Quick answer
Gemini API documents batch concurrency and enqueued work as 100 concurrent batch requests; enqueued tokens vary by model and tier. Batch token capacity is separate from the fixed concurrent request count.
Verified Aug 22, 2026Official source
Current limits
| Constraint | Current value | Scope | Verified source |
|---|---|---|---|
| batch concurrency and enqueued workBatch token capacity is separate from the fixed concurrent request count. | 100 concurrent batch requests; enqueued tokens vary by model and tier | Per project | GoogleAug 22, 2026 |
Why does this limit matter?
A job can fit the concurrency count and still exceed the model-tier enqueued-token allowance.
This value is scoped to Per project; a different plan, runtime, model, endpoint, region, or account can produce a different effective constraint.
What should you check?
- Compare active batch count and enqueued tokens with the current model-tier table.
- Confirm the exact plan, model, runtime, endpoint, region, and account that serve the failing workload.
- Record the observed value, response headers or configuration, timestamp, and source without logging secrets.
Important caveats
- Batch input storage and model availability impose additional independent constraints.
- Treat the official source and live account configuration as authoritative if they differ from this verified snapshot.
HyperObserve reports the documented platform constraint. Your application, SDK, gateway, provider, region, or account can impose a lower effective limit.
Related