You are on Antigravity Pro but you keep hitting a sudden seven day lockout or your five hour refresh does not come back as expected. You might also see Pro features acting like the Free tier.
The Problem
Users on Antigravity Pro run into a confusing quota situation. The five hour rolling refresh for sprint work exists, but a hidden seven day baseline cap can override it and force a full week lockout. Some users also see a Pro to Free tier glitch due to authentication, with Output logs showing OAuth or Login errors.
For a deeper explanation of how Pro quotas are counted and why the weekly cap triggers, check our plain language walk through here: Antigravity Pro quota guide.
Solution Overview
Aspect
Detail
Root Cause
A hidden weekly baseline cap can override the five hour refresh. Authentication glitches can also misclassify a Pro account as Free. Intensive agent activity drains the weekly pool faster.
Primary Fix
Check the Quota Monitor and Usage Dashboard, stop any stuck agents, switch to lighter models for simple tasks, and resolve the OAuth tier glitch by sign out and sign in.
Complexity
Easy
Estimated Time
10 to 20 minutes
Fix the Antigravity Pro seven day lockout and restore the five hour refresh
Open the Quota Monitor to understand what is actually limiting you right now.
Navigation:
Agent Manager > Settings > Models
Review both meters:
– Rolling five hour units shows what you spent in the last five hours. Units here return exactly five hours after they were used.
– Weekly baseline shows the hidden seven day pool. If this is exhausted, the lockout is expected until the next seven day window starts.
If the baseline is low for example below twenty percent pause heavy tasks.
Step 2. Stop any agent that is stuck in a loop
Infinite loops drain units very quickly. If you see an agent making many tiny edits without progress, stop it now.
Navigation:
Activity sidebar > Running tasks > Stop
Step 3. Switch to a lighter model for simple tasks
For quick Q and A or small fixes, switch from a heavy model to Gemini 3 Flash to preserve the weekly baseline for real sprints.
Step 6. Accept the weekly wait if the baseline is already empty
If the baseline pool shows empty, the lockout is expected. The account will unlock at the start of the next seven day window. Plan around that reset and keep your heavy work inside the next cycle.
Alternative Fixes and Workarounds
Use the IDE Output quickly to confirm the cause
If Output shows no OAuth or Login errors, it is a real quota hit. If it shows errors, fix auth first since that is faster than waiting a week.
Avoid third party prompt chaining
Automated prompt chains and external macro tools can trigger abuse filters that cause an immediate seven day lockout. Remove these tools and retry after the window resets.
Reduce compute heavy actions
Turn off or limit Browser and Terminal actions during light work days. Save them for your intended sprint window.
Split large refactors
Break a repo wide refactor into smaller chunks across multiple five hour windows so the weekly baseline drains more predictably.
For an example of triggers that trip the cap, see our short explainer: common quota triggers.
Troubleshooting Tips
Know the two clocks. The five hour meter refills exactly five hours after each unit was used. The weekly baseline is a rolling seven day pool and it can override the five hour refresh.
Confirm your subscription tier. In Account or Billing pages verify you are on Pro and not temporarily flagged as Free.
Re auth resolves many false lockouts. A clean sign out and sign in usually fixes the tier misalignment bug.
Watch for background tasks. Long linting or test runs driven by the agent can rack up units even when you are not typing.
Keep an eye on the meter. When the baseline drops below twenty percent, switch to lighter tasks and models.
Best Practices
Plan heavy work inside one focused window but watch the weekly baseline so you do not hit the cap mid sprint.
Switch models based on task type. Use heavier models for complex edits and lighter ones for explanations or quick diffs.
Stop loops fast. If an agent looks stuck, stop it before it burns the pool.
Review the Usage Dashboard daily. A quick look prevents surprises.
Avoid external automation that triggers abuse filters.
If you want a simple crib sheet for planning usage across the two quotas, check this short read: how the two quotas work.
Final Thought
The fix is straightforward. Check the quota meters, stop wasteful tasks, switch to a lighter model when appropriate, and refresh auth if the app forgets your Pro status. Once you align your workflow to the five hour and seven day clocks, the lockout stops being a surprise and your Pro time delivers steady output.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
