Bug, ai-studio, build

Hello,

I am unable to continue working on an existing, business-critical project in Google AI Studio Build.

Every Build prompt immediately fails with:

“An internal error occurred. Retry.”

The failure occurs after “Ran for 0s”, even with extremely short prompts such as:

“Reply only with OK.”

Billing and quota status:

  • Paid Tier 3
  • Prepay billing
  • Existing API key linked to the correct project
  • Monthly spend: EUR 221.02 of a EUR 1,000 project spend cap
  • RPM usage is far below the limit
  • TPM usage is far below the limit
  • RPD usage is below the daily limit
  • The problem continued after the daily quota reset

API verification:

I tested the same project, API key and Gemini 3.8 Flash model through the Interactions API.

These direct requests succeeded with HTTP status 200 and returned the expected response. They are visible in the AI Studio Logs dashboard.

However, the failed requests submitted through AI Studio Build do not appear in the API Logs. This suggests that Build fails before the request reaches the Gemini API or that the paid API key is not being routed correctly by the Build agent.

Troubleshooting already completed:

  • Tested the original app
  • Tested a Remix of the app
  • Started a new Build chat
  • Tested multiple browsers
  • Tested after the daily reset
  • Confirmed active Paid Tier 3 billing
  • Confirmed sufficient prepaid balance
  • Confirmed that project spend and API rate limits are not exhausted
  • Confirmed that direct API requests succeed with HTTP 200
  • Checked the official AI Studio status page

The issue persists in all cases.

Impact:

The AI Studio Build agent is completely unusable for this project. The project contains a large and complex business application, so rebuilding it elsewhere is not a practical workaround. Starting a new Build chat has also made the previous detailed conversation inaccessible.

Please investigate the Build agent entitlement and API-key routing for this account and project. This does not appear to be a normal Gemini API quota, billing, model-access or API-key problem.

For confidentiality reasons, I am not posting the full project ID, app ID, app URL, source code or company information publicly.

Project ID:
gen-lang-client-0949…

I can provide the full project ID, app ID, timestamps and additional screenshots privately to a verified Google employee.

Relevant timestamps:

  • Problem began: September 15, 2026
  • Retested after daily reset: September 16, 2026, approximately 09:06 CEST
  • Direct API tests returned HTTP 200

Thank you.

Hello @rainer_unruh ,

DM’ed you for more details.

Hello Sai_Deepika_K,

I have same problem about An internal error occurred. I did everything what users are suggeting to get it fix, but the error still stands. What could be a problem and is there any quick fix.

Hello @LongMiles ,

DM’ed you for more details.

Hello Sai_Deepika_K,

I have same problem about “Encountered retryable error from model provider: Agent execution terminated due to error.” I did everything what users are suggeting to get it fix, but the error still stands. What could be a problem and is there any quick fix.

Hello @Rain_Forest ,

DM’ed you for more details.