I am currently subscribed to the AI Ultra for Business plan. I am writing to report a discrepancy between the stated usage policy and my actual account limits.
1. Explicit Exemption for Ultra Users In early January, the Antigravity team announced the implementation of weekly usage limits for the Pro plan but explicitly stated that the Ultra plan would be exempt.
According to the official post on X (formerly Twitter):
“As we balance giving the best possible quotas and maintaining fairness between users, especially under incredible demand, we will be establishing generous weekly limits for all models. This will only affect a minority of Google AI Pro users. These limits do not apply to Google AI Ultra, which continues to be the best plan for power developers!”
2. Current Issue Despite the statement above, I am currently facing a usage limit with a reset timer of “1d+”. A few days ago, I even encountered a “4d+” reset timer, which briefly reverted to the standard 5-hour cycle but has now returned to a multi-day lockout.
3. Impact on Workflow This limitation is severely impacting my development workflow. During the lockout, I attempted to use Gemini 3 Pro and Flash as fallbacks. However, these models failed to adhere to my specific instructions, repeatedly making unauthorized and out-of-scope code modifications. This forced me to rollback a full day’s worth of work.
Given that the official policy states Ultra users are exempt from these weekly caps, could you please clarify if this is a system error or an unannounced policy change?
The situation has escalated. I just received another restriction without any warning, and this time it is a full 7-day lockout.
The UI currently displays this message:
“Model quota reached. Your plan’s baseline quota will refresh on 3/3/2026, 9:13:41 AM. You are subscribed to the best Workspace plan.”
It is incredibly frustrating that the system explicitly acknowledges I am on the highest tier (“best Workspace plan”), yet still enforces a week-long restriction. This completely contradicts the stated policy that Ultra users are exempt from weekly limits.
This multi-day lockout makes the service practically unusable for ongoing professional work. Is there any official update or a fix being worked on for this bug?
I’m not exactly sure what just happened, but all models have suddenly refreshed and are now fully available to use again.
Could it be that the system temporarily misidentified my account as being on a Free or Pro tier? I didn’t even restart my IDE. Right after posting my previous comment, I checked the newly updated model usage menu (from the first screenshot) and saw that my quota had completely reset.
While I’m glad the issue is resolved for now, the root cause remains completely unknown. If this is indeed a system bug affecting Ultra tier recognition, I really hope it gets patched quickly to prevent these random lockouts.
Another Update: The issue is highly unstable and keeps recurring.
It keeps bouncing back and forth between working and locking me out. I have attached new screenshots that show a clear bug and inconsistency within the system:
1) Settings Menu (Model Quota): As you can see, the settings menu shows that I still have quota remaining for the models, and it indicates they will refresh in 19 minutes.
2) IDE Chat Window: At the exact same time, the IDE chat window throws a “Model quota reached” error and enforces a 7-day lockout (refreshing on 3/3/2026).
The system is completely contradicting itself. This level of instability makes it absolutely impossible to get any actual work done.
Could the team please look into why the IDE is falsely triggering a 7-day restriction while the backend/settings clearly show available quota?
Today 24 February 2026, we all got the Weekly limit on Ultra Plan, when I check in the same page as before they applied the Weekly Limit on Ultra plan !
Gemini 3.1 model has reach the weekly limit too even I just used when Opus 4.6 is reach the limit by today ! It’s abnormal !
I hope this is only BUG and should be fix ASAP, it’s really annoyed and doesn’t make any sense !
Honestly, it doesn’t make much sense to change a policy without giving us a heads-up. Usage caps are a huge deal, so shouldn’t there be some sort of user consent? Or at the very least, shouldn’t they honor the original terms until our current billing cycle ends since we subscribed under the old policy?
Also, it’s hard to just write this off as “you used up your quota.” For one thing, my usage isn’t even that heavy. And even if it were, getting slapped with a 7-day lockout implies that I somehow burned through an entire week’s worth of quota in just a few hours, which is completely unrealistic.
