The worst part for me was that I waited 40+ hours for the quota to lift, and then after immediately got 168h lockdown without even using one action.
I have of course tried to work with Gemini since we are forced to after this breach of contract. And honestly, it is confused, but mostly badly prompted. It is so annoying that it thinks its a one man show that plays god and jumps to the next step with no interaction all the time. Especially all mistakes and the arrogant tone.
Anyway. I can work with it on less important tasks. But this is just a rugpull and a thing we should build a class action lawsuit if not fixed soon.
Gemini is not stupid. It is very competent. It is just a narcissist.
2 Likes
What a shenanigans offering Claude models on Antigravity.. I wonder why it was available in the first place..
Or is Google investigating telemetry and traffics from the Claude models to update Gemini into a proper coder? 
Unfortunately it’s mostly a “trust me bro” kind of scenario, but it can also be derived by scouring the Antigravity Subreddit and there is at least one Google official team member who comments there occasionally; it has been mentioned that the reason they keep needing to tighten restrictions is because of capacity issues (“Agent terminated due to error” issues)
1 Like
Gemini can code. That is not the problem. The problem is that it is prompted to be a narcissist who does not listen to anyone else. I think the ones who made the prompt doesnt really know the craft, and just told it to ‘be the best, and solve all problems’.
But sure, it is easy to lash out after a rug pull, and it might be more to it
same here :
Well they sure take their times, I’ve been dealing with this since january 21
I just got off the phone with a Google representative and he told me that’s how it is and that ‘it depends on demand’, I don’t know, I’m going back to windsurfing.
@samiralibabic
Did you try downgrading? I think I read someone saying that in version 1.14.2 this problem doesn’t occur.
Done.
Does not change a thing.
I hope this made the trick for you, because I just tried and it made just nothing..
My2 cents: the issue appear clearly centralized on the account/plan and has nothing to do with IDE settings/version/etc..
I also made to reach out support few days ago and they only try to do some normal steps: uninstalling, login/out, re-check age, etc..
1 Like
Uninstall Antigravity download the lastest version 1.14 hit update wait a little and it will start working. mine is working now
Same issue and no reply from google
I don’t think it’s version-related. My limit was updated today after a 5-day block, and so far everything’s fine. It’s back to 100%, and there’s no cyclical transition to a new limit. Maybe they finally figured out the problem?
I am a subscribed Ultra user. I’m writing to report some severe stability issues that are hindering the workflow, along with a specific anomaly I observed that might help you debug the backend logic.
1. The “Agent Execution Terminated” Loop (The Core Issue) I am constantly encountering the “Agent execution terminated due to error” message when using Claude 4.5 Opus.
-
Symptoms: Every single step requires a “Retry”.
-
Contrast: This issue rarely happens when I switch to Gemini models.
-
Diagnosis: This strongly suggests a backend timeout configuration flaw. Claude 4.5 Opus (and high-reasoning models) has a longer Time-To-First-Token (TTFT) and generation time compared to Gemini. It seems your backend kills the process before the model finishes its CoT (Chain of Thought), treating it as a “hung” process.
2. A Strange Anomaly (Potential Debugging Clue) Here is a very counter-intuitive observation I made: When the Context Window is extremely large and the generation speed is visibly slow, the error actually triggers LESS often.
- Hypothesis: This implies your system might have a different timeout policy for heavy-payload requests, or the continuous stream of tokens from a large context somehow keeps the connection alive (heartbeat) better than a “thinking” state where no tokens are generated for a while. This inconsistency confirms the timeout logic is brittle.
3. Feedback on the Quota System (Community Pain Point) While I personally benefit from the Ultra quota, I’ve noticed significant frustration in the developer community regarding the rigid 7-day refresh cycle. The logic seems prone to glitches where the 7-day timer resets incorrectly, locking users out. This unpredictability is damaging the user experience for non-Ultra developers.
Proposed Technical Solutions:
-
Dynamic Timeouts: Adjust the backend
timeoutthreshold dynamically based on the selected model. Claude Opus should have a significantly higher tolerance than Gemini Flash. -
Heartbeat/Keep-Alive: Implement a “thinking” signal or heartbeat mechanism. If the model is processing but hasn’t output a token yet, the agent should not be terminated.
-
Soft Fails: Instead of terminating the whole agent, allow the frontend to wait longer if the backend reports “processing”.
I enjoy the tool when it works, but I currently have to rely on my own Rust-based workaround because the platform’s stability with SOTA models is unreliable.
Looking forward to a fix.
2 Likes
Yes, same issue for me.
This is kinda hilarious. Did Google vibe code this release? I’m on a Pro plan. Hit the rate limit yesterday on Gemini 3 Pro. My cooldown is until tomorrow evening. Not the five days. I must be lucky. I honestly can’t imagine how such a thing can slip
My main issue with this is that Google shows that they can not provide continuation. Thank goodness I am not depended on AI being available, but if I was, I would definitely not make Antigravity a vital part of my process. As a user, not having any form of communication from Google makes me very unsecure. I have no idea:
A. What caused this
B. What Google is doing about this
C. How long it will aprox take before this gets resolved
Not having the answers would just nudge me to switch to a different IDE
4 Likes
Still experiencing auth failures as of Jan 31. Any ETA on the billing migration fix? This is affecting paid Google One Pro and Ultra subscribers’ ability to use cloud features.
I understand this is in experimental phase, but a public acknowledgement would at least help people understand how much time they have to wait.
What’s particularly frustrating is I submitted a detailed bug report through the official channels (Because the customer support chat said so), and there has been zero acknowledgment, no email confirmation, no ticket number, no automated response. For a paid service (Google One Pro), the complete silence on bug reports is unacceptable. Even a simple “We’ve received your report and are investigating” would go a long way.
At minimum, could we get: A public status update (even if it’s just “still working on it”) An estimated timeline for the fix Or Confirmation that bug reports are actually being received and reviewed.
7 Likes
I experienced this issue as well. Here are my usage details:
Over the last 15 days, I have been on vacation and dedicated my time to learning more about IDEs and Agents. Using the OpenSpec lib, I managed to develop several features during this period.
I was blocked on Jan 29, 2026, and was only unblocked on Feb 02, 2026. I went back to using Antigravity to develop on Feb 04, and after drafting just a few plans with OpenSpec, I hit my limit again. Now, it says I will only be unblocked on Feb 09, 2026.