We’re in useage tier 5 and got this error with the Batch API:
“Enqueued token limit reached for gpt-4o-mini in organization X. Limit: 1,000,000 enqueued tokens. Please try again once some in_progress batches have been completed.”
Is this a bug or do we misunderstand the limits?
Getting the same error though we are in tier 5. I see several issues reported on this topic…anyone from OpenAI monitoring this and could they provide some suggestion?
Doesn’t anyone care or answer this question? I’m on tier 5 and still can only do one batch task at a time.
I’m experiencing a similar issue. My organization is on Tier 5, yet occasionally, even a single batch fails due to exceeding the 1,000,000 enqueued token limit. Interestingly, when I retrigger the same batch a few hours later, it processes without any problems and doesn’t encounter any rate-limiting constraints. This error consistently seems to occur around midnight. Does anyone know if OpenAI imposes any time-based system constraints that might explain this behavior?
Same here, did you ever find a solution to this ?
I’m having this problem as of April 2025.
I have no other batch jobs running and yet I’m told:
Please try again once some in_progress batches have been completed.
Has this been confirmed to be a bug by anyone anywhere? Is there a solution?
I’m having this problem as of May 2025. I’m waiting for any request to complete before sending the next one. Even the smallest request gets rejected. I created a new API key and it’s working for now. We’ll see how long that lasts. Worst case I can create multiple API keys.
Nope. Weird. I believe it’s a bug. Changing API Keys worked for a very short time before stopping. I have no batches in progress and can’t submit any without errors. I’ll wait 24 hours before giving up.
I have the same issue. API always reject any requests even I gradually changed it to a smaller number of sample which definitely not beyond the limit.
Update: I tried to create different keys but it still not works. When I tried another account, everything works smoothly. I think the potential reason is that either the enqueued token limit takes account into the tokens of failure batch. Just a guess at this moment.
I kind of believe there is a bug for calculating the enqueued toekns for batch api.
Background: I sliced my data equally to 152 samples per slice and submit the batch one by one. I checked the average token usage (in+out) from finished batch is ~1050/sample. I tried it with a Tier 1 account and using GPT-4.1-mini.
First time it failed, I tried to resubmit that slice and it works. But when I submitted the next one, it always failed. It is very frustrating that you don’t know what is happening, and you have to wait 24h to continue.
That happens not the first time and I try to understand how to avoid it.
I ran into this before. I believe that it’s assuming all of your max outputs tokens will be used.