to see such poor customer service for a professional platform. Because of this, I have officially canceled my PRO subscription. They will need to provide a serious explanation and a permanent fix before I even consider resubscribing. The lack of attention to this matter is unacceptable.
Very frustrating, a whole wasted week, literally. Last week I worked for under 1.5 hours and not continuously and I was hit with the message of multi days lockout, don’t recall exactly but believe it was like 5 days. I have done way way more prompting and chatting the two weeks earlier and was just fine and would get hit with a few 5 hours refreshes. A little under the 5 days I checked and saw I only have to wait 10 more hours before the multi days is over. Next day I saw it change from 10 hours to two days or so with no work what so ever. !!! Seriously, with every one complaining, why isn’t google officially respond if they are 1# Working on a known bug or 2# dropped the 5 hours refresh rate and will make it weekly refresh !!!
2 Likes
[URGENT BUG] Google AI Pro Quota Sync Failure – Incorrect 48h/120h “Baseline” Lockout (Ref: AG-859)
Issue Summary: I am a Google AI Pro subscriber (active via the Jio 18-month promotional offer). While my subscription is correctly recognized on gemini.google.com (showing the Gemini Advanced badge and 1,000 monthly credits), the Antigravity IDE is failing to sync this entitlement.
Instead of the advertised 5-hour Pro refresh cycle, the IDE has erroneously placed my account into a “Weekly Baseline Lockout” of 4+ days (formerly 5 days). This is preventing me from using Gemini 3.1 Pro and Claude models for my development work.
Environment Details:
-
Account Tier: Google AI Pro (Jio Managed)
-
IDE Version: Antigravity v1.20.5
-
Operating System: [ Windows 11 ]
-
Affected Models: Gemini 3.1 Pro (High/Low), Claude 4.6 (Thinking), GPT-OSS 120B.
Observed Behavior:
-
Tier Misalignment: The IDE shows a yellow warning icon next to all Pro models with the message: “Your plan’s baseline quota will refresh in 2 days, 2 hours.”
-
Incorrect Limit: A friend with the same Jio Pro plan is correctly receiving a 5-hour refresh, confirming this is an account-specific sync bug (likely related to AG-859).
-
Usage Context: As a Full-Stack Developer (Laravel/React), I am currently blocked from critical integration work on the E-KYC projects.
Troubleshooting Already Taken:
-
Logged out and re-authenticated via Google OAuth multiple times.
-
Cleared local IDE cache and browser cookies.
-
Verified that Gemini 3 Flash correctly refreshes every 5 hours, while the Pro models remain locked behind the multi-day “Free Tier” timer.
-
Confirmed that 1,000 AI Credits are available in my Google account, but the IDE ignores them unless I manually toggle “Overages,” which shouldn’t be necessary for the standard Pro refresh.
Requested Action:
-
Please perform a Manual Quota Reconciliation for my account to restore the 5-hour “Priority” refresh.
-
Force a backend re-sync between the Jio/Google One entitlement server and the Antigravity metadata database.
-
Confirm if there is a known latency issue with Jio-managed accounts regarding the March 2026 “Baseline” quota rollout.
1 Like
Same issue here:
I dont used antigravity the whole week, then used up the tokens in one 5h session and got locked out for multible days. I realy like antigravity, but this is not usable for me, when this dont changes I will cancel my pro subscription.
2 Likes
Which models are you referering too?
Its not a bug this is how gooogle will manage quota from now.
I’ve moved my workflow to Claude and Cursor, and seeing this happen is frustrating. The most bizarre part is facing a full weekly lockout even though I haven’t even initiated any requests. There is a clear discrepancy between my actual usage and what the system reports. Google needs to investigate if idle sessions or background syncs are incorrectly eating up the Pro quota.
It’s not a bug everyone. Their documentation has changed since they launched. It used to be 5 -6 hour refreshes. Now it’s 5 - 6 hour refreshes until you reach your weekly limit. Then you have to wait until the next week.
