As a Pro subscriber, I’m experiencing the same issue. My daily workflow primarily involves using the Flash model via the Gemini Code CLI—and even then, my usage is only around 200 requests for 2.5 Flash and about 70 for 3 Flash. I only use the Pro model and Claude occasionally for development or when I don’t feel like opening the web interface. I’ve only ever hit the limit once when I first started using Antigravity; once the cooldown ended, everything was back to normal.
For me, bypassing this issue is simple: I could just register multiple accounts and aggregate the free-tier APIs from AI Studio. Since the usage per project would be low enough, I wouldn’t run the risk of being banned, and I wouldn’t even have to pay.
If this is how Google treats its paying customers, then I think it’s time for me to switch to Anthropic.
If you are still getting this 403 error, you aren’t alone. I finally got a support agent on record admitting this is a systemic bug, but they are completely powerless to escalate it and keep sending me to enterprise paywalls. Meanwhile, my $250/mo Ultra subscription is still actively billing.
I just wrote up a timeline of this entire broken support loop over on the google_antigravity subreddit. We need to consolidate our voices there to force a DevRel PM to actually bypass this broken routing system. Don’t let them keep billing you in silence. Go find the thread and drop your trace IDs.
I am experiencing the exact same issue (Error 403 / Gemini disabled on Antigravity IDE). I sent an appeal email to gemini-code-assist-user-feedback@google.com several days ago, but I haven’t received any response or acknowledgment yet.
My development work is completely blocked right now. Could you please help escalate this or check the status of the backlog? Any help would be greatly appreciated!
They probably banned thousands of their most loyal users/subscribers and this ban forces people to go to the competition altogether. I would suggest them to improve their process and communication. Set deadlines, ban people for 72 hrs on first violation, 1 week on the second etc. Don’t just perma-ban WITHOUT warning. Not everyone is ill intentioned.
The reality of the situation is that Google turned on the flood gates
We just saw Anthropic disable their subscription for 3rd party, OpenAI is still playing nice bc they are “behind” but expect them to do the same soon. Google is heading in the same direction, subscription for our ecosystem and API(full price) for everything. Subscription calls keeps idle servers busy, and when there is too much demand, just stop serving the subscriptions.
Quota’s across all providers will continue to tighten, we are current in the “hook em” phase where they make us dependent on the project, price increases come later.
The reason for the abrupt ban was someone actually looked at $ coming in vs $ coming out; and correctly identified that bad actors were costing them ALOT of money in compute and degrading the service for people using their services normally.
There is limited parallelisation in their current product stack(Antigrav and Gemini CLI) but once people had access to the raw API; they could spin up 5-100 API calls under 1 sub at the same time. So 1 user consuming as if they were 5-100 people.
While I truly understand that actions needed to be taken against people abusing the ssystem - especially with the free accounts - I lost some trust here as some individual with a single AI pro subscription, who “just” tried out OpenClaw and OpenCode, and maybe also some quota monitors (which also used the login of course).
I’d love the Antigravity team to reconsider some decisions. I’ve been a paying customer for 5+ years now (Google One plan) and would’ve loved at least some warning before banning me.
The whole “who will win the AI race” is not just about having the first model. It’s also about trust. And I’ve lost quite some trust through how Google’s handling this. Banning people - okay. Not reacting to any feedback or emails for 1+ week for paying (!) customers? Sorry, that’s a no go.
Switching to my Github + OpenAI subscription in both OpenCode and OpenClaw now. Will probably just cancel my Google One subscription which luckily is only a monthly, if I cannot use Gemini CLI + Code Assist + Antigravity anymore. There’s no reason to keep it, if this is how paying customers are treated and I have to worry to get banned again.
Hi Guys, i have the same problem, i tried everything i found, the last thing i do was uninstall vscode and visual studio community.. i have no idea if was that or where i read something about vs code credentials but suddenly works, i can log in to antigravity again.
Hope it works for some of you
btw, i did a lot of more stuff, so maybe was a combination.. but the last thing before works was uninstall vs code
Has anyone tested this yet? I’m about to start testing and cleaning everything. After all, someone on Reddit recommended cleaning a few days ago. But it might be an individual thing.
I see this the same way. I hope they will fix it soon.
When I first read somewhere that it was against the TOS, I immediately stopped using Antigravity / Gemini CLI OAuth and disabled it. But in OpenClaws setup process it tells you nowhere that you risk getting a ban or anything similar. However I still got banned now, even though it was probably just a couple requests in one day or two.
Some people (especially on Reddit but also here) act like it’s super obvious and an active decision to violate the TOS in these cases like with OpenClaw or OpenCode. The reality is many people just want to try out these popular tools and get no hint of it being against Googles TOS while doing so. Additionally, most users do not fully read the TOS of the software they are using. That is just completely unfeasible in the time we live in, where you often have to accept similar agreements on a daily basis and nobody can convince me that they actually fully read the TOS of everything they are using.
Unfortunately, it’s not working for me. I’ve cleared everything possible, multiple times. Apparently some detail is escaping me, or maybe they lifted the ban on you themselves.
It’s quite upsetting that I am facing the same issue if logging in with is against terms of service then I honestly didn’t know but I don’t think that testing out the tools where you’re going to get to train your AI to be better the more people use it in similar tools especially considering that anthropic opus and sonnet perform way better in tools like openclaw then Gemini does data could be used to train Gemini better I’ve stuck with Gemini and bard even when everybody else will critics of it if I or anyone else you’re actually had malicious intent it’s very easy to open up a couple of thousand Gmail accounts and take the free API keys and put them on tumblers none of us have done that please restore our accounts.
In all honesty I was just testing out the thing to see why old people were so hyped about it after I realized that there was no persistent vector memory in it I lost interest if it’s such a crime to try out new AI tools fine
i am having the same issue and i havent even used open claw. i set up a google cloud project (billed!) as suggested by the antigravity agent to experiment with communicating with antigravity via telegram. so the api was billed and i also had the ultra plan! and i still get the elbow. raised tickets in antigravity, and cloud projects but no one replies. anyways cancelled ai ultra plan and i’m using Claude now.
almost as soon as I post them so they are definately “paying attention” guys