Hi team,
I just encountered this error indicating that I am locked out of the model for the next 72 hours!!
I am an Ultra plan subscriber, how is that even possible??
Can you please clarify why this is happening and resolve the issue?
Thank you
Hi team,
I just encountered this error indicating that I am locked out of the model for the next 72 hours!!
I am an Ultra plan subscriber, how is that even possible??
Can you please clarify why this is happening and resolve the issue?
Thank you
I think they are adding hidden quotas now. Even the AI Pro now have weekly limit, that I didnāt know (I thought the docs only mention about 5 hours cooldown timer)
Iām an Ultra plan subscriber and Iāve just been hit with a 72-hour lockout for Claude Opus 4.6 (Thinking), while the UI still says āYou are subscribed to the best plan.ā
There was no clear warning, no visible limit, and no explanation before this happened.
Seeing a 3-day lockout on a core model at the highest subscription tier is unacceptable.
Multiple users are reporting the same issue, so this looks like either a bug or an undocumented hard cap.
This needs immediate clarification. If such limits exist, they must be clearly documented and shown in the UI.
I honestly donāt understand why people are paying $250+ a month for AI Ultra when it still feels unreliable.
For less than half the price, Claude Max costs about $100 a month and gives a more stable coding experience with better practical limits. You can actually work for long sessions without random throttling.
Paying $250+ for something that still feels like a beta product doesnāt really make sense when cheaper options work more reliably. Not every expensive tool is automatically the better one, even if the marketing says so.
Absolutely agree with you, it is not even the first time I am getting issues with it but this time is different, they literally prevent me from doing any work for 3 daysā¦
It feel like a soft banā¦
⦠which give me ZERO reason to stay a customer.
I am honestly super disappointed, because the UX of Antigravity is very good, I liked using the tool everyday and much better than the Claude Code VSCode plugin but enough is enough⦠time to switch.
Yeah same. I upgraded to Ultra for my research lab. Some of my colleagues upgraded to Claude Max. We are doing test runs right now and if Ultra hits a weekly rate limit under the same tasks, our whole lab is going with a big Claude Max plan. Same if Claude Max bottoms out first, we will go with Ultra for our lab.
I purchased based upon the statement on the plan website that there are no weekly limits for Ultra users, only 5 hour limits. But apparently this is not true and not all of the plan info has been updated correctly across Googleās websites.
For the most expensive tier, it really need to be unlimited or itās not worth. Or having a way to buy additional usage.
Hope Google handles this. Best wishes to them.
Any update on the run? I will prob. buy 100$ max plan, instead of asking my manager for ai ultra plan.
Iām facing the same usage limit issue. A few days ago, I got hit with a ā4d+ā restriction, then it went back to the 5-hour cycle. (link) But now, Iāve been hit with a ā1d+ā limit again. I saw an announcement saying the Ultra model doesnāt have weekly usage limits (link), so I donāt understand why this is happening.
Update: Itās great so far.
I havenāt hit any rate limits putting in 8-16 hour days using Opus 4.6.
I believe this is partially due to my workflow. I break things into tasks (sometimes pretty big tasks). I have separate agents solve them. I have throughout but concise documentation that each agent uses to get itself familiar with the project.
Also if a task is going really long, I check in with the agent about the size of the context window. If the context window is getting really big then I find a good stopping point, update documentation, and have it create a handoff doc for a new agent.
So far very happy, but Iām sure you would be equally happy with Claude Max, I just really enjoy the Antigravity IDE. Hope it takes off as a Google product and doesnāt get enshittified <3.
Using AI Ultra btw if that wasnāt clear
Dude Iām really interested to learn more about your workflow of using AI.
Can you elaborate more?
uuuuh kinda, but you can also ask the agent to build its own rules and workflows.
I just have some work flows for starting a new session and reading start up docs and a workflow for ending a current session and updating the docs before ending
is it more like hook implementation or just instructions in somewhere (GEMINI.md)?
idk what hook implementation is so itās not that
look at the antigravity documentation Google provides. specifically the rules and workflows section.
I have rules set to keep certain coding styles and documentation standards. and workflows built for repetitive tasks like starting and ending session.
Re Work Flow, Claude mentioned one time that it minimizes the usage if you can bundle several inquiries into one response. So I number them. Number one, first response, number two, second question, number three, next question. And it seems to work. It does keep the usage a bit lower. And of course, as youāve said all of us swear, keeping the context window to a reasonable size is a big help as well, across the board for all AIs.
Basically we now need another ultra sub for the locked period. Or, completely stop using Google AI Products. Iām tired and stopped using Antigravity and trying to avoid all Google AI Products altogether.
Right now i canāt even know what exact model i am running on. For all i know my prompts might be using a cheaper model even if i select opus 4.6 or gemini 3 pro.
Iām in the same boat. I am livid and blocked from working
AI Ultra sub, just got locked out for 168 hours.
I monitor my quota with an extension. 5hour never got below 80%, Iām not a heavy user. Any comms would be appreciated.
This is insane
Finally bought google business standard for Gemini cli, and canāt even use it/itās throttled me out due to āhigh demandā on their servers.
So bought AI ultra and STILL Iām throttled and unable to use the service - not even one prompt! (And yes I am updated and actually using Ultra in the CLI terminal). I havenāt even used it yet!
Disgusting and shocking. I donāt think we should be accepting or tolerating this tbh.
I canāt be the only one being affected and fuming like this? Can we all co-sign a formal complaint letter or something to their senior management to make sure they are actually providing us the service we have paid for?