Issue Summary: I am a paying Google AI Pro subscriber, but the Antigravity IDE is completely ignoring my tier’s 5-hour refresh cycle. I am currently facing an 81-hour (3+ days) hard lockout on Gemini 3.1 Pro models.
Expected Behavior: According to official documentation, Pro tier users are allocated a high-capacity quota with a rolling 5-hour refresh window.
Actual Behavior: After normal workflow usage, my quota was exhausted and the system assigned a reset timer of 81+ hours (over 3 days) for Gemini 3.1 Pro. This looks like the system is erroneously applying the Free-tier weekly penalty to my paid Pro account.
Additional Context & Troubleshooting:
I have verified my Pro subscription is fully active.
The issue persists even if I log out, re-authenticate via the browser, and log back into the IDE.
I attempted to use a completely fresh, secondary Pro account, and it instantly threw the exact same “Quota Exhausted” server-side error, confirming this is a global MODEL_CAPACITY_EXHAUSTED failure, not a local user limit.
Lightweight models like Gemini 3 Flash are still operating normally on the correct 5-hour cycle.
I don’t think they care. If they did, the updates would be better, but instead, they’ve gotten worse. Antigravity IDE is a huge failure, it would have been better if they had just left it as it was. It got worse after every single update. They’ve forced me to go back to VSCode. To be honest, VSCode is much better than this IDE. Also, their current model is a complete disaster, it barely even works.
The issue seems to be that a whole lot of users decided to point their OpenClaw instances at their Gemini Pro subscriptions and Google is still wrestling with the onslaught. To be fair Anthropic is also showing signs of pushing back on the use of their Max plans with OAuth (vs. API use) to address the same issue.
As we balance giving the best possible quotas and maintaining fairness between users, especially under incredible demand, we’ve established weekly limits for all models. These limits do not apply to Google AI Ultra, which continues to be the best plan for power developers!
I have AI ultra and it just started telling me i have to wait a week to use the models again. So, something is definitely wrong here. Support has been useless, any ideas what to do? I don’t pay for ai ultra to wait a week to use it
I’m using claude opus 4.6.
In the chat it errors out and says quota exceeded, if i click retry it lets a few messages come out before stopping again, like maybe it’s hitting different servers and some are letting it through and some aren’t and when it hits the one that isn’t it stops.
It says:
Error
You have exhausted your quota on this model.
Your plan’s baseline quota will refresh on 3/2/2026, 3:58:41 PM. You are subscribed to the best Workspace plan. See plans.
And then vscode log:
2026-02-24 16:09:04.283 [info] E0224 16:09:04.283073 15482 log.go:380] agent executor error: RESOURCE_EXHAUSTED (code 429): You have exhausted your capacity on this model. Your quota will reset after 142h49m36s.: RESOURCE_EXHAUSTED (code 429): You have exhausted your capacity on this model. Your quota will reset after 142h49m36s.
Hey guys, I wanted to share something that when I downgraded the version of antigravity, everything is working fine. You can also try, It worked for me.
Apparently, with a 5-hour quota, there is a weekly quota too. So, if you reach your limit a second time, then it will restore after a week in the pro plan, and in 72 hours in the ultra plan.
Think youself lucky I have been off for over 10 days, tried all the usual plus 1 day with gpt and still nogo I also am a paying pro member! All I get is Error messages!
Thats not what I signed up for though, I specifically paid for Pro due to the 5h refresh. So are you going to refund us all based on how much less we are able to access the product now compared to when we started the subscription?
