Gemini Models Has Become Completely Unusable in Antigravity – Extremely Slow Responses and Constant Model Unavailability (Paid Pro User)

I’m experiencing a serious issue with Gemini 3.8 Flash and the rest of gemini models in Google Antigravity. At this point, the model has become practically unusable.

This isn’t simply a performance issue with complex coding tasks. Even sending a basic message like “Hi” takes an extremely long time to receive a response, and frequently fails altogether. The model either gets stuck generating, takes an unreasonable amount of time, or returns errors such as “Model unavailable.”

I’m a paid Pro subscriber, and this level of reliability is unacceptable, especially when the model is supposed to be available as part of a paid subscription.

I’ve already tried virtually every troubleshooting step I can think of:

  • Switched between multiple Google accounts to rule out an account-specific issue.

  • Tested Gemini 3.8 Flash with completely simple prompts, including just saying “Hi”.

  • Tried using the Antigravity CLI (AGY CLI) instead of the regular application.

  • Tested using Hydrogen as well, but experienced exactly the same problems.

  • Completely uninstalled and reinstalled Antigravity.

  • Retested the model after reinstalling, but the problem remained exactly the same.

None of these attempts made any difference.

The most concerning part is that this happens even without any complex tasks, coding, large context, or demanding instructions. Basic interactions are already extremely slow and unreliable.

The fact that the same issue occurs through different interfaces and tools, including Antigravity, AGY CLI, and Hydrogen, makes me question whether there is a deeper problem with Gemini 3.8 Flash’s infrastructure, API access, model integration, or backend services.

I would appreciate an actual investigation from the Antigravity engineering team, not generic troubleshooting suggestions.

  1. Is Gemini 3.8 Flash currently experiencing infrastructure or backend issues?

  2. Are there known problems with model availability or response latency?

  3. Why is the model practically unusable even for basic prompts?

  4. Is this a widespread issue affecting other paid Pro subscribers?

  5. Could there be an underlying issue affecting Gemini 3.8 Flash across different clients and interfaces?

I’m paying for the service and expect the advertised models to be functional and reasonably responsive. Right now, Gemini 3.8 Flash is effectively unusable, and reinstalling the application, changing accounts, switching to the CLI, or trying Hydrogen makes absolutely no difference.

Please investigate this issue and provide a concrete explanation or fix.

Bro are you jio subscriber?

Im actually facing problems with my clients because of this problem , for a basic adding to the header its still working for hour ago i mean since yesterday i have not worked properly my productivity went to zero because of this problem and no one is talking right now we need explanation and a solution and when ur gonna fix this , please understand our frustration and say something why gemini 3,8 flash is broken we need an answer.

even ultra is same issue.

Hi @Mazen_Mahmoud,

Welcome to the Forum and thank you to everyone in the thread for sharing what you’re seeing. Since you’ve already ruled out your local setup across both the app and CLI, this points to server-side Gemini capacity spikes combined with client retry issues we just patched:

  • Update to our Nov 17–18 releases: Please update to Antigravity App v2.33.0 and Antigravity CLI v1.5.11 (agy update). v2.33.0 fixes stale pooled network connections that caused retries to fail and 5-minute thinking timeouts, while CLI v1.5.10+ improves capacity retry handling.
  • Separate backend during Gemini spikes: "Model unavailable" reflects temporary serving demand rather than your Pro or Ultra quota (Plans). On a non-trial paid Google AI Pro or Ultra plan, switching temporarily to Claude Sonnet 5.5 (thinking) or Claude Opus 5.5 (thinking) (Choose a model) routes to a separate serving pool.

If Gemini latency still persists on v2.33.0 / v1.5.11, please share your region so we can flag it to our serving team.

Hi! @Rebbanpalli_Naveen It still the same issue, all my pro accounts error persists
latest antigravity version

i’ve had enough of this joke, every few weeks antigravity is just a joke, i am cancelling my pro subscription. Latest models are worst for my tasks and 3.8 became like a brainless model… Enough