Hi,
I submitted a bunch of batch jobs that stayed in BATCH_STATE_RUNNING for approximately 24 hours before finally transitioning to SUCCEEDED. However, all results in the output file contains “Internal error encountered” instead of valid responses.
Details:
- Example batch name: batches/t68sodku084bqp26sypbio0qc7nnjt2ult5e
- Model:
gemini-3.5-flash-lite
This started abruptly — batches submitted the previous day processed normally within minutes. Nothing changed on my side (same file format, same request structure, same batch sizes).