[Critical Bug] Total IDE lockout: "state syncing error: key not found" leading to "MODEL_PLACEHOLDER_M16" fallback

**Context:** I am a Google PRO account user using the native Antigravity IDE. I am currently completely locked out of sending any prompts (getting a persistent “Failed to send” UI error), even though I have available quota on my selected models.

**How the issue triggered (Cascading Failure):**

  1. I was using a model that ran out of quota and hit a `429 RESOURCE_EXHAUSTED` error (Claude Opus with an 87-hour wait).
  2. I switched back to **Gemini 3 Flash** (which has its quota bar completely full).
  3. The IDE failed to update the model state properly. From that point on, sending any message fails instantly.

**The Underlying Logs:**
Looking at the internal Output logs, there is a severe state corruption regarding the OAuth token and model fetching. The IDE fails to sync the state, fails to load the available models cache, and then panics by attempting to use an empty placeholder:

`failed to get OAuth token: error getting token source from auth provider: state syncing error: key not found`
`Cache(availableModels): Singleflight refresh failed: error getting token source: state syncing error: key not found`
`/exa.language_server_pb.LanguageServerService/SendUserCascadeMessage (unknown): unknown model key MODEL_PLACEHOLDER_M16: model not found`

**Troubleshooting Steps Already Taken (None Worked):**
To rule out local caching issues, I have performed a complete “hard nuke” of the local environment:

  1. Logged out of the PRO account.
  2. Closed the IDE and completely renamed/deleted the `%APPDATA%` and `%LOCALAPPDATA%` Antigravity folders to force a clean credential vault generation.
  3. Reopened the IDE and logged back in.

**Result:**
The exact same `state syncing error: key not found` persists immediately upon logging in. The IDE cannot fetch the model list for my PRO account, breaking the entire agent executor.

It seems my account state is corrupted on the backend servers or the IDE’s authentication handshake for PRO users is currently experiencing an outage after a 429 error. Any immediate workarounds?