After running a small set of (5) batches, each only with around 20 tasks, they all expired (failed) with 0 progress after 24 hours. This was using the ‘gpt-3.5-turbo’ model, with no temperature value (which was known to give errors last week). Previously running these small batches would see progress within a few minutes or hours. Is anyone else having similar issues with batch jobs not being completed / no progress being made?
7 Likes
Yes, I am having a similar issue. Yesterday I run several sets of batches with gpt-4.0-turbo, and they would run successfully in 1 or 2 minutes. Today I rerun the same batches, but they are stuck in progress, even after 1 hour.
I’ve noticed several batches that expired yesterday with 0/8000 progress. I restarted them today as new batches, but they’re still showing 0 progress after 10 hours. What’s going on?
@OpenAI, using the batch API has been quite frustrating lately. Is there anything you can do to improve its usability?
I have similar problem from yesterday on 4o-mini
1 Like
I have the same issue with 40-mini. I tried to run something for a few days now, but the batches keep expiring over and over again without any progress. This is really annoying.
Did you find any solution?
1 Like
Similar issue. We are using the gpt-4o-mini model, and all of our batches have been expiring after 24 hours since 24 October.
2 Likes