Hi Google team,
I’m experiencing a bug where the model quota in Antigravity
IDE is showing abnormally long refresh timers, even for
models that have not been used.
[Environment Info]
- Antigravity Version: 1.20.5
- VSCode OSS Version: 1.107.0 (user setup)
- Commit: 4603c2a412f8c7cca552ff00db91c3ee787016ff
- Date: 2026-03-07T01:34:20.929Z
- Electron: 39.2.3
- Chromium: 142.0.7444.175
- Node.js: 22.21.1
- OS: Windows_NT x64 10.0.26200
- Language Server CL: 879885162
[Account Info]
- Plan: Google AI Pro (confirmed in Settings → Account)
- Billing Country: South Korea (KR)
- Last Payment: ₩29,000 on March 10, 2026 (Completed)
- Payment Profile ID: 9954-2008-3323
[Bug Details]
Model quota refresh timers are stuck at abnormally long
intervals for unused models:
| Model | Refresh Timer | Status |
|---|---|---|
| Gemini 3.1 Pro (High) | 6 days 23 hours | |
| Gemini 3.1 Pro (Low) | 6 days 23 hours | |
| Claude Sonnet 4.6 Thinking | 5 days 22 hours | - |
| Claude Opus 4.6 Thinking | 5 days 22 hours | - |
| GPT-OSS 120B (Medium) | 5 days 22 hours | - |
| Gemini 3 Flash | ~4 hours 44 min |
Gemini 3 Flash refreshes normally, but all other models
are stuck. Gemini 3.1 Pro (High/Low) also show a warning
icon (
).
[What I’ve Already Tried]
- Verified subscription is active and recognized:
Settings → Account → “Your Plan: Google AI Pro”
- Cleared %APPDATA%\Antigravity cache and restarted IDE
- Submitted Bug Report via IDE Provide Feedback feature
- Emailed apps-help@google.com
[Expected Behavior]
All models should refresh on a regular short cycle as
provided under the Google AI Pro plan quota schedule.
Please investigate. Thank you.