If you look at my Settings > Model Quota, I clearly still have usage left, and it’s doing the normal 5-hour refreshes. Plus, when I run commands in the IDE, they actually execute perfectly fine. But then, completely at random, this error message just pops up. If I had actually hit a hard limit, shouldn’t there be a warning right there on the chat model selection screen, and shouldn’t the settings menu show that same multi-day restriction? Neither of those is happening.
At this point, I strongly suspect there’s some kind of internal conflict going on within Antigravity. The UI is clearly displaying that I have an Ultra account, but it feels like one of the internal usage monitoring logics is mistaking me for a Free tier user and keeps tripping me up.
Complete conjecture on my part, but reading between the lines, I suspect they’re doing something a bit like Uber’s demand pricing, and that’s why the terms lists nebulous things like “highest, most generous.” When things get busy the rate limit ceilings get lowered.
And yeah, the UX is so horrible on the model menu. From a bit earlier today:
Looking at that and that alone, I’d assume it’s saying that I have some Claude/GPT quota left, but it’s getting low, which is why the bar’s yellow. Except that wasn’t the case. My 5h quota was exhausted, and I had to wait the 2h 21 min.
But where can I view the weekly quota? How much of that do I have left? How much of either quota was consumed by the action that I just took? How can I improve the efficiency of my usage if I have to go hunting for this information.
Frustratingly, it now shows 162h and 35m until the next Claude/GPT refresh.
I was so close to upgrading from Pro to Ultra earlier today because of the annoying 5h quota. Now after looking at the recent silent policy changes I just can’t bring myself to do it - especially with the utter lack of comparative information about how much additional usage I’ll get from upgrading.
Even with the best plan, the agent stops responding up to 50 times a day. I have to stop and restart Antigravity, but with the quota, it’s really too much. The service is not stable at all; I’m completely stuck… even after changing models!
I can’t even continue my work due to these repeated, random “quota reached” errors, despite the settings clearly showing I have usage left. I’ve actually had to switch over to my personal account temporarily just to get things done. I really hope a fix is deployed soon.
I need some clarity on this: are we actually hitting a hidden weekly limit, or is this purely a system bug? If there really is a weekly limit being enforced, there absolutely needs to be a progress bar for weekly usage in the Model Quota menu so we aren’t left in the dark.
Also, don’t Gemini and Claude have separate quotas? I barely used Gemini at all. I only switched to it for a few hours after Claude locked me out, but now Gemini is also intermittently halting with the exact same 7-day limit message. It’s honestly hard to believe that I could burn through an entire week’s worth of Gemini quota on an Ultra plan with just a few hours of work.
My best guess is still that the system is mistakenly identifying my account as a Free tier user, which keeps triggering this 7-day reset loop. Whatever the root cause is, this bug is making it completely impossible to do any actual work.
The model quota screen in Antigravity is bugged. It shows full quota for me but gives me quota reached when I use Claude models. When I restart, it will show them as empty in the model quota screen, but if I click to another setting page and back they are full again with refresh of 4 hours.
This needs resolution and there needs to be a place to easily see how many credits I have and am using in each thread. These bars aren’t informative enough.
When I (and many others) paid for the Ultra plan this month, we entered a transaction based on the explicit promise of “No weekly rate limit,” as clearly shown in the official documentation from February 4th. Google took our money for that specific tier.
Silently changing the terms to “The highest weekly rate limit” right in the middle of our paid billing cycle—and immediately bricking our accounts for a week—is an extremely serious problem. It is deceptive.
If Google realized their infrastructure couldn’t handle the “unlimited” promise, the ethical or correct approach would have been to apply the new limits only at the start of the next billing cycle for each user, or offer an immediate prorated refund. You cannot retroactively change the fundamental features of a product after the customer has already paid for it.
You need to roll this back for current subscribers or issue automatic refunds immediately.
Just downgrade the version, It’s working fine in older version, although you will not get Gemini 3.1 pro but Claude models are working fine with 5 hours refresh limit. Try it will work.
