Batch
batch_68cc8dd557cc81908d0e86686caf59b6
Status
In-progress 0 / 1
Model gpt-5
/v1/chat/completions
Completion window
24h
1 Like
Some time yesterday, all batch jobs stopped, and all newly created jobs stayed at 0 progress. It must be some problem and the batch pipeline is stuck.
I am sorry for the double topic. I have the exact same problem. Jobs are stuck at zero, and sometimes yesterday simply stopped mid-progress and expired, and new jobs do not move.
2 Likes
No worries. I’ll see if I can find out anything.
1 Like
batch_68cc99966e68819094d81012e2b7dac8
still stuck in progress even if my company is paying top dollars for service.
1 Like
I’ve passed it along, but I haven’t heard anything yet.
Hopefully, someone from the team will stop by.
Thanks for reporting!
Hi Harry,
Sorry about the delay (working through a bit of a backlog!) and disruption. We did see intermittent batch pipeline slowdowns earlier this month that sound similar to what you described. Engineering observed some jobs stalling during the final output/aggregation step and periods of worker instability that reduced processing capacity, which can leave new uploads at zero progress or cause jobs to expire. We also saw instances where a processing stage timed out without emitting metrics, which can make jobs appear stuck.
How are things looking on your end now? If you're still affected, can you share the affected batch IDs and approximate timestamps?
Thank you!
1 Like
I’ve seen this happen before — sometimes the batch gets stuck due to a network delay or internal queuing issue. It helps to check the batch job logs and ensure that the callback endpoint (if you’re using one) is responding correctly.
If the status stays “in-progress” for the full 24 hours, try splitting the job into smaller batches or re-running it with fewer requests. That often clears it up.