⚠️ Gemini API Instability During New Model Releases

Google Cloud Compute Infrastructure
1

Overview

I have observed instability in the Gemini API coinciding with the release of new model versions by Google. These disruptions directly impact production applications, particularly those relying on features such as function calling and low-latency responses.

Issue After Gemini 2.5 Pro Release

  • Event: Following the release of Gemini 2.5 Pro by Google.

  • Impact: The function-calling feature in Gemini 2.0 Flash began failing intermittently for approximately three days.

  • Observation: No changes were made to the code or the app, resulting in inconsistent behavior that is unacceptable in production environments.

Similar Problem After Gemini 2.0 Flash Launch

  • Scenario: When Gemini 2.0 Flash was introduced:

    • Applications using Gemini 1.5 Pro experienced a drastic increase in response times—from milliseconds to over 15 seconds for identical inputs.

    • The issue persisted for about two days and then resolved without any modifications to the code.

  • Implication: New model rollouts appear to affect older models that are still actively used.

Why It Matters

  • Unreliable Performance: There is significant instability during model transitions, leading to:

    • Sudden latency spikes.

    • No changes made from the user side.

  • Production Impact: The unexpected behavior in production makes it challenging to depend on Gemini for critical use cases.

Community Feedback

  • Observation: The issue aligns with reports from other community members.

  • Reference: Developers in a Google AI forum thread have also reported major slowdowns during new model releases.

Final Note

Based on my observations, these issues have repeatedly occurred during new model releases. Although the problems seem to resolve on their own without any intervention from my side, the instability raises serious concerns about trusting the Gemini API for stable production use.

Can anyone help, please?

2

Hi @Mahmoud_Zaher ,

Thanks for raising your concerns with the Gemini 2.5 Pro Experimental model.

Just a heads up, the Gemini 2.5 Pro Experimental model is currently on the Preview launch stage, which means that its features are intended for testing by users. Preview offerings are often publicly announced, but are not necessarily feature-complete, and no SLAs or technical support commitments are provided for these. Unless stated otherwise by Google, Preview offerings are intended for use in test environments only.

For now, I recommend you provide feedback or create a bug ticket about your issue. Include steps to replicate, configuration settings, screenshots, and other details to help us investigate. While there isn’t a specific timeframe for resolution, once we’ve fixed an issue in production we’ll indicate this and then update and close the bug.

You may also consider checking the Gemini API Status dashboard regularly for any possible incidents or outages.

For more visibility, I recommend cross posting your thread on these channels:

If you’re an Admin of your Google Cloud Organization, you can open a case here for more assistance.

Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.

3

Hello @Rhett ,

Did you actually read the description?

This has nothing to do with Gemini 2.5 Pro. The issue is clearly about instability during new model releases — specifically with Gemini 2.0 Flash, which is supposed to be a stable version.