We believe the problem lies in the fact that when Antigravity first launched, we were attracted to the fact that the credit limits for all AI models automatically reset every 5 hours after we purchased the Google AI Pro plan. This feature made us willing to pay for such a plan, especially since it attracted many developers worldwide.
However, starting in mid-to-late January of this year, this reset mechanism suddenly changed. This has resulted in many users having to wait 7 days for their credit limits to reset once theyāve used them up, significantly reducing AI collaboration capabilities. We are very disappointed with this outcome. We know the Antigravity team is working on resolving this issue, and we hope for a positive result soon. Thank you sincerely for your hard work!
Thank you for the update and for your continued work on Antigravity.
I understand the need to manage demand and maintain fairness between users. However, the price gap between the Pro plan and the Ultra plan is currently very large. For many serious users and independent developers, Ultra is simply out of reach, while Pro becomes limiting once the weekly quotas are hit.
It would make a lot of sense to introduce a middle-tier plan ā for example around ā¬40/month ā offering higher quotas than Pro without requiring the full commitment of Ultra. Such a plan would likely attract many active users who want to rely on Antigravity daily but cannot justify the Ultra pricing.
I believe this would improve accessibility while still keeping Ultra as the premium option for heavy professional workloads.
When we took the decision to go with Google AI Pro plan, we took very seriously their offer statements and we decided based on those facts.
That was false advertising to say the least.
I see the only option is to actually to drop the subscription as the Ultra plan is simply too expensive and due the way treat Pro users I cannot thrust in their policies anymore even for the Ultra.
A reminder bellow about what was promissed in the first place:
Possible workaround for Claude Opus 4.5 limits in Google Antigravity (Pro Users)
Iām currently testing something regarding the 5-hour reset timer, because many users report that their limits donāt reset correctly anymore.
Hereās what I noticed so far:
It might be important to NOT run the usage down to 0%.
During my last tests, my quota reset back to 100% after ~5 hours when I still had around 40ā60% remaining. This has now worked 3 times in a row for me ā but I need more data from others to confirm.
My current assumption:
⢠If you fully exhaust the Opus 4.5 quota, you may trigger a longer cooldown or a different limit tier.
⢠Staying above 0% could keep you inside the normal rolling reset window.
Important:
This is NOT a real solution. If you rely on Opus for production work, switching to another model temporarily might still be necessary ā because stopping mid-workflow isnāt realistic.
If youāre on Pro:
Please test this and report back:
Use Opus normally but stop before hitting 0%.
Wait for the 5-hour timer.
Check if your quota resets to 100%.
Iām trying to figure out whether this is an actual workaround or just inconsistent rate-limit behavior.
You didnāt use much of your quota, so it was restored. The 5-hour limit itself is now being restored mostly normally, but the problem is that the weekly limit, which you donāt see, is just insanely small. Iāve already tried stopping at 20%, but it doesnāt help. Once you reach your weekly quota, youāll be limited.
Itās real pain, especially when you pay for Ultra. It also made me take the Claude Max sub, that has real benefits now I really using it⦠google we need a fix, or weāll move to a competitorā¦
For 2+ weeks, a known bug caused Antigravity IDE users on Pro/Ultra (including paying customers and trial users) to be treated as Free tier, getting locked out after 1ā2 prompts. The issue was acknowledged and later fixed, but the real problem is what happened during the outage: people who paid (or were on Pro/Ultra via a promo) effectively couldnāt use the service they were entitled to.
When contacting support, the response was a script: āengineering is investigating / we canāt adjust anything / check future promotions,ā with no meaningful compensation path and no ownership of the business-side impact (lost access, wasted subscription time).
If you were affected:
Were you on paid Pro/Ultra or on a promo?
Did you get any refund/credit/extension or a real escalation?
Whatās the best way to push this beyond frontline scripted support?
Iām trying to start a thread so affected users can compare experiences and figure out what actually works, because āsorry + nothing we can doā isnāt an acceptable resolution when access was broken for weeks.
I didnāt paid pro on promo, I was just before, but now, Iām paying, and Iām affected, I thing itās just a question of timing but still
I didnāt have any refund or anything (btw where can you contact the support, itās so hard to have someone or even a chatbot)
See previous answer, I even tried on Instagram on different page but nothing
When it comes to Gemini 3 Pro, Flash, and Claude, I have no complaintsāthe limits refresh every 5 hours. Claude and GPT token limits are counted together. Is the GPT-OSS not working for you at all?