Here is Why Your Quotas Are Not Resetting

While researching my issue(!), I noticed that all Antigravity users are suffering from the same situation. Thousands of comments have been made under hundreds of threads on this forum and Reddit. While all Pro and Ultra users are struggling with the exact same issue(!), there hasn’t been a single statement from Google. Users are literally being ignored.

Now, if we apply some quick logic: it’s incredibly obvious that these quota reset issues aren’t actually unmanageable or unsolvable bugs. If this were a bug they were actively trying to fix in the background, they would have definitely made an announcement and informed the users.

The real issue is a marketing trick used by Google.

A few months ago, no one even knew about Antigravity; tools like Cursor, Copilot, and Bolt were already being heavily used. In a classic move, Google gathered people onto their app by offering free usage quotas, and by providing overly generous limits, they managed to convince us all that it was a great IDE. Let’s admit it, in the early days, it had higher limits than any other IDE, and many of us canceled our existing subscriptions to switch to Google.

Then, once Google was convinced they had gathered enough users, they started reducing the limits. I literally experienced this firsthand; it’s impossible not to notice the initially allocated limits slowly shrinking. Later, they gradually started ignoring the promised renewal times, pretending the algorithm was broken. If you’ve noticed, they neither deny nor confirm it; they are literally not saying a single word. Google is a massive company that should be extremely sensitive about customer support. I have to say this: your quotas are not going to get fixed, because there isn’t even an official claiming that they are broken.

I know why you chose Antigravity, and Google knows it too… You will never get that initial efficiency again. That’s exactly what they are working towards; if you are efficient, they won’t make enough money. And at this point, who cares about the to-do apps you are developing anyway? :smiley:

The best thing to do is to cancel your subscriptions, rent the Claude API via OpenRouter, and try to stay away from these tricks with a pay-as-you-go approach. Stop asking questions, my friends, they are not going to answer.

18 Likes

I agree entirely. Well said.

2 Likes

@that.Steve I’ve been considering Clause API via OpenRouter I guess it’s time to do some research. If i make the switch i have to rebuild all my MCP’s and “Skillz” etc, i guess. But before I do I want to research the pro’s and con’s. I stil think Google is “figuring out the sweet spot”. They are paying developers to build it so I’m sure there is a cost involved and they need to recover their investment and figure out how much they’re going to have to spend on “ongoing support” of the product. Afterall , they’re in business to turn a profit and I get that.

3 Likes

I’ve been on the Pro plan and would upgrade to Ultra if I could reliably know what I could get out of it, but with how uncommunicative Google is and with how they’ve played a shadow game with the usage, it’s really making me look favorably towards Cursor or Claude Code. I was using that before and they’re great.

What’s really sad is that Google Workspace users can’t even use the 25,000 AI credits in Antigravity. So why in the world would someone upgrade?!

@Google_Google you’re killing the momentum that you have with early adopters and evangelists like myself. On my personal account I’ve been a paying subscriber of 3-4 of your services (Google One, AI Pro, YT Premium, etc) for 10+ years, through all the pivots and iterations your services have taken. But not enabling a developer tool like Antigravity to use Ultra AI Workspace credits makes it an impossible tool to use.

3 Likes

Yeah, the amount of tokens to use should be clearly shown, now it is not transparent; it could be anything, and it could be less tomorrow than it was today - there is no way to measure it lol.

I used Antigravity maybe 1.5 days for several hours, then hit the weekly limit, had to wait 7 days, and now when it was supposed to reset - it didn’t - the quota only reset partially to 20%… so now I have to wait 7 days again? Not really usable.

So after all this quota isn’t even a weekly quota, maybe a monthly or something - who knows. I already got account from another company. l will see how that works for me.

2 Likes

Hello @sarp_efe @Eses, welcome to AI Forum!

To provide you with greater control and a seamless path to scale, we are evolving our Google AI plans by introducing built-in AI credits that can now be utilized directly within Antigravity.

Under this updated structure, our Google AI Pro tier is tailored for practical builders, offering generous limits for Gemini Flash alongside a baseline quota to experience our most advanced premium models. For developers operating at the highest scale who require consistent, high-volume access to our most complex models, Google AI Ultra serves as the ideal comprehensive solution.
Furthermore, we’ve designed these plans with maximum flexibility in mind; if you are on the Pro plan and require additional capacity or deeper premium access for intensive development sprints, you can seamlessly top up your AI credits to customize your limits and continue shipping your best work without interruption.

5 Likes

Thanks - it is not the answer I need. Google as company should clearly communicate this kinds of upcoming changes beforehand, and not trying to explain it afterwards.

6 Likes

This does not answer the questions we users have, seems there’s no reason to use Google products anymore

7 Likes

The way it was before was good, it was in a good balance face codex and claude.
But now, it is just ridiculous. Codex offers much more for you 20 bucks, even cursor does.

Going to cancel my subscription, that’s it…
This is a language Google understands.

7 Likes

To be honest, none of your plans inspire confidence. I have a few developer friends with Ultra accounts, and they are dealing with the exact same issue. These people didn’t subscribe just to burn through credits. That wasn’t what you promised; we had no idea you’d try to sell us extra credits, or that we’d face long wait times if we didn’t buy them.

So, dear Abhijit, can we just sum it up like this? “The tasting menu is done, time to pay up.’ :)”

8 Likes

i didnt touch for almost a week to my ide (antigravity) but it still not reset my usage quota! extend it for more 1 week.

I’ve been on the Ultra plan using it almost daily for two months, and I just hit my first limit last night. I usually juggle 3-4 projects at a time, relying mainly on Opus 4.6. I never came close to hitting the ceiling before, but the throttling started yesterday and is still continuing today.

It really feels like usage quotas have been quietly downgraded across the board. Honestly, this might be an industry-wide shift. Running AI infrastructure is incredibly expensive, especially with rising global energy prices tied to the ongoing conflicts in the Middle East.

Brace yourselves for a lower-compute reality. We all got hooked and might be addicted to AI workflows by now. Time to start using open-source models locally and train our own agents.

I started using QWEN 3.5 9B via LM Studio. I also use Cline via VS Code (they have free models for limited period occasionally).

Anybody can share honest views on how Claude Max $200 stack up vs Antigravity Ultra? I still prefer IDE vs CLI.

2 Likes

Qwen 3.5 9B is not very good to integrate into tools except as a chat agent I think. Been considering the 5x Claude myself as I feel it is too expensive the top tiers as a hobby developer.

hey, i encountered this 7 days waiting period with my pro subscription around 3rd of March. Since i need to work, i switched to Ultra plan. Since then - i have no limits that i actually reached, smooth sailing. i don’t know what i’ll do after 3 months when subscription price doubles but i’ll worry about it when i get there. Right now - i am working without limits.

1 Like

can’t compare to frontier models for sure but good to have a lightweight model that is free to use for simple tasks. I’m also using it to experiment with autoresearch. So far not yielding much results.