Google AI Studio & Gemini API — Current Stability Issues
Many developers are currently reporting serious problems with Google AI Studio and the Gemini API. From a technical perspective, this appears to be a combination of rollout instability, capacity constraints, and backend issues.
Commonly reported problems:
-
Requests stuck indefinitely in “Thinking”
-
Batch jobs remain in “Pending/Running” without completing
-
503 Service Unavailable and 429 Rate Limit errors, even at low usage
-
Daily budgets consumed without meaningful output
-
Generic “Internal error” messages with no diagnostics
-
Performance varies significantly by time of day and region (EU users especially affected)
Particularly concerning is that official status pages often show “Operational” while users experience severe degradation. Third-party monitoring services and developer forums report multiple incidents and regional disruptions.
Technical context:
AI Studio is only the frontend. The service depends on a multi-layer infrastructure:
Frontend → Gemini Serving Layer → Vertex AI → Scheduler → TPU/GPU Compute Pools
Failures or bottlenecks at any layer can result in:
-
Endless jobs
-
Incorrect quota or rate-limit errors
-
High latency
-
Stalled or lost requests
-
Unexpected cost spikes
Many reports indicate these issues started after a major ground-up rewrite of the Build experience that is still being rolled out.
Bottom line:
This does not appear to to be an isolated case, but a broader stability problem during rollout…..
It is completely unusefull at the moment