Critical Quota Sync Mismatch: IDE says "Exceeded" while Monitor shows 20% usage

I am experiencing a significant discrepancy between the Antigravity Quota Monitor and the IDE’s actual behavior regarding the Claude Opus 4.5 (Thinking) model.

The Discrepancy:

  • Quota Monitor: Shows only 20.00% used (80% remaining). Reset date: 01/29/2026.

  • IDE Error Pop-up: Claims “Model quota limit exceeded”. Resume date: 1/24/2026, 9:33 PM.

The Problem: After re-logging, the reset timer jumped from “today” to “5 days later,” but the IDE still blocks access despite the monitor showing plenty of remaining quota. This seems like a backend sync bug between the subscription status and the local IDE enforcement.

Environment:

  • Subscription: Google AI Pro

  • Model: Claude Opus 4.5 (Thinking)

  • Location: KST (South Korea)

I’ve attached a side-by-side screenshot showing the conflict. Has anyone else seen this “ghost quota” issue? Any manual sync triggers available?