I’m reaching out because I’m confused and frustrated with the current quota system for Gemini 3 Pro.
I recently switched from Windsurf to Antigravity after a very positive 1-month free trial. During the trial and my first few days as a paid user, my quotas were generous and refreshed every 5 hours, which was perfect for my workflow.
However, things have changed drastically. I started seeing people on Reddit and here in the forums complaining about wait times of several days, and I honestly thought it was a bug—until it happened to me. Now, I’m being forced to wait days to use Gemini 3 Pro again.
What’s strange is that Claude Opus still shows a 5-hour refresh, but Gemini (the model I actually want to use) is basically locked out for days. Why is this happening to paying subscribers? It feels like the service was downgraded the moment I started paying.
To make matters worse, I keep getting the following error mid-task:
“Agent terminated due to error. You can prompt the model to try again…”
I’m concerned that these crashes are consuming my credits/quota, and since I have to wait days for a refresh, every error feels like a massive waste of my subscription.
Is this a known bug or a permanent change in policy? I really like Antigravity’s UX, but if the quotas are going to be this restrictive compared to Windsurf’s refill system, I’ll have to consider switching back.
Any clarity from the team would be greatly appreciated.
That’s right. The quota’s too low now, so I’m considering switching to another paid AI. It reset yesterday, but it’s already showing me to wait until next week.
I wrote four prompts and it barely did anything, yet my quota went from 100% to 0%. It feels like the limit is extremely low. Now I have to wait five days.
It seems like with the rise of Opus 4.5 they pulled everyone onto the platform, and now we’re getting crumbs. It’s sad because a lot of people are frustrated.
today it was really bad at first i got on every task agent terminated error and then after this i hit by this 5days reset plan this is really bad for PRO plan user ….
Pro account, didn’t think it was an issue to have two accounts under same household, both accounts are paid Pro and 140h? woot almost 6 days?… Really Gee’? (Google)
Its weird being Gemini Pro user for … for surely up to 2 years or more, have active Workspace account, containers running, use Firebase alot.. Totally Google product that we are making sorta speak also we use various services on GCP.. and .. we get this?
Same problem here, there has been a sudden decrease of limits post the long downtime on the Agent terminated due to error? Google should state this policy change so that paid users are aware.
I have been so happy with Antigravity , compared to Codex and other tools that my colleagues have been using but unfortunately its back to square one now.
Hope to get clarity from the google team on the quota changes
Facing same issue. 5 hrs refresh changed to days with that multiple issue related to agent error. I just started my laptop after 10 hrs breal and this.
I’m or .. was thinking that; Is this some “crowed control” because users are using Claude models above Gemini? Which doesnt make sense if the user @ment in this thread is on Pro acount and getting 5+ hours reset window..
I started thinking this because I noticed alot of users are using the Claude models who are experiencing this jump in reset window… And I can speak for myself when I say; Ive mostly used Claude due to complexity of the work, and Gemini via Stitch, Jules, NotebookLLM etc.. Both have their strengths and weaknesses’, but could that be a hint? Users are using Claude above Gemini and therefore those users get longer reset window?
Shouldn’t that be announced and Google just needs to adjust to the consumer rate? Instead of having users confused and , complaining about a bought service…
We had barely been working this whole week me and my brother, and we did some work yesterday Friday, and boom… we both get above 5 hour reset window? (he saw 140h).
I wonder if having more than one instances of the IDE running might trigger some Bot Detection at Google? “This is almost certainly a Concurrency/Bot Detection Trigger, not a volume limit.” ← I asked Gemini if this could be because having more than one instance running…
Can we have an official answer about this by someone from google support? It’s unbelievable, i’ve updated antigravity hoping for a fix for the "agent error” and my reset is now 4 days?
Thank you for bringing this to our attention. We have escalated the matter internally for a thorough investigation. Any additional context or visual documentation you can provide would be greatly appreciated as we work toward a resolution.