oh.you are right. So it’s Just a week.
Please check the Models tab in the Antigravity settings.
There seems to be a bug related to account, capacity, or quota detection.
I suspect that some servers are unable to communicate with the microservice responsible for proper account detection, which may explain why the issue occurs randomly.
More details here:
https://discuss.ai.google.dev/t/bug-resource-exhausted-429-on-ultra-plan-despite-available-quota/126232/4
If you click Retry a few times, it starts working again…
I aways receive message that said Error,You have exhausted your quota on this model. And It will never restore quota after 5 hours. I can’t use gemini 3.1 and claude both. But In setting model quota, it still displays full quota. I need support. This problem is affecting my work, which makes me feel like I wasted my money subscribing to Google Ultra.
If this continues, what’s the point of my Google Ultra subscription? Should I switch back to Cursor? Every time I submit, it says my quota is exhausted. For Both gemini and claude model.
I am experiencing the exact same issue. It really feels like there’s some kind of internal conflict within Antigravity regarding account authentication and usage monitoring.
The system clearly recognizes that I am on the Ultra plan. If I check the Model Quota in the settings, it shows the normal 5-hour refresh cycle and indicates that I still have usage left. I can also select models from the dropdown without any warnings, and when I actually send a command, it executes perfectly fine.
But then, completely at random, it throws a “quota reached” error and halts the process. If I send another command right after, it works again—only to randomly spit out the quota error and stop later on. My best guess is that Antigravity has multiple layers of usage monitoring logic running in the background, and one of them is mistakenly identifying my Ultra accounts as the Free tier, which keeps interrupting my workflows.
To make matters worse, this issue isn’t just limited to the Claude models; it’s happening with Gemini Pro as well.
After Claude started becoming unstable, I temporarily switched over to Gemini Pro. Despite barely using it, Gemini also started cutting me off randomly mid-chat with the exact same error. And just like before, sending another command immediately after the error works perfectly, until the system randomly blocks it again.
This definitely confirms it’s a system-wide monitoring/quota sync bug, not just an isolated issue with one specific model.
retries don’t help anymore, so I assume they’re trying to figure it out
and are working on some fixes
Hi Everyone,
Thank you for bringing the issue to our attention. Our engineering team is currently investigating the matter, and we appreciate your patience as we work toward a resolution.
this is what I’m returning from yesterday night, I’m subscribed to AI Ultra plan. not able to use claude at all. it always return this message.
Same here on Workspace Business account with Ultra, all the sudden yesterday I get quota problems on Claude, even tho the Model tab shows 100% quota… and I can retry (I dont get the yellow triangle) I can try again and again but nothing…
When you combine that hypersensitive, bugged backend with an extension that is aggressively pinging the server every few seconds to check your token count, Google’s anti-abuse system flags it. The system sees this rapid, continuous polling, assumes it is massive automated “agent” activity, and instantly slams the emergency brake—giving you that absurd 155-hour cooldown.
everyone, remove the 3rd party quota viewers. that is the problem.
I have 1 ultra and 1 pro account , i just did an A/B test. and found that the quota viewer is the reason i have been stopped at ultra.
now in the pro account i have no limits . at 5hrs refreshal . no any problem.. i have been slammed with 170 hours waiting at ultra account just because of that weird third party quota viewer